Everything people ask before they hire someone.

Roughly €400–€600 a day in France, £500–£600 a day in the UK, and $60–$100 an hour in the US, for an experienced solo developer. Different countries quote in different units, and the units are not interchangeable.

By market:

• France — €575/day for experienced developers, rising from €317 (0–2 years) to €671 (15+ years). These are rates freelancers declare on their own marketplace profiles.
• UK — £525/day contract median for React as of 4 August 2026 (n=839); £508/day for front-end developer (n=126). Both are medians of advertised contract rates on a rolling six-month window.
• Germany — priced hourly, not daily: €90/hour median for software and web development, the lowest-paid IT specialism in that survey (n=356).
• US — $50–$100/hour reported by firms in the US, UK and Canada; $61–$80/hour is the median band for freelance WordPress work on one global vetted marketplace.

Whole projects are priced differently again. Of 300+ web development firms surveyed across 31 countries, 59.9% put a basic website or MVP at $1,000–$3,000, and 45.4% put a CMS-based site at $5,000–$20,000. Those are firms answering about their own pricing, not audited invoices.

Three warnings. Day rates and project prices are not convertible unless you know how many days the job takes. No currency here has been converted, so €575, £525 and $61–$80 cannot be lined up against each other. And every figure above is what someone says they charge — none of these sources publish settled transactions.

TJM stands for taux journalier moyen — average daily rate. France prices by the day in the mission market, where a company buys days of a freelancer's time. It prices by the job in the site-vitrine market, where a small business buys a finished website. Both conventions run side by side.

• The mission market — longer engagements with companies, priced by the day. Malt indexes 625,852 French freelancers and puts an experienced developer at €575/day, from rates freelancers declare on their own profiles.
• The site-vitrine market — priced au forfait: one fixed price for the job. Codeur.com measures this side from bids placed on its own board: €769 average posted budget on WordPress projects, €1,441 across all web development, and 40 competing quotes arriving inside 48 hours.

Neither is "the French way". Which one you meet depends on what you are buying.

One trap, because both numbers circulate as "the French TJM". Malt says €575/day. Codeur says €135/day. That 4.3× gap is not noise — it is two different populations. Codeur's own senior tier tops out at €210/day, which is below what Malt reports for a freelancer with 0–2 years' experience (€317). Two indices cannot be measuring the same market when one's ceiling sits under the other's floor. Codeur is a project-bidding board and a large share of the providers bidding on it are outside France; its €135 figure is not a French professional rate and should not be quoted as one.

£500 to £600 a day for an experienced contractor. The figures below are medians of advertised contract rates as of 4 August 2026 — they run on a rolling six-month window and move.

• React — £525/day, n=839, up 2.44% year on year. The largest sample behind any figure on this page.
• Next.js — £523/day, n=120. Within a few pounds of generic React.
• Front-end developer — £508/day, n=126, up 6.44%.
• WordPress — £390/day, n=25, down 15.68% year on year while React rose. Small sample.
• "Web Developer" — £550/day, but n=14. Indicative only.

London pays more, and how much depends on which title you look at. React is £598 in London against £500 outside it (n=419 and n=314), a gap of about 20%. Front-end developer shows about 3%, on a London sample of 40. Both are as published; the sample sizes are the reason to read them together rather than pick one.

For scale on how much this moves: one freelancer trade body's member survey recorded £547, then £457, then £576 across the first three quarters of 2024. That 16% single-quarter fall is larger than any regional gap in this data. Those 2024 figures are two years old and are here for the volatility, not the level.

$60 to $100 an hour for a solo freelance developer, and $100 to $149 an hour once you are buying from a company.

The one government figure available covers employed web developers, not freelancers: a median of $44.54/hour ($92,650/year), with the 10th to 90th percentile running $23.12 to $78.03. California's median is $57.48. That is a salary comparison, not a quote you will receive.

On the freelance side, every source is commercial:

• $50–$100/hour reported by firms in the US, UK and Canada, from a survey of 300+ development firms across 31 countries.
• $61–$80/hour — the median band for freelance WordPress on one global vetted marketplace, and the same as that platform's figure for general web development, so no WordPress discount shows up in its data. Freelance React sits in the same median band there, with a higher average band of $81–$100.
• $100–$149/hour for US web development companies, self-reported to a directory that sells those companies leads.

One honest gap. No US-only freelance marketplace publishing checkable transaction data could be used for this page — Upwork, Fiverr and Toptal do not make rate data publicly retrievable, and Toptal publishes no rates at all. The widely-quoted "$30 median for web developers" could not be checked against its source, so it is not used here.

Mostly cost of living and business overhead, not difficulty of the work. A rate has to cover rent, tax, insurance, healthcare and pension where that developer lives, and those costs differ by far more than the work does. Nothing below measures skill by country, because no source found for this page does.

The scale of the cost difference, from whole-economy data rather than developer data: employer hourly labour costs across the EU run from €12.0 in Bulgaria to €56.8 in Luxembourg against an EU average of €34.9 — a 4.7× spread. A self-reported developer survey shows the same shape inside a single job title: an engineering manager's median is $200,000 in the US, $118,000 in Germany and $52,000 in India.

Senior freelance rates by region, from one marketplace's own blended benchmarks: $82–$130 in the US and Canada, $70–$120 in Western Europe, $40–$85 in Eastern Europe, $35–$75 in Latin America, $28–$65 in Southeast Asia. Read those as directional — the publisher calls them directional benchmarks itself, not measurements.

Two things worth knowing before you conclude that cheaper means worse. Inside one country, remote work has flattened the geography: German rates move about €10/hour between regions, and the source's own explanation is that location matters less than it used to. And the spread by experience inside one country is large in its own right — France runs €317/day at 0–2 years to €671/day at 15+, a 2.1× range. A country average tells you very little about the individual you are actually talking to, in either direction.

About 1.4× to 2.4×, depending on what you are buying. The gap is smaller by the hour and larger on a monthly retainer.

The only figures here from a study with a stated sample are for SEO and marketing services rather than website builds — 439 providers:

• Hourly — $71.59 freelance, $98.90 agency, $171.18 consultancy. A 1.4× gap between freelance and agency.
• Monthly retainers — $1,348.63 freelance against $3,209.00 agency, a 2.4× gap, from the same study. In local work specifically, $1,150 against $1,819, 58% more.

For website projects, the only published freelance-versus-agency splits are vendor editorial with no method behind them. Treat them as impressions, not measurements:

• UK — £1,000–£3,000 freelance against £2,000–£7,000 agency, from a lead-generation marketplace that publishes no methodology at all.
• France — €500–€3,000 freelance against €2,000–€10,000+ agency, also vendor editorial with no stated method.

There is one measured number for what the extra money buys, and it is scale rather than quality: in one PPC survey, agencies manage an average of $2.9M/month in ad spend against $545K for freelancers — about five times.

When the agency is the right buy: when the work needs more than one pair of hands at the same time, when you need someone answering while your developer is ill or away, or when your procurement requires a company on the other side of the contract. One freelancer cannot supply those, and I would not pretend otherwise. Nothing in these sources measures whether the output is better.

Be careful with ranges that look far apart and are not. On SEO audits, one French source quotes €450–€600 freelance against €500–€2,500 agency. Those overlap. The freelance price sits at the bottom of the agency range; it is not a fraction of it.

They can be, yes. My own work starts from €500, so read this as a warning about my own tier rather than a pitch against it.

The caveat first, because it matters: there is no prevalence data. No neutral source publishes how often cheap builds fail, and every "X% of businesses have had their website held hostage" statistic in circulation traces back to an agency selling website-rescue services. What follows is the shape of the failure, not its frequency.

Three ways it actually costs you:

• Abandonment mid-project. You pay a deposit, the work stops, and you restart from nothing. Bidding boards make this easy to walk into — one French board shows 40 competing quotes arriving within 48 hours, and public profiles on it advertising €5–€18/hour.
• No handover. In the US, the UK and France alike, copyright in commissioned work usually stays with the person who created it unless a written assignment says otherwise. A cheap quote that never mentions ownership can leave you without the right to reuse what you paid for.
• The subscription tail. Hosting, domain and third-party services opened in the developer's name and billed to the developer's card, which you inherit or lose when they stop answering.

None of those are about the price. They are about what the quote does not say — which is why the checklist in "How do I tell whether a quote is fair?" asks about scope, ownership and accounts before it asks about the number.

Check it against six things, and check the number last.

• Scope in writing. How many pages, what is on them, and what happens to anything not listed. In a 2018 survey of 4,455 project practitioners, 52% of projects had experienced scope creep in the previous year — the quote that does not define scope is the one that grows.
• Whose labour, whose bills. Hosting, domain, database and third-party subscriptions should be named, separated from the developer's fee, and opened in your name. An account in your own name is one you keep when the relationship ends.
• Ownership, stated. Copyright in commissioned work stays with the person who made it by default, in the US, the UK and France alike. A US "work for hire" clause does not fix that on its own — website code is not one of the nine statutory categories that clause covers. Only an express written assignment moves ownership.
• Revisions and scope changes, counted separately. Revisions are rounds of changes inside the agreed plan. A scope change is something that was never in the plan, or that reverses it. They are priced differently, and a quote that merges them into one line will surprise you later. Ask which one you are being sold.
• Payment terms. In France, B2B terms default to 30 days and cannot exceed 60 days from the invoice date, so "net 90" is unlawful there. On deposits: no published survey supports "50% upfront is standard". A deposit is normal practice; the percentage is negotiated.
• The number, last. Divide the price by a local day rate — €575/day in France, £525/day in the UK as of 4 August 2026. A five-page site quoted at half a day of work is not a bargain; it is a description of how much work is planned.

