“How long will you need?” is the first question asked in any first meeting, and the honest answer always comes as a surprise: how long a web project takes depends far less on your supplier than on you. On a standard brochure site, the production work accounts for about a third of the schedule; the rest is waiting, deciding and writing. Here are the real durations stage by stage, some benchmarks by type of project, and the three things that send a schedule off the rails.
The six stages of a web project, and what each one takes
The figures below are for a brochure site of eight to ten pages, run without incident, with a responsive client. They are given in working days and as a share of the total schedule, because it is that share which holds steady as the project grows.
| Stage | Share of the schedule | Typical duration | Who holds the pen |
|---|---|---|---|
| Scoping and site structure | 8% | 3 to 5 working days | Both of us, over 1 to 2 meetings |
| Content and editorial decisions | 35% | 2 to 5 weeks | The client, for the most part |
| Visual design | 15% | 5 to 8 working days | The agency, 2 rounds |
| Build and development | 25% | 8 to 12 working days | The agency |
| Acceptance testing and fixes | 12% | 5 to 8 working days | The client tests, the agency fixes |
| Launch and final settings | 5% | 1 to 2 working days | The agency |
The most interesting row is not the longest one, it is the second. Clients almost always assume that visual design is the bottleneck: in reality it takes a sixth of the schedule, and the agency gets through it in a week and a half. What stretches projects out is writing the copy and settling the editorial decisions, two tasks that can only be delegated in part.
Benchmarks by type of project
Every range below rests on an assumption about how available you will be. Without that assumption a quoted duration means nothing, which is why we write ours down beside each range.
| Type of project | Realistic timescale | Assumption about your availability |
|---|---|---|
| Single presentation page | 1 to 2 weeks | Copy and photographs already in hand |
| Brochure site of 5 to 10 pages | 5 to 8 weeks | Copy delivered within two weeks, one decision maker |
| Rich brochure site, content still to produce | 8 to 14 weeks | Copywriting and a photo shoot to schedule |
| Store, small catalog (under 50 items) | 8 to 12 weeks | Product pages and images supplied, carriers chosen |
| Full store (several hundred items, options, customer accounts) | 12 to 20 weeks | Catalog exportable, stock management settled |
| Redesign with content migration and redirects | 8 to 16 weeks | An inventory of the existing addresses is feasible |
| Client area or custom application | 10 to 24 weeks for a first version | Scope written down and settled before we start |
Two remarks these ranges do not show. First, a redesign is never faster than building from scratch: it adds the inventory of addresses, the redirect plan and the migration of the existing content, which is two to three weeks more. The seven signs that you need a website redesign set out that extra time. Second, the budget question and the schedule question are not answered together: what a professional website really costs depends on scope, not on speed.
The three things that send a schedule off the rails
The content that never arrives
This is cause number one, by a very long way. A project announced at six weeks that runs to twenty two almost never has a technical problem: it is waiting on copy, on usable photographs, on a Kbis extract (the French certificate of company registration) for the legal notice, or on an up to date catalog. The parry fits in a single sentence: hand over a page by page outline at the start, with a target character count and a delivery date. And accept this uncomfortable rule: decent copy delivered on the agreed date beats perfect copy delivered six weeks later. We rework it during acceptance testing, when we can see it in place.
Sign off spread too thin
Five people give an opinion, none of them decides, and every round of feedback contradicts the last. The project stops making way; it just rolls. The parry: one contact with the authority to decide, feedback gathered into a single document, once per stage, and a 48 hour turnaround agreed from the outset. Internal opinions are collected before they are sent on, not during.
The scope that creeps
The famous small feature added along the way. Taken on its own it costs two days; repeated five times it costs a month, and nobody notices until the end. The parry is not to refuse: it is to keep a log of out of scope requests, dealt with after launch. A request written down is not a request refused, and the schedule stays readable.
Six moves that save weeks
- Prepare the content before the start. A shared folder with the copy, the full resolution photographs and the legal documents is three weeks saved.