Apply all six to my quotes too. If one of mine cannot survive the checklist, do not buy it.

Four starting prices, and each one is a floor rather than a final price:

• from €500 — WordPress from a premium pre-built theme. One page, you control the content from an admin panel, delivered in about a week. Not a shop.
• from €1,000 — WordPress built from scratch and shaped around your business. This is the tier for a shop or a store.
• from €1,200 — a custom Next.js build. Fast and exactly right, and nothing in it is editable — no panel, no content changes.
• from €800 — a custom admin panel added to that Next.js build, if you do need to edit it. On top of the build, not instead of it.

All of them cover my work only. Hosting, the domain and any third-party service are excluded and billed to you directly, in your own name. More pages raise the price. More features raise the price.

For context, two measurements that disagree. In GoodFirms' 2026 survey, 59.9% of the 300+ web development firms surveyed across 31 countries put a basic website at $1,000–$3,000. In France, the average budget clients actually posted for a WordPress project on the bidding marketplace Codeur was €769 in August 2026.

Those are two different populations — firms quoting, and clients posting — in two currencies, unconverted. The gap between them is most of why this question has no single answer.

My work on the build. Design and layout, your copy placed properly, the site working on a phone, and it going live. It also includes two to three rounds of changes inside the plan we agreed.

It does not include anything bought from someone else — hosting, the domain, a database, or any third-party service the site needs. Those are billed to you directly, in your own name, and I never mark them up.

It covers one page. If you need five, that is five times the layout work, and the price says so.

Two different things get confused here, so plainly: rounds of changes inside the agreed plan are included. Changing the plan itself is a scope change — you get one of those free, and after that it is priced and added to the quote. Saying what you want before we start is cheaper than either.

One page. All three build prices — from €500, from €1,000 and from €1,200 — are for a single page, a landing page, and they cover my work on it only. (The custom admin panel is the fourth price and is not a page at all; it is added to a build.)

Most businesses need more than one, which is fine, but it means the number goes up from the floor. Pricing it this way is deliberate. A floor quoted for a "whole site" is a number nobody can hold, because a whole site is five pages for one business and forty for another, and you would find that out at the quote stage instead of on the price list.

Tell me how many pages you need and what each has to do, and you get a real figure before any work starts. In my experience the count comes out smaller than people expect once we go through it page by page — but if it comes out larger, the quote says that instead.

More pages, and features that are not just pages. A second and third page is more layout work. A shop, a booking system, customer logins, or anything that stores and changes data is a different class of build — it needs somewhere to keep the data, and testing that a static page never needs.

The market shows the same steps. In GoodFirms' 2026 survey of 300+ firms, 59.9% put a basic site at $1,000–$3,000, 45.4% put a CMS-based site at $5,000–$20,000, and 35.5% put a custom web application at $20,000–$50,000. Those are firms quoting their own work, and that top band describes agency-built applications rather than a small business's website — read it as the shape of the steps, not as a price list.

Changing the plan after we have agreed it also moves the price, and that is handled separately: one scope change is free, then it is priced. Rounds of changes inside the agreed plan are separate again, and included.

Worth saying plainly: most small businesses never need the second or third step. If a page and a contact form does the job, that is the build.

from €500, for one page, delivered in about a week and a week at most.

What that is: a premium pre-built WordPress theme, set up properly with your text, your branding and your images — never handed over with the demo content still showing. It is meant to look expensive, because a landing page that looks cheap costs you the visitor before they read anything.

You control it. There is an admin panel, and you change the words, the prices and the photos yourself without paying me for a text change.

What it is not: a shop. This tier does not sell things. If you need to take payments, that is the custom build. It also is not a multi-page site — each extra page raises the price. A one-page portfolio fits here perfectly.

The week is real, but it assumes your text and photos arrive when you say they will. Changes on your side move the date, not the price.

The comparisons people ask for: WordPress.com charges a $499 one-time design fee for a done-for-you five-page site on its own platform, on top of annual hosting. In France, the average budget posted for a WordPress project on Codeur was €769 in August 2026. On timing, 71% of firms in GoodFirms' 2026 survey take two to four weeks for a basic site.

If what you want is a small site on a platform that handles everything under one bill, WordPress.com's own service is a straightforward buy and you do not need me for it. What you trade away is control of where the site lives and what it can later become.

from €1,000. This is the tier for a shop, a store, or anything built around how your business actually works.

Custom means built from scratch and adapted to what you do — not a theme someone designed for a generic business, bent until it roughly fits yours. It is the right answer when your list of requirements keeps hitting things a theme will not do. It is the wrong answer, and an expensive one, when it does not.

You control the content here too. The difference from the template tier is what the site can be, not whether you can edit it.

Where that sits against the market: Codeable, which sells WordPress development, starts website design and development at $2,500 and e-commerce at $3,200. In GoodFirms' 2026 survey, 45.4% of firms put a CMS-based site at $5,000–$20,000, though another 15% quote $1,000–$3,000 for simpler builds. On France's Malt barometer, freelance WordPress developers declare €437 a day.

I will say the obvious thing: that floor still sits at or below the published bottom of most custom-WordPress ranges in the research behind this page, and it stays there even without converting between currencies. If a low number reads as a warning sign to you, that is a sound instinct. The real risks with a cheap quote are someone stopping mid-project, handing over nothing at the end, or leaving you with subscriptions you did not know you had. Ask what the price buys, ask what happens if I disappear, and look at work I have already shipped before you decide.

from €1,200 — and at that price the site is fixed.

Read that plainly, because it is the part people miss until it is too late: there is no admin panel, no login, and nothing in the site you can change yourself. It is built, it is fast, and it stays exactly as delivered until you pay someone to change it. For a site whose content is settled, that is a feature — there is less to break, less to maintain, and nothing to keep updated.

If you do need to edit it, that is from €800 on top, for a custom admin panel. Not instead of the build — in addition to it. See the next answer for what that panel actually is.

You should also know the market data here is thin. No barometer or survey publishes a project price for a freelance Next.js build. France's Malt barometer does not index Next.js at all — its front-end list stops at React, at €564 a day. The only published Next.js day rate anywhere is IT Jobs Watch's UK contract median of £523, from 120 advertised contracts, as of 4 August 2026.

So the only honest construction is a day rate multiplied by a number of days, not a published price band. A confident "Next.js sites cost X" is not coming from a measurement, because there is not one to come from.

And if what your site has to do is say what you offer and let people contact you, Next.js is not what you need. Say so, and take the template tier.

It is an admin panel built for your site specifically, priced from €800 on top of a Next.js build. Its whole advantage is what it leaves out.

A WordPress admin shows you everything WordPress can do. Posts, pages, media, comments, plugins, themes, users, settings, and whatever else the plugins you installed decided to add — whether or not your site uses any of it. That is not a flaw in WordPress; it is a general-purpose tool being general. But it is why owners tell me they are frightened of breaking something, and why they end up phoning someone instead of changing a price themselves.

A panel built for one site shows the things that site actually has. If your site has a hero, four services, opening hours and a gallery, that is what the panel lists. There is no menu item that does nothing for you, and no setting that can break a page you did not know existed.

The honest trade: it costs from €800 you would not spend on WordPress, which arrives with an admin included. It is worth it when the site has to be a Next.js build for other reasons and the owner still needs to edit it. It is not worth it as a reason to choose Next.js in the first place — if editing is the priority, take WordPress and keep the money.

There is a working example on this site: the owner of State Auto Glass edits every page, price and photo on his own site from a panel like this, with no content system to license and no invoice to me for a text change.

Because more of it is written and less of it is assembled. A WordPress template hands you a working site to adapt. A Next.js build starts closer to empty — layout, components and behaviour are built rather than configured — so it is more of my time.

Two markets price the same gap. On Malt's French stack table, WordPress is the lowest-paid development stack listed, at €437 a day against €564 for React. In the UK contract market, IT Jobs Watch puts WordPress at £390 a day, from 25 advertised contracts, against £525 for React, from 839, as of 4 August 2026.

Against my own interest: that gap is not universal. On Arc's global freelance marketplace WordPress carries no discount at all — its average sits level with general web development. So do not let anyone tell you WordPress is inherently cheap labour, and do not read the price difference as a quality difference. Costing more is not the same as being right for you. If WordPress does what your site has to do, the cheaper build is the better one.

Fixed price, agreed before I start. You know the number in advance and I carry the risk of it taking me longer than I thought.

Hourly is normal in some markets and there is nothing wrong with it. The closest published evidence on billing culture is The State of PPC 2026 global report — a survey of paid-advertising agencies rather than web builders, so read it as a regional pattern, not as web pricing. It found hourly billing used by 25% of respondents in Continental Europe and 23% in the UK and Ireland, against 12% in North America, and called the billable hour "a cultural preference, not an operational necessity". France's two largest freelance platforms measure the two conventions side by side: Malt publishes day rates, and Codeur publishes project budgets — €1,441 average for a web project in August 2026.

Fixed pricing only works if the scope is fixed too. That is why the plan gets written down first, why one scope change is free and further ones are priced, and why rounds of changes inside the plan are counted separately and included.

When hourly is the better deal for you: open-ended work where nobody can say in advance what is needed — an ongoing fix list, or "make this faster and stop when the budget runs out". A fixed price on work that vague is a fixed price on a guess, and you would be paying for my guess.

No. A deposit at the start, the balance on delivery, and for anything running longer than a couple of weeks we can split it into stages.

The percentage is negotiable, and I will not dress it up as an industry rule. "50% up front is standard" is repeated everywhere and no published survey supports it. AIGA's standard design contract recommends taking a deposit and deliberately names no number. Freelancers Union offers 30/30/40 as, in its own words, "one rough example", for projects over two weeks.

Why a deposit exists at all: IPSE's 2024 survey found 35% of UK freelancers had been paid late in the previous twelve months, with an average of £5,230 outstanding. IPSE campaigns on late payment, so it is not a neutral party to its own number — but it is the only survey figure available.

The staged schedule protects you in the same direction: you never pay far ahead of work that has actually been delivered. And if the deposit is the sticking point, a small first piece of work paid on delivery is a reasonable way for both of us to find out whether this works.

Yes, and it is the easiest version of this conversation. Tell me the number first and I will tell you what fits inside it — or tell you it does not fit, and what would.

Sometimes the answer is that you should not hire me. If the budget stretches to a one-page template site but you actually need a shop, buying a good theme and setting it up yourself is a better use of that money than paying me for half a build you cannot use. I would rather say so than take the job and hand you something incomplete.

What a fixed budget needs is a fixed scope. PMI's 2018 survey of 4,455 practitioners found 52% of projects had experienced scope creep, up from 43% five years earlier — that covers projects of every kind rather than web builds, and it is the closest thing to a measurement I could find. A budget survives that only if the plan is written down, one scope change is free, and further ones are priced when they arrive instead of absorbed quietly until the quality drops.

Cost of living where I am based, and no agency overhead. That is the whole answer and there is no third reason hiding behind it.

I am based in Constantine, Algeria, and I work remotely. Where work is done moves the price far more than how well it is done. Eurostat puts average hourly labour costs at €12.0 in Bulgaria against €56.8 in Luxembourg — a 4.7× spread inside the EU alone. That is whole-economy employer cost, not developer rates, so I am reading its shape rather than quoting a rate. Stack Overflow's 2025 survey puts the median engineering manager on $200,000 in the US against $118,000 in Germany — one identical job title, salaried, self-reported.

The rest is overhead. An agency quote carries salespeople, project managers, an office and margin; a freelance quote carries far less of it. Bark's UK guide — a lead-generation marketplace that publishes no methodology, so treat it lightly — puts freelance small-business sites at £1,000–£3,000 against £2,000–£7,000 for agencies.

It is not that I do less. But it is also not that an agency is the wrong buy: an agency has people who cover for each other, someone to call when one person is ill, and a company that still exists if an individual walks away. That is a real thing to pay for, and if it matters to you it is worth the difference.

No. Hosting is excluded from all three prices and is billed to you directly, by the host, in your own name.

That is not me offloading a cost onto you. An account in your own name is an account you keep when we stop working together. You hold the login, you see the real price, and nobody needs my permission to move the site. If hosting sat on my account, you would effectively be renting your own website from me.

I will tell you exactly what the build needs and what it costs before you commit to anything, so the monthly figure is never a surprise after launch.

The trade is real and you may want the other side of it: an all-in-one platform puts the site, the hosting and the domain on one bill and one login, and for some people that convenience is worth more than the control.

No. You register the domain yourself, in your own name, and pay for it directly — usually a small yearly fee to the registrar.

This matters more than hosting does. The domain is the one asset that stays yours no matter who built the site, and being the registered holder is what gives you control of it. Under ICANN's transfer policy, a registrar must give the registrant the transfer code and remove the transfer lock within five calendar days of a request, and must not refuse solely because there is a dispute over payment — though it may refuse where the registration fee itself is unpaid. Those protections belong to whoever is named on the registration. If that name were mine, they would protect me, not you.

If you have never registered one, I will walk you through it. It takes about ten minutes, and you do not need me to do it for you.

Most one-page sites do not need one, and if yours does not, I will not sell you one. You need a database when the site stores things that change — bookings, customer accounts, orders, stock, or a panel where you edit the site yourself.

When it is needed, it is billed to you directly by the provider, in your own name, like everything else that is not my labour. No markup, and nothing routed through me.

A real example: State Auto Glass has a page per town and a panel where the owner edits every page, price and photo himself. All of that has to be stored somewhere, so yes, that build needs one. The whole site runs on four paid services in total, database included.

They are paid services that each do one job the site would otherwise need built from scratch. In plain terms: one handles user logins and passwords, one stores and delivers your images, one runs the site itself. There are dozens more for email, payments, maps and so on.

You only pay for the ones your site actually uses, and most small sites use very few. A one-page template build typically needs none of them beyond hosting and a domain.

I treat the number of them as something to keep down, not something to collect. Every one you add is another monthly bill, another account that can lapse, and another company that can change its pricing on you. State Auto Glass, which is a substantial build, runs on four in total.

You do, directly to each provider, on your own card. I do not bill them through me and I do not add a margin.

Bundling hosting and services into a single monthly fee is a legitimate way to work and some clients prefer it — one bill, one person to chase. What it costs you is visibility: you cannot see which part of that fee is the service and which part is the margin. I would rather you saw the real number, so you pay each provider what they charge and you pay me for my work.

The practical effect is that your running costs are yours to control. You can change plan, switch provider, or cancel something without going through me. And if you ever stop working with me, none of it switches off.

Because an account in your own name is an account you keep when we stop working together.

That is the whole reason. Everything else follows from it. You hold the logins. You see what things really cost. You can change the card or the plan. You can hand the entire setup to another developer without asking me for anything. If those accounts were mine, ending the relationship would mean untangling your business from my billing.

It matters legally on the domain too. ICANN's transfer policy obliges a registrar to hand the registered holder the transfer code and remove the transfer lock within five calendar days of a request, and not to refuse solely over a payment dispute — with an exception where the registration fee itself is unpaid. Those rights belong to the name on the registration. It should be yours.

The cost of doing it this way lands on you: a handful of accounts to set up, a card on each, and the renewal emails arriving with you instead of me. Some people would rather not deal with that and hand the lot to whoever builds the site. That is a fair trade to make — just make it knowingly.

For a one-page template site: a domain and hosting, and nothing else. For anything larger it depends on what the site does, and I give you the exact figures for your build before you commit to anything, not after launch.

Costs climb when the site starts doing things — storing data, sending email, handling logins, serving a lot of images. Each is a separate provider with its own bill in your name, and I keep the count as low as the build allows. State Auto Glass, a multi-page site with an owner-editable panel, runs on four paid services in total.

One distinction worth holding onto: running costs are not a maintenance retainer. In GoodFirms' 2026 survey, the commonest maintenance retainer band was $500–$1,500 a month, at about 30% of firms, with $1,500–$3,000 close behind at 28% — and only 40.8% of firms offer a retainer at all. If a quote merges running costs and maintenance into one monthly number, ask which part is which.

Close to none, never actually none — you always need a domain and somewhere for the site to live. Past those two, yes: a straightforward site can run on nothing else.

Keeping the count near zero is how I build by default, not a special request you have to negotiate. Every paid service added is a recurring bill you carry indefinitely, an account that can lapse, and a company that can change its pricing on you. A feature that needs one has to earn it.

The real example: State Auto Glass runs on four paid services in total. That is a site with a page per town, a panel where the owner edits every page, price and photo himself, and a paid campaign attached.

The honest limit: sometimes a paid service is the cheaper answer. A small monthly fee for something that already works and is maintained by someone else can beat paying me to build a version of it and then keep it running. Where that is true I will say so, rather than pretend zero is always the target.

Two to three rounds, included in the price. A round covers changes inside the plan we agreed — wording, colour, spacing, swapping a photo, moving a section up the page. You send your notes in one batch, I work through them, and we do that two or three times.

There is no industry standard here. The widely used Contract Killer template, in its 2008 first version, offers "one main design plus the opportunity for you to make up to two rounds of revisions". The AIGA standard agreement names no number at all. So treat 2–3 included rounds as a common convention rather than a measured norm — I could not find a published survey that counts revision rounds in web work.

If you expect to iterate more than that, say so before we start rather than after. Open-ended iteration is a different shape of job, it gets priced differently, and someone who sells it that way may suit you better than I do.

Something that is not in the agreed plan is a different thing entirely, and it is answered next.