- Appoint one contact with the authority to decide. One person who settles things, not a committee that comments.
- Gather the logins on day one. Domain registrar, hosting, email, search console, analytics. Those five get hunted down in a panic on 90% of projects, always at the worst possible moment.
- Supply usable photographs. The originals, not screen grabs pulled off a social account. Reckon on 2,000 pixels wide at the very least, or the shoot has to be done again, and a shoot takes two to three weeks to arrange.
- List three sites you love and three you loathe, and say why. On average that list removes one full round of visual design.
- Book two sign off slots in advance in the calendar, one after the design, one during acceptance testing. A meeting looking for a date is eight days lost.
Acceptance testing and launch: the phase everyone underestimates
Acceptance testing is not proofreading. It is walking the site as a customer would, not as its owner, and trying to catch it out. Reckon on half a day for a brochure site, two days for a store, and do it with two people.
- Fill in every form and check that the message really lands, junk folder included.
- Test on a real phone, not by shrinking the browser window.
- Check the legal notice, the terms and conditions, the privacy policy and the consent banner.
- Read the copy aloud: mistakes show up to the ear before they show up to the eye.
- On a store, place a full order with a real payment method, then refund it.
- Check the most visible accessibility points, contrast and keyboard navigation. The French general accessibility framework, whose pages are in French, is dealt with at scoping, not the night before launch.
Testing rushed through does not save time, it moves it: the faults nobody saw come back as incidents within the following six weeks. As for the launch date, it is a choice: never a Friday at 5pm, never the day before a trade show, never while you are away. A Tuesday morning, early in a quiet month, costs exactly the same and spares a good many sleepless nights.
The special case of the rush job
“I need it in three weeks” is a perfectly admissible request, provided you know what compresses and what does not.
Compresses: the number of pages, how much the site has to do, the number of design rounds, the degree of visual customization. Does not compress: producing the content, acceptance testing, the propagation of a DNS zone when you change host (up to 24 to 48 hours depending on the cache durations configured), and the approval of an online payment account, which commonly takes the banking provider 3 to 10 working days to process.
The right answer to a rush is therefore not to go faster, it is to surface in stages. A first version, short and serious, in two to three weeks, three or four pages written properly, then something added every month. The result is online on the date you wanted, it works, and it does not carry the scars of a project done in a hurry.
Where we stand: the project does not stop at launch
The first weeks after launch are part of the project, and we say so up front. There are always adjustments to the copy, indexing to follow in the search console, someone to walk through the controls, and the first audience figures to read together, usually after a month.
We prefer that rhythm to the usual cycle of a grand redesign every five years, which costs a lot, ties everyone up and produces a site already dated by the halfway mark. A site that receives one improvement a month stays young indefinitely, and every change is decided on figures rather than on an impression. That is the principle behind a website on a monthly subscription for small businesses, and it is what our website packages cover, hosting and follow up included.
The timescales set out above are the ones we hold to: the full sequence is described on the page for our web agency in Pau, south west France.
The detailed schedule for a local project is on the website design in Pau page.
Frequently asked questions
Can you have a site in a week?
Yes, if it is a single presentation page and the copy and the photographs already exist. No for a complete site: the acceptance testing phase alone, done seriously, takes several days, and it does not compress.
Why did my last project take a year?
Almost always because the content never came and because sign off had no single decision maker, rarely because of the technology. Go back through the message history: the silent stretches often run longer than the productive ones.
Is a redesign faster than building from scratch?
No, at equal scope it takes longer. It adds the inventory of the existing addresses, the plan of permanent redirects and the migration of the content, which is two to three weeks more.
When should you start writing the copy?
Before the visual design, from the scoping stage onwards. A mockup built on dummy text always has to be redone when the real content arrives, because the real lengths never match.
What to take away
A web project does not take the time it costs, it takes the time the client needs to decide. Prepare your content, name a decision maker, book your sign off slots: those three moves are worth several weeks. The next useful step fits into a one hour meeting to set out your schedule, with dates on the table before any commitment.
Also worth reading: how to choose a web agency