A revision changes something we already agreed. A new request adds something we did not agree, or reverses a decision the build already sits on.

Revisions: the headline, the colours, the photo, the button text, the order of the sections. Those come out of your 2–3 included rounds.

New requests: a booking form, a second language, a shop, a page that was not in the plan. Reversing counts too — if we agreed one layout, built it, and you now want the other one, that is a new request even though nothing is being added.

The line is the plan we both signed off, not how big the change looks. A five-minute job outside the plan is still outside the plan, and a long day of tweaks inside it is still a revision — that second half goes in your favour more often than mine.

The first new request is free. After that it is quoted before any work starts. That is the next two answers.

The first one is free — it is the one free scope change, the same allowance described in the answers either side of this one. After that, it is quoted before I touch anything.

Changing your mind is normal and I would rather hear it than watch you live with something you dislike. But it is not a revision, because the work it undoes has already been built. So the first reversal costs nothing, and the ones after it get a price and a date impact before any work starts.

Timing matters more than size. A change of mind while we are still on the plan costs nothing at all. The same change after the build is the expensive version. Say it early.

One comparison you can check yourself: on changes, the 2008 Contract Killer template says "You will be charged the daily rate set out in the estimate we gave you", and names no free one. That is one widely used practitioner template, not an industry standard.

I quote each one before I start it, priced on the time it takes. There is no fixed per-change fee, and nothing gets built until you have the price and the new date and have said yes.

Be careful with anyone who quotes you an "industry standard" figure here, because I could find no source that prices change orders in web work. The two mechanisms that appear in the widely used templates are both open-ended. The AIGA standard agreement charges out-of-scope changes "on a time and materials basis, at Designer's standard hourly rate of ___ per hour". The Contract Killer template charges the day rate from the original estimate. Even AIGA's threshold for triggering a fresh signed proposal is written as changes "in or near excess of ____ percent (____%) of the time required" — a blank you fill in yourself, not a number the trade agreed on.

That cuts against me as much as for me. Time-based pricing means you are trusting my estimate of my own hours. So ask for the number in writing before I start, and ask what happens if it runs longer than I said. The answer should be that it is my problem and not yours — and if a quote you are comparing this against will not answer that, the price is not the thing to worry about.

A short written note saying what you are now asking for, what it adds to the price, and what it does to the delivery date — agreed before the work starts. That is the whole thing. It exists so neither of us has to rely on remembering a conversation from three weeks ago.

I use one for anything outside the agreed plan. The AIGA standard agreement does the same job with a threshold: once changes run "in or near excess of ____ percent (____%) of the time required", the work needs a new signed proposal — and in the template that percentage is a blank, not an industry figure.

For the scale of the problem it addresses: in PMI's 2018 survey of 4,455 practitioners, 52% of projects completed in the previous 12 months had experienced scope creep, up from 43% five years earlier. A change order does not stop that happening. It makes it visible while there is still time to decide about it.

Yes. More pages raise the price, and I quote the addition the same way as the original — before the work, not after. A page that reuses a layout we already built costs less than one that needs a new design.

Often you will not need me for it at all. On the State Auto Glass build, the owner adds and edits pages, prices and photos himself from a panel on his own site — a page per town, without a developer touching it. If you would rather not be paying someone for small changes indefinitely, that is the thing to ask for, and it is worth asking any developer you are talking to whether they will build it that way.

Say so before we start, though. Building for it up front is cheap. Retrofitting it onto a site that was never designed for it costs more than building it in would have.

Then we go back to the plan rather than nudge a layout you do not want — and redoing the direction once is free. It is the one free scope change, and it does not come out of your revision rounds.

Say it plainly and say it straight away. A first draft you dislike usually means we agreed the wrong thing, or I heard the wrong thing, and the slow way to discover that is to keep adjusting the wrong draft.

If it happens a second time, that is a signal about fit rather than about the draft. I would rather say so than take your money for a third attempt.

One thing worth knowing before you ask for several directions at once: in the AIGA standard agreement the designer keeps the rights to rejected work — "Designer retains all rights in and to all Preliminary Works." Mine works the same way, so a draft you did not buy is not yours to hand to somebody else. If owning every concept matters to you, that is a thing to agree and price at the start, not after.

About a week for a one-page build from a pre-built template. Custom work takes longer, and you get the date with the price rather than after it. Changes on your side move the date.

The published evidence on build times is thin, and both figures I could source describe entry-level builds. In a survey of 300+ development firms across 31 countries, 71% said they deliver a basic website or MVP in 2–4 weeks. A French project marketplace's cost guide puts a one-page site at 2–3 weeks.

Both describe roughly the tier my template build sits in, so one week is shorter than either. Neither measures a custom build, and I could not find a source that does — which is why I give you a date for your job rather than a band taken from someone else's.

Because one week is the template tier, and the two-month figure usually is not.

The sourced timeline figures both describe entry-level builds: 71% of 300+ firms across 31 countries deliver a basic website or MVP in 2–4 weeks, and a French marketplace's guide puts a one-page site at 2–3 weeks. Every "8–12 weeks" or "4–6 months" figure I tried to trace led back to an agency or vendor blog with nothing published behind it.

Both numbers describe something real, though. One week is one person building one page from a pre-built template, with nobody to route approvals through. Two months is a custom build with an agency's process around it — and that process buys you things I cannot: cover when the one person is ill, and more hands when the job outgrows one head.

So compare tier to tier, not headline to headline. And if your content does not exist yet, or three people have to sign off on it, two months may be the honest number for your project and one week is not.

Content, nearly always. Text, photos, logo files, the opening hours you want on the page — the build waits on those, and they tend to arrive in pieces rather than all at once.

Second is decisions. If two people have to agree and one of them is travelling, a review that should take two days takes a fortnight.

Third is scope. In PMI's 2018 survey of 4,455 practitioners, 52% of projects completed in the previous 12 months had experienced scope creep, up from 43% five years earlier. Every added thing has a date attached to it, and the date is the part people forget to ask about.

And sometimes me. I underestimate jobs like everybody else does. When that happens the delay is mine to absorb, the price does not move, and you hear about it before the date rather than after.

What rarely causes a delay is the code or the hosting. Any date I give assumes you come back within a couple of days, and it moves honestly when either of us does not.

Four things, and none of them are technical:

• The words, or a clear sense of what they should say.
• The images and your logo, in the best files you have.
• The page list — what pages exist and what each one is for.
• One person who decides. Not a committee that replies in a week.

If a site already exists, add access to it, the domain and the hosting.

If you sell into particular places or particular services, tell me which. That shapes the page list before anything is built — on the State Auto Glass job it is what produced a page per town, decided up front rather than bolted on afterwards.

Missing content does not stop the start. It stops the finish. Send what you have and we fill the gaps as we go. But if none of it exists yet, that is worth both of us knowing early: writing it is a job in itself, it is the thing most likely to move your date, and you may want somebody who does that part properly rather than me filling the holes.

No. Paying more does not compress the parts that take the time.

A week is already near the floor for a template build. The things that stretch a project — waiting on your text, waiting on a decision, a change of mind halfway through — sit on your side of the line, where paying me more changes nothing at all.

What does move a date is cutting scope. Fewer pages. Launch with three and add the rest later. Ship without the form and add it next month. That costs nothing and it works.

If you are against a hard deadline, tell me the date before the wish list, and tell me while a bigger team is still an option. Sometimes the right answer to a fixed date is an agency with three people on it rather than one person working weekends, and I would rather say that now than miss the date for you.

You hear it from me before the date, not after — with a new date and the real reason.

The price does not move because I was slow. A fixed quote is fixed, and my delay is mine to absorb. What I will not do is let a slip go quiet or bury it in a status update, because a date you cannot trust is worse than a later date you can.

If the hold-up is on your side — content that has not arrived, a decision still open — I will say that plainly and early too, while it can still be fixed rather than discovered at the end.

What I do not offer is a penalty clause, and there is no refund for lateness on its own. If your launch has real money riding on one specific day, say so at the start: either we agree something in writing, or you hire someone who sells that guarantee. And if the date matters more than the scope, tell me — I would rather cut scope to hold a date than hold the scope and miss it.

Yes, and it is written into the contract, because it does not happen by itself. Copyright starts with the person who made the work, not the person who paid for it. UK law makes the author the first owner, and the UK Intellectual Property Office states that when you commission someone, the first legal owner is *usually* the person or organisation that created the work and not you, unless you agree otherwise in writing. France and the United States default the same way for a freelance commission.

So the transfer has to be an actual clause. Mine assigns the site to you when the final invoice is paid. That condition is not something I invented — the standard US design-services agreement makes any grant of rights conditional on receipt of payment in full. Pay the invoice and the site is yours on paper.

If a developer answers this question with "of course you own it" and nothing else, ask to see the clause that says so, then read it. A clause that licenses the work to you is not the same as one that assigns it.

Yes, the code as well, and it is assigned to you in writing rather than licensed to you. That distinction matters, because contracts in this trade do split it. The first version of one widely used practitioner template, published in 2008, gave the client the graphics and kept the markup and code with the developer, licensed for that one site only; later versions of that same template changed the split. Separately, one well-known designer has described offering full ownership as a separately priced option, reflecting what it might be worth to the client later.

Both are legitimate arrangements if the price reflects them and you know which one you are signing. A licence limited to one site can be the right deal. I assign instead, and I would rather you knew the difference than took my word for it.

One thing nobody can hand over is third-party parts. WordPress itself, plugins, a bought theme, open-source libraries — they come with their own licences and stay under them. I tell you which parts those are before you pay for any of them.

And in the United States, a "work made for hire" clause on its own does not move ownership of website code. The US Copyright Office lists nine categories of commissioned work that can qualify, and software and web design are not among them. Only an express written assignment does it.

You do, because you register it yourself, in your own name, on your own account. The domain is not in my price and does not go through me. You pay the registrar directly, with your own card. If you would rather not do it alone, I walk you through it on your own account and you keep the login.

The reason is the same one that governs hosting and every other paid service: an account in your own name is an account you keep when the relationship ends. I do not register a client's domain under my name or my business, and I would ask hard questions of anyone who offers to.

Two things then stay with you: renewing it, and keeping the registrar contact details current. If it lapses, getting it back is not guaranteed, and it is not something I can fix afterwards.

Yes. The owner of State Auto Glass edits every page, every price and every photo on his site himself, from a panel on his own site. Nothing to license, nothing to ask me for, no invoice. That is the normal arrangement on my builds, not a paid tier.

How much you can change depends on what gets built, so be blunt at the start about what moves often — prices, opening hours, photos, staff, a seasonal offer. Planning for that at the start is cheap; adding it afterwards is not. What stays a developer's job is layout, structure and how the site works.

Two honest alternatives. If you would rather never touch it, say so and I will build it simpler and cheaper. And if changing the layout yourself is the thing that matters most to you, a hosted site builder fits that better than hiring me does — I would tell you that rather than sell you a build.

No. Editing your own content is not a billable event. The State Auto Glass owner has never had an invoice from me for changing a price or swapping a photo, and he changes both regularly. Handing over the panel is the point — if I charged for text edits, the handover would be theatre.

What is different is work that needs code written: new pages, new features, a build change. Those are quoted before they start, never after.

Two things this answer does not cover, because they are separate and are often wrongly merged:

• Revisions — tweaks inside the plan we agreed, like colour, copy, spacing or a swapped photo during the build. Two to three rounds are included.
• Scope changes — something that was not in the plan, or that reverses it. The first one is free; anything after that is quoted before it starts.

If you would rather not do the content edits yourself, I will make them and bill the time. That is your choice, not a condition of the build.

The test: could you have done it yourself in the panel? If yes, it is free.

You go, and you take everything with you. No exit fee, no notice period, nothing to unwind — the hosting, the domain, the database and every paid service were in your name and on your card from day one. I do not have to release them, because you already hold them. Once the invoices are settled you have the source files, and I will answer the next developer's questions rather than make them guess.

Sometimes leaving is the right call. You may want someone in your own working hours, or a specialism I do not have, and I would rather say so than keep the work.

The harder case runs the other way. If you are arriving from a build where the accounts are in someone else's name, that move is genuinely difficult, and how difficult it is was decided when those accounts were opened, not now.

No, and not because I promise not to. There is nothing for me to hold. Hosting, domain, database and every paid service sit in your name, on your account, paid with your card, from the beginning. I am never the only person with a login.

The mechanism behind these stories is simple: someone else is the account holder, and control follows the account, not the invoice. If you are the account holder, there is no hostage to take.

One honest distinction. Until the final invoice is paid, ownership of the work has not transferred. The standard US design-services agreement conditions any grant of rights on receipt of payment in full, and my contract works the same way. That is a different thing from switching off a live site, and I do not do the second.

Be sceptical of the statistics here too. I could find no independent published figure for how often this actually happens; the numbers in circulation come from firms selling website-recovery services.

Yes, all of them, on final payment. The code, the working design files, the images at full size, and the configuration needed to run it. They go somewhere you control, not into an archive I hold on your behalf. Tying delivery to final payment follows the standard US design-services agreement, which conditions transfer of ownership on receipt of payment in full.

Two limits, said out loud. Third-party parts — a licensed theme, a paid plugin, a stock photo — transfer on their seller's terms, not mine, and I list which parts those are in writing before you buy any of them. And files only help if someone can work with them, so ask any developer, me included, whether you are getting the source or a copy of the finished pages. They are not the same delivery.

The site keeps running, because none of it routes through me. Your hosting account, your domain, your database and your subscriptions are yours. As long as those bills are paid, nothing switches off because I stop answering. You hold the source files from final payment, so another developer has what they need to take it over rather than start again.

That is the real answer, and it applies to every freelancer and every agency, not only to me. The test is not whether someone promises to still be here in five years. It is what breaks on the day they are not. Ask it before you hire anyone, me included — and if that risk weighs heavily on you, wanting a firm with more than one person in it is a reasonable thing to want.

What I cannot offer is a guarantee of being permanently available. Nobody honest can, and a solo developer least of all. What I can do is make sure my absence means finding another developer rather than rebuilding the site.

Constantine, Algeria. I work remotely. Every project runs the same way: remote, in writing, with the finished work delivered into accounts opened in your own name.

Two practical consequences. Constantine is UTC+1 and the clocks here never change, so the gap between us moves only when yours does. And there is nothing to visit — no office, no on-site day. If you need someone who can turn up in person, that is a real reason to hire locally instead.

The site gets built, deployed to your hosting account, and you hold the logins from the day it goes live. For the overlap with your working day, see the timezone answer below.

Yes. All the work is remote, and where a client is based doesn't change how a project runs: the same written agreement, the same staged invoices, the same accounts in your own name.

Distance is priced into this market less than people expect. In freelancermap's 2026 Freelancer-Kompass survey of 5,412 IT freelancers, hourly rates across the German states span about €10 — roughly €85 at the bottom, €95 at the top — and the publisher attributes the narrow spread to remote work. That measures freelancers inside one country, not cross-border hiring, and freelancermap runs a freelance marketplace and publishes the survey itself.

What distance does change is the time of day we can talk, and how you pay. Both are answered below. If the job needs someone in the room — a shoot to art-direct, staff to train face to face — hiring locally is the better call.

UTC+1, all year. Overlap against a normal working day:

• UK — I'm an hour ahead of you in winter, level in summer. The whole day overlaps.
• Continental Europe — level with Paris, Berlin and Madrid in winter, an hour behind in summer. The whole day overlaps.
• US Eastern — five or six hours behind me. When it's 15:00 here it's 09:00 in New York in winter, 10:00 in summer. My afternoon is your morning: several hours, with neither of us keeping odd hours.
• US Pacific — eight or nine hours behind. When it's 18:00 here it's 09:00 or 10:00 in Los Angeles. Enough for a morning call your side, not enough for a shared working day. West Coast projects run mostly asynchronously, and that's worth knowing before you hire rather than after.

Clocks here don't change, so the gap shifts by an hour when yours does. Working hours: [TO CONFIRM].

English, French and Arabic. English is the default for everything — calls, written updates, the agreement, the invoice — unless you'd rather use one of the other two.

For site content it means I read the copy on the page myself instead of pasting in text I can't check. That is a check, not a guarantee: it catches what a careful reader catches, no more.

One limit. I'm not a translator. If your site needs a language none of us reads, hire someone for the words. I'll build the page to hold them, Arabic included — it runs right to left, which is a layout job rather than a translation one — but I won't vouch for wording I can't read.

In writing, in whatever tool you already use — email, WhatsApp, Slack. Calls when a call is faster. No standing weekly meeting.

One rule: anything that changes the plan, the price or the date is confirmed in writing before I act on it. Six weeks later neither of us reliably remembers what was said on a call. Putting change requests in writing is old advice in this trade — the Contract Killer template, published in 2008 and widely copied since, says the designer may ask for requests in writing. It is one practitioner's template rather than an industry standard.

I answer messages within [TO CONFIRM] on working days. If you want daily stand-ups, a shared project board or a named account manager, that isn't how I work — better known before you hire than after.

In stages, against an invoice. A deposit to begin, payments tied to points in the work, the balance at delivery. The schedule is written into the agreement before anything starts, and the transfer method is agreed there too — what is cheap and fast differs by country, so it is settled in writing rather than assumed.

Two things worth knowing. Hosting, domain, database and any third-party service are paid by you, directly, on your own card and in your own name. They never route through me — no markup, no exchange rate set in the middle, no bill you can only settle through me. And a cross-border transfer costs a fee on your side and can take a few days to land; that belongs in the schedule rather than being discovered on the day.

If you're a French business, the terms are not open-ended. Code de commerce article L441-10 sets B2B payment at 30 days from delivery of the service by default, and caps an agreed term at 60 days from the invoice date — or 45 days end-of-month where that is expressly stipulated and not grossly unfair. A "Net 90" term sits outside that cap.

You don't, at the start — and nothing I write here should convince you. So structure the job so that my disappearing costs you as little as possible.

Pay in stages. A deposit, then payments tied to points in the work, so what is at risk at any moment is one stage rather than the whole build. Freelancers Union sets out 30% up front / 30% at first draft / 40% on completion as, in its own words, "one rough example", and suggests milestones for anything running longer than two weeks. The "50% up front is standard" line you'll hear quoted, I could not trace to any published survey — AIGA's standard agreement recommends a deposit and deliberately names no percentage.

Start with something small. See the last answer in this section.

Keep every account in your own name. Hosting, domain, services. On the State Auto Glass build the owner edits every page, price and photo himself from a panel on his own site, and the whole thing runs on four paid services, all in his name. If I disappeared he would lose me and nothing else.

I won't quote a figure for how often developers vanish. Every one I found traces back to an agency selling website-rescue services.

Yes — a written agreement on every job, and an NDA if you want one, both signed before work starts.

The clause worth reading isn't the NDA, though. It's the one about ownership. In the US, the UK and France the default is that copyright starts with the person who made the work, and paying for it does not by itself move it. The UK Intellectual Property Office's guidance, published in 2014, says that when you commission someone the first legal owner is the person or organisation that created the work and not you the commissioner, unless you agree otherwise in writing — and that someone working under a contract for services will usually retain copyright.

In the US a "work made for hire" line alone doesn't fix it either. The Copyright Office lists nine categories of commissioned work that can qualify, and software and web design are not among them. What moves ownership there is an express written assignment.

That is what the published guidance says, not legal advice. The assignment clause is in my agreement — read it, then read the same clause in every other quote you're holding.

Yes, and it's often the right way to start. One page from a pre-built template starts from €500 and takes about a week, unless changes on your side move the date. You end up owning a page, on your own hosting, whether or not we work together again — a deliverable, not a trial run.

What you get for it is information about me: how I answer, whether the date holds, what the work looks like, for the price of the smallest thing I sell.

Two caveats, both against my interest. Splitting a build into two engagements usually costs a little more than doing it in one go, so if you already know the whole scope, going straight at it is the cheaper route. And one page doesn't prove anyone can run a thirty-page build with a shop attached; it proves they turn up.

For context on why smaller pieces stay on the rails more easily: PMI's 2018 survey of 4,455 practitioners found 52% of projects had experienced scope creep in the previous twelve months. That covers projects of every size — it isn't a finding about small ones. The reason to start small is simply that there is less in it to go wrong.

A template is good enough for most small businesses, and I build on premium themes too — that is real work, not a consolation prize. The test is simple: write down everything the site has to do, then check whether any of it falls outside a box the theme already has. If nothing does, take the theme and keep the difference.

My template builds start from €500; a custom build starts from €1,000. That gap is real money for a business that does not need to cross it. The clearest line between them: the template tier is one page and does not sell things. The moment you need a shop, or the site has to be shaped around how your business works rather than around a theme, you are in the custom tier and the price follows.

The same split runs through the market, much wider. WordPress.com — Automattic, selling the same service — charges a $499 one-time design fee for a done-for-you five-page template site, hosting billed annually on top, and starts custom development at $5,000 (WordPress.com). Same direction, bigger jump.

WordPress, for most of the businesses that ask me — and the deciding question is usually editing, not speed.

WordPress costs less to build, costs less to change later, and hands over more easily because it is a standard product rather than a codebase only I know. It comes with an admin panel, so you can change your own content from the start.

Next.js earns its price when the site has to do something a page-and-post system fights you on: a lot of moving parts, real data, something closer to an application than a brochure. But a Next.js build from me starts from €1,200 and is fixed — nothing in it is editable. If you need to change your own content, that is from €800 on top for a custom panel. Add those together before you compare tiers, or you are comparing a site you can edit against one you cannot.

Mine start from €1,000 for custom WordPress against from €1,200 for Next.js.

The day-rate indexes show the same gap. On Malt's French barometer, WordPress sits at €437/day against €564/day for React (Malt, WordPress · Malt, React) — no French barometer lists Next.js separately, so React is the nearest French comparison. In the UK contract market, WordPress medians £390/day against £523/day for Next.js, as of 4 August 2026 (IT Jobs Watch, WordPress · IT Jobs Watch, Next.js).

One caveat, because it cuts against the neat story: that discount is a contract-market pattern, not a universal one. On Arc's global freelance marketplace, WordPress hourly rates carry no discount against general web development at all (Arc). So the saving is less work, not automatically cheaper labour. If nobody can tell you what the extra buys, it is not buying anything.

Nothing, for a lot of people. If you want a decent-looking site up this week, you want to run it yourself, and you do not intend to move it, those platforms do the job and you do not need me. The trade-off is renting rather than owning, and it is honest rather than sinister: you can normally take the domain and your text and images with you if you leave, but the build itself stays behind, because it only exists inside their editor. That is a fair deal if you never leave and a bad one if you do. Shopify in particular is a sensible answer for a real shop with real stock. The one thing I would push on, whichever you use: keep the domain and every account in your own name. An account in your name is an account you keep when the relationship ends.

Usually I would rather fix it, and fixing is usually cheaper for you. A rebuild is the right call when the thing underneath is broken or abandoned: an unsupported platform, a theme nobody maintains any more, a build nobody has the logins for. Short of that, most of what people want fixed is speed, wording, a form that does not send, a layout that falls apart on a phone. That is hours of work, not a project. Published rate cards split it the same way — Codeable, which sells WordPress development, publishes a $420 floor for a speed and security job against a $2,500 floor for a full design and build (Codeable). Send me the URL and I will tell you which of the two you are actually looking at, including when the answer is to leave it alone.

Yes — that is the custom WordPress tier, from €1,000, not the template one. The template tier (from €500) does not sell things.

But check whether you should, first. A shop is not a website with a buy button. It is stock, payments, tax, shipping, refunds, and somebody answering emails when a parcel goes missing. If you sell a handful of products, Shopify or a WooCommerce setup on a template will usually cost you less and get you selling sooner than anything I build from scratch.

The gap shows up inside a single rate card: Codeable, which sells WordPress development, publishes a $3,200 floor for e-commerce against $2,500 for a standard site (Codeable). My own floor is for a shop built from scratch and shaped to your business, and it moves with the number of products and the features you need.

Probably not. Most of the blogs I get asked to look at have three posts and nothing newer than two years ago, and to my eye that reads worse to a customer than no blog at all. What businesses usually need instead is a handful of permanent pages that answer the questions they get asked on the phone every week: what you do, where you do it, what it costs, what happens next. Write those once and keep them accurate. That is not a blog, that is the site doing its job. A blog earns its keep when you genuinely have something new to say on a schedule you will hold. If you cannot name who is writing it in month six, do not start it.

Almost certainly not. Nearly every small business that asks me for an app wants something a website already does — bookings, prices, a contact form, a gallery — and an app adds two app stores, two builds, review queues, forced updates, and a download you have to persuade a stranger to install. A website has none of that, and the browser is already on the phone. An app makes sense when people use it repeatedly and need something a browser cannot do: working offline, a hardware feature, notifications people actually want. If your honest answer is that a customer would open it a few times a year, build the website properly and keep the difference.

Usually because there is nothing to show, or nothing that matches: no Business Profile, or a profile and pages that never say plainly what you do and which towns you cover. What it is almost never is a payment you missed. Google says so directly: "There's no way to request or pay for a better local ranking on Google" (Google Business Profile Help). You can pay for the slots above the unpaid results — those carry a Sponsored label — and for service trades there are Local Services Ads, where you "Pay only for leads related to your business and the services you offer" (Google). Everything below the label is earned, and the same Google page names three things that decide it: relevance ("how well a Business Profile matches what someone is searching for"), distance ("how far each business is from the customer who's searching") and prominence ("how well-known a business is"). You cannot move your premises, so the work sits in the other two — and in my experience most of that is wording rather than spending.

Usually the pages never tell a stranger, fast enough, that you cover their town and handle their exact problem. Being found and being chosen are two different jobs, and you have already done the first one. What breaks the second is nearly always that a visitor cannot see in a few seconds whether you are for them, or how to reach you without hunting for it. So before touching anything, read your own pages as that stranger would: on a phone, in a hurry, with no patience. Reviews matter at the same moment — in a BrightLocal survey of 1,002 US adults, 97% said they read reviews for local businesses and 47% said they would not use one with fewer than 20 (BrightLocal, which sells review software). Often the fix is wording, and a phone number where the thumb already is. That costs nothing.

Ask every customer, the same way, after the job is finished, and offer them nothing. Google's policy names both halves of that. Offering "payment, discounts, free goods and/or services" in exchange for a review is prohibited — and so is choosing who you ask: "Discourage or prohibit negative reviews, or selectively solicit positive reviews from customers." Both are bullets in the same policy (Google). It is enforced at scale rather than on paper: Google says it blocked or removed over 292 million policy-violating reviews and removed over 13 million fake Business Profiles in 2025 (Google). So incentivised and cherry-picked reviews are not a quiet shortcut — they are the category being actively removed, along with the profiles behind them. Ask everyone, offer nothing, reply to all of them.

Usually one of three ordinary things, and the first costs nothing to check. They may be in a paid slot — a Sponsored placement or a Local Services Ad, labelled as such and sitting above the unpaid results (Google). They may simply have been nearer to whoever ran that particular search. Or their listing and pages describe the job in the words customers type, while yours use the words your trade uses. Google names relevance, distance and prominence as the three factors, and review count is only one input into the third — "Prominence means how well-known a business is" (Google Business Profile Help). Distance changes with every person who searches, so the order on your phone is not the order on theirs. The wording is the part anyone can fix.

I will not give you a date, and you should be careful with anyone who does. I have no measured result from past work to base one on, and every version of the "three to six months" figure I tried to trace led back to someone selling the service rather than to a study. What I will do instead is tell you what I changed and when I changed it, in writing, so you can hold it against your own phone log and your own diary and judge it yourself. Some things are quick to fix because they were simply missing — a profile nobody ever filled in, a page that never said which towns you cover — though fixing something quickly is not the same as a result arriving quickly. Other things depend on other people and are slow for that reason. Nobody controls the timing, and nobody should pretend to.

Roughly $1,000–$2,500 a month from a freelancer and roughly $1,800–$3,200 from an agency — but the published figures genuinely disagree, and you should know that before you compare quotes. Ahrefs, which sells SEO software, surveyed 439 providers in August 2024: freelancers averaged $1,348.63 a month, agencies $3,209, and the single most common band was $501–$1,000 (Ahrefs). Clutch's directory, where agencies self-report and which sells them leads, lands almost on the same agency figure at $3,199.19 (Clutch). Then SE Ranking, which also sells SEO software, surveyed 260 agencies and found 64% charging under $1,000 a month (SE Ranking). Those cannot all describe the same market. The likeliest reading is that it is bimodal — a large cheap tier and a smaller expensive one — so a $400 quote and a $3,000 quote can both be normal, and the number alone tells you nothing about quality.

Often no, and this is the answer that loses me work. At a freelance average of $1,348.63 a month — from a survey of 439 providers by Ahrefs, which sells SEO software (Ahrefs) — SEO has to bring in more than that, every month, before it has done anything for you at all. For a one-van trade with a full diary, it usually will not. Cheaper things are usually worth doing first: a listing that is actually filled in, pages that say plainly which towns you cover, and reviews — in a BrightLocal survey of 1,002 US adults, 47% said they would not use a business with fewer than 20 reviews and only 9% would use one with five or fewer (BrightLocal, which sells review software). Spend there first. Paying for SEO makes sense when you have room to take more work, a margin per job that clears the fee, and a service area worth competing over.

Local SEO is the work of getting your business to show up when someone nearby searches for what you sell — the map results and the listings, rather than the whole web. Google publishes what decides that order: relevance, distance and prominence (Google Business Profile Help). Relevance is how well your profile matches what was searched. Distance is how far you are from whoever is searching. Prominence is how well known the business is, and reviews feed into it. Two of those you can influence; one of them you cannot. Most of the rest is filling things in accurately and keeping them accurate — the profile, the hours, the service area, the wording on your pages. Be careful what you pay for here: no independent survey publishes what monthly local SEO should cost, so every figure in circulation is somebody's own rate card, mine included.

Yes, if customers can visit you or you travel to them — it is free, and it is the thing I most often find missing. Google's own local ranking guidance turns largely on the profile: how well it matches what was searched, how far you are from the searcher, and how well known the business is (Google Business Profile Help). You do not need me for it. Claim it, verify it, fill in every field honestly, and keep the opening hours right. One caution: Google polices this hard — it says it removed over 13 million fake Business Profiles in 2025 (Google) — so use your real name, real address and real category, and expect to prove them. If you sell only online and have no local customers, it matters much less.

Yes, most of it — and for a small business the parts you can do yourself are the parts that matter most. You can claim and fill in your Google Business Profile. You can ask every customer for a review and reply to all of them. You can write, in plain words, what you do, where you do it, what a job involves and what it costs, because that is what somebody searching is trying to find out. None of that needs a tool or a subscription, and Google's own documentation is clear that there is no extra technical step you have to buy your way into to be eligible for Search (Google Search Central). What is harder to do alone is the slow, boring part — keeping all of it accurate for years, and noticing what has quietly gone missing. That is what you would actually be paying someone for.

The clearest tell is a guarantee. Nobody can promise a ranking, and Google says it plainly: "There's no way to request or pay for a better local ranking on Google" (Google Business Profile Help). Anyone guaranteeing you the top spot is either describing ads or describing nothing. Four more:

• They will not tell you, in ordinary words, what they are going to do.
• They report impressions and "visibility" instead of calls, enquiries or sales.
• The accounts, the domain or the site are in their name instead of yours.
• They lean hard on new acronyms. Google's Search Advocate John Mueller: "The higher the urgency, and the stronger the push of new acronyms, the more likely they're just making spam and scamming" (PPC Land).

And if anyone offers to sell you reviews, walk away — Google prohibits incentives in exchange for reviews outright (Google). The same goes for anyone promising to delete a bad one: Google's stated position is that it "doesn't get involved in conflict between businesses and customers", and only reviews that break its policies come down (Google).

You can buy the ad slots; you cannot buy the rest. Google sells Sponsored placements, and for service trades it sells Local Services Ads, where you "Pay only for leads related to your business and the services you offer" (Google) rather than paying for clicks. Those sit above the unpaid results and carry a label saying what they are. Everything below the label is ordered by relevance, distance and prominence, and Google states that "There's no way to request or pay for a better local ranking on Google" (Google Business Profile Help). Practically: ads switch on today and stop the day you stop paying. The unpaid side is slower and has no monthly invoice, but it is not permanent either — it moves when your listing goes stale or when someone else does the work. If you need work this month, test ads. Do the unpaid work anyway, because it is the part that does not end with the budget.

There is no submission form and nobody sells the placement — what an assistant says about you is built out of what your public pages already say. Assistants answer from text they can read, so a business that states plainly, in public, what it does, where and for whom gives them something to repeat; one that leaves it to be inferred does not. Google says the same about its own AI features: there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary", and those features run on its ordinary search ranking systems.

Nobody can promise the recommendation itself, and the output is unstable. SparkToro ran 2,961 test runs in late 2025 and found under a 1-in-100 chance that any two answers to the same question name the same set of brands. Being describable is in your control. Being named on any given run is not.

And if your customers arrive by word of mouth and phone calls, this is not the first place I would spend money.

The work behind the name is real; a good deal of what is sold under the name is search work with a new label on it.

The real part: the academic paper that named GEO (KDD 2024) reported that content changes could raise how often a source was cited inside a generated answer by up to 40%. That was measured on a pipeline the authors built themselves, and it measured citation visibility only — no traffic, no sales, no revenue — and the authors warn the methods will need to change as the engines do.

The part to be careful with: the idea that this is a separate discipline. Google's documentation says its generative AI features are "rooted in our core Search ranking and quality systems". Google's John Mueller was blunter on Bluesky in August 2025: "The higher the urgency, and the stronger the push of new acronyms, the more likely they're just making spam and scamming."

So judge it as search and content work, and price it as search and content work. If someone already does that work for you and does it properly, you probably do not need a second supplier for the same thing under a new acronym.

Very little in the work; quite a lot in what you can count at the end. SEO aims to get your page listed so someone clicks it. AEO aims to get your business named inside an answer, where there may be no click at all. In practice the same page serves both — Google says its generative AI features are "rooted in our core Search ranking and quality systems".

The difference shows up in what happens afterwards. Pew Research watched 68,879 real searches by 900 US adults in March 2025 and found that people who saw an AI summary clicked a traditional result on 8% of visits, against 15% when no summary appeared, and clicked the summary's own source link on 1%.

So a mention can arrive without a visit. What that mention is worth to you is the part nobody has measured well yet — which is worth holding on to when someone quotes you a number for it.

No, and Google says so in writing. Its own Google Ads help documentation states: "No, you can't directly target ad placements in the AI Overviews." No assistant publishes a submission fee, a partner list or a price for being named in its answers, so anyone offering to put you in AI answers for a fee is selling something they do not control.

What you can pay for is the work of making a business publicly describable — what it does, where, for whom, at what price — on pages anyone can read. That is not a placement and it is not a guarantee. It is a bet that a clear set of public pages gets used more often than a vague one. Anyone who prices it as a guaranteed outcome, me included, would be lying to you.

If what you need is something you can count this quarter, ads and reviews are easier to measure than this is.

Not so far, on the numbers that exist — but it is already changing the clicks you get.

Direct traffic from assistants is still small. Ahrefs measured 0.17% of the average site's visitors (0.12% of views) arriving from AI chatbots across 3,000 sites in February 2025. Conductor put AI referrals at 1.08% of all traffic across 13,770 domains between May and September 2025. Both companies sell software into this market, so read those as their own panels rather than as the market.

The clearer effect is on Google itself. Pew Research found people who saw an AI summary clicked a traditional result on 8% of visits, against 15% without. What that measures is fewer clicks when a summary appears — none of these sources measured whether people search less overall.

This is a young measurement area and the data is thin. I would not bet a business on any of these figures still holding in a year. If your traffic is falling, the first thing worth establishing is whether fewer people are looking for you at all, or the same number are looking and fewer are clicking through. Those have different causes and different fixes.

Nobody outside the companies building these systems can tell you why one business was named and another wasn't, and the first honest fact is that the result is unstable. SparkToro ran 2,961 test runs in late 2025 and found under a 1-in-100 chance that any two answers to the same question name the same set of brands. Two businesses can be described equally well and get different results on the same afternoon.

What is in your control is whether your public pages say plainly what you do, where, and for whom. An assistant can only repeat what it can read, and a site made of slogans gives it nothing to repeat. Google states there is nothing extra to add for its AI features beyond what it already asks of any page — worth knowing before you pay someone for a technique.

So judge the description. Any single answer you happen to see is one run of an unstable process, in either direction.

Because there is no published market rate for it and no agreed way to check the result.

When I went looking for a rate, there wasn't one — no marketplace, survey or platform breaks this out as a category, so every monthly figure in circulation traces back to an agency's own blog. The most-quoted price guide is WebFX's, which discloses its own method on the page: "We researched the cost of generative engine optimization by scouring data from across the web." It arrives at $1,500–$5,000 a month for a small business.

For comparison, the AI-visibility tracking tools publish their prices openly: Otterly lists $29, $189 and $489 a month (retrieved August 2026). That is the tooling cost. The rest of a retainer is labour, and the labour is search and content work — Google says its AI features run on its ordinary search ranking systems.

There is also no settled way to attribute the result. SparkToro's researchers — whose own sector sells this tracking — wrote that any tool giving a "ranking position in AI" is "full of baloney", because what it claims to measure changes on every run.

None of which makes a retainer automatically a rip-off. Someone doing genuine content and search work is worth paying, and if you have the budget and a real content operation, a monthly arrangement can make sense. The test is whether they can show you what they changed and what it cost, rather than a dashboard number.

Only if the phone needs to ring this month, and only if your pages already answer the question a buyer arrives with. Ads buy attention and nothing else: the day you stop paying, they stop, and you keep nothing. Reviews, listings and the pages you have written stay up. So ads suit a quiet stretch, a new service you want tested, or a season you cannot afford to miss. They do not fix a site that loses people — if a stranger cannot tell in a few seconds that you cover their town and handle their problem, ads pay for more strangers to find that out. Scale matters too: in the State of PPC 2026 survey (1,306 respondents, fielded Q4 2025), freelancers manage around $545,000 a month in ad spend against $2.9M for agencies. If you are spending near the top of that, hire a team, not me.

There is no published market rate for it, and the model most people expect — a percentage of your ad spend — is not the most common one. The State of PPC 2026 survey (1,306 respondents, fielded Q4 2025, and neither a tool vendor nor an agency behind it) found the most popular arrangement is a flat fee set by your monthly spend band: 20% of the 922 people who answered that question. Percentage-of-spend is used by 15% of North American respondents, and North America is where it is most common — the global share is lower. Hourly billing runs at 25% in Continental Europe, 23% in the UK and Ireland and 12% in North America; the report calls the billable hour "a cultural preference, not an operational necessity". What nobody publishes is the fee itself, so the "10–20% of spend" figure you will see quoted has agency blogs behind it and nothing else. And the ad budget is not a management fee — that money goes to Google, on an account in your name.

Ask every customer, the same way, after the job is finished, and offer them nothing. Google's own policy names what gets a listing in trouble: incentives of any kind — payment, discounts, free goods or services — in exchange for posting a review, revising one, or removing a negative one. The same policy bans review gating: discouraging negative reviews, or selectively asking only the customers you already know are happy. Google says it blocked or removed over 292 million policy-violating reviews and 13 million fake Business Profiles in 2025 — the enforcement is real, and it lands on the profile, not on whoever advised the shortcut. The routine I set up for State Auto Glass asks everyone and offers nothing. What that produces is not something I or anyone else can promise you in advance, and it is not work you need to pay anybody for: asking is free, and you are closer to your customers than I am.

No — not a genuine one, and nobody else can either. Google's own help page says it "doesn't get involved in conflict between businesses and customers", and says plainly: "Do not report a review just because you disagree with it or dislike it." What can actually be done is narrower: a review that breaks Google's published policies can be reported, and Google decides. Reporting is not pointless — Google says it blocked or removed over 292 million policy-violating reviews in 2025. But the decision is theirs, not mine, and I will not sell you a removal I cannot deliver. The rest is the reply. In BrightLocal's consumer survey, 80% said they would be likely to use a business that responds to all its reviews. The reply is the part you control, and it sits under the review for as long as the review does.

No. In the US the FTC's fake-reviews rule (16 CFR Part 465) prohibits businesses from providing compensation or incentives conditioned on a review expressing a particular sentiment, positive or negative, and it carries civil penalties of up to $51,744 per violation — that amount is the 2024 inflation adjustment, and the FTC re-adjusts it every January, so the current figure is not this one. Per violation, not per campaign. Google bans incentives under its own review policy, and says it blocked or removed over 292 million policy-violating reviews and 13 million fake Business Profiles in 2025. The wasted money is not the real cost. The listing is — it is what enforcement acts on, and it is where the honest reviews you would have earned anyway live.

Google first, and for most local businesses Google is the only one worth the effort. I should be straight about the evidence: there is no independent research I can point to that compares review platforms against each other, so anyone quoting you a number for what Trustpilot does is quoting marketing. That is thin ground and I will not pretend otherwise. What is measured is that people read reviews for local businesses at all — 97% of respondents in BrightLocal's consumer survey — and that thin profiles put them off: 47% said they would not use a business with fewer than 20 reviews, and only 9% would use one with five or fewer. So use a simpler test: have your own customers ever mentioned Trustpilot? If they have, be there. If nobody has ever asked, a second profile is mostly a second place to look thin. And do not open one expecting it to move your position in Google's unpaid results — nobody can promise that.

You get the site and you get the keys, because the hosting, domain, database and any paid service were in your name from the first day, never mine. Nothing needs my permission afterwards, and nothing switches off when we stop working together — as long as those bills get paid, and since the accounts are yours, that part is yours too. Day-to-day content is yours from the start: on the State Auto Glass build, the owner edits every page, every price and every photo himself, from a panel on his own site, and has not needed me for a price change. What I hand over at the end, and what the first weeks after launch look like, is set out here: [TO CONFIRM]. It is written down rather than promised as "ongoing support", because you should know exactly what you are getting before you agree to it, not on the day you need it.

Yes — and plenty of sites do not need it. My terms are here: [TO CONFIRM]. For context on what the trade does: in GoodFirms' survey of 300+ web development firms across 31 countries, only 40.8% offer a retainer at all, and the two most common monthly bands are $500–$1,500 (about 30%) and $1,500–$3,000 (28%). Those are firms self-reporting their own prices, and they cover development maintenance, not marketing work. A brochure site that changes twice a year does not justify a monthly fee — you would be buying availability rather than work, and you should turn that down, including from me. A shop taking orders, or a site with logins and a database behind it, is a different question, and there a standing arrangement earns its keep.

Tell me and I look at it. Two things are worth keeping separate. If something I built does not do what we agreed it would do, that is mine to fix, and you do not pay twice for the same work. If it broke because a third-party service changed, a plan lapsed, or the site was edited, that is new work and it gets quoted as work. There is a third category that belongs to neither of us: a hosting or database outage is the provider's problem. Because that account is in your name, you can raise it with them directly instead of waiting for me to pass the message along — which is one of the reasons the accounts are in your name in the first place. What I commit to, and how fast: [TO CONFIRM].

Yes, if the hosting plan you are on includes them — backups come from your host, not from me, and that account is in your name. Before you pick a plan, I tell you what its backup arrangement actually is, and I set it up. That distinction matters more than it sounds: a backup sitting in my account is not a backup you can use; one in yours is. How much to spend on it depends on what would be lost. A small site with no database is quicker to rebuild than to insure, and paying extra to protect it is usually money wasted. A shop with orders and customer accounts in it is the opposite, and should be treated seriously. The specific arrangement on your build: [TO CONFIRM].

Yes — and most of it you will not need me for. Prices, photos, opening hours, new wording on an existing page: those are yours, from a panel on your own site. That is how the State Auto Glass build works, and it was built that way on purpose. What does need me is structural — a new page, a new section, a feature that did not exist before. More pages raise the price and extra features raise the price, so each change is quoted as its own small job rather than buried inside a monthly fee you may never use. Terms for ongoing work: [TO CONFIRM]. If the only things that ever change are words and photographs, you should not be paying anybody a monthly fee for it, me included.

Yes. The build is not finished until you can change your own prices and photos without asking me — that is the standard I hold it to, and it is what the State Auto Glass owner does: every page, price and photo, from a panel on his own site. Training runs in English, French or Arabic, whichever you actually think in. The format, and how much of it is included: [TO CONFIRM]. It is worth naming the alternative plainly. If a developer's handover is "email me when you need a change", you have been signed up to a subscription nobody quoted you. That is a business model, not a technical necessity, and you are allowed to ask for the other one — from me or from anyone else.