A small business website is typically quoted at four to eight weeks with a professional, or one to two weeks if you build it yourself. Very little of that is construction. Most of the calendar is waiting: for content, for feedback, for a slot in someone’s queue.
The published ranges are broadly right and worth knowing. What none of them explain is where the weeks actually go, which is the part that decides whether a quoted timeline is honest or padded.
What the published timelines say
| Site | Do it yourself | With a professional |
|---|---|---|
| Brochure, 1 to 5 pages | 1 to 3 days | 1 to 2 weeks |
| Small business, 5 to 15 pages | 1 to 2 weeks | 4 to 8 weeks |
| Small ecommerce, under 50 products | 3 to 6 weeks | 8 to 12 weeks |
| Large ecommerce, 500+ products | 6 to 9 months | 4 to 8 months |
| Custom application | not realistic | 6 to 12 months+ |
Figures as published by Elementor, September 2026. Other guides converge on the same ranges: HubSpot and Hostinger both land in the same place for a small business site.
So the number to hold in your head is four to eight weeks for a normal small business site built by somebody else. That is the honest market answer and we are not going to argue with it.
Where the weeks actually go
Here is the part that is missing everywhere else. Take a six week project and split the calendar into time somebody is working on your site versus time your site is sitting still.
| Phase | Calendar | Actual work | What the gap is |
|---|---|---|---|
| Discovery and planning | 1 week | 4 to 6 hours | Scheduling two calls around two diaries. |
| Waiting for your assets | 1 to 2 weeks | 0 hours | Logo, photos, copy, access. Nothing can start. |
| Design | 1 to 2 weeks | 1 to 2 days | Mockups, then waiting for sign off. |
| Build | 1 to 2 weeks | 2 to 4 days | Genuine construction. This is the real work. |
| Revisions | 1 to 2 weeks | 4 to 8 hours | Feedback arriving in pieces over days. |
| Launch | 2 to 3 days | 2 to 4 hours | Redirects, DNS, testing. |
Our own breakdown of where calendar time goes on a standard small business build, based on ten years of running these projects.
Add the right hand column up and a six week project contains roughly five to eight working days of actual work. The rest is queue, diary and inbox.
When someone quotes you six weeks, they are usually not quoting six weeks of effort. They are quoting the shape of a normal project with normal waiting in it. That is not dishonest, but it does mean the timeline is mostly negotiable if the waiting is removed.
The single biggest predictor of how long a website takes is not the developer. It is how fast the client sends the photos.
I have had a five page site sit untouched for nine weeks waiting on a logo file, and I have had the same site go live in four days because the owner had everything in one folder before we started. Same work, same person doing it, a nine week difference in calendar time. If you want a fast website, the lever is almost entirely on your side of the table.
How to make a normal build fast
Whoever you hire, these four things compress the timeline more than anything the developer does.
- Have the assets ready before day one. Logo in a vector file, real photos, team names and bios, a plain list of your services. One folder, one link, sent before the first invoice.
- Write down the three sites you like and why. Not a mood board. Three URLs and one sentence each on what you like. This replaces a week of design exploration.
- Give feedback once, in one document. Seven emails over five days is the slowest possible way to do one round of revisions.
- Get the access sorted early. Domain registrar, hosting, Analytics, Search Console. Chasing credentials at launch is the most common way a launch date slips.
A rebuild is not the same as a new site
Almost every published timeline quotes the figure for a site built from nothing. If you already have a site, four things get added to the calendar, and only one of them is design work.
- The redirect map. Every old URL needs to point somewhere sensible. A fifteen page site is an afternoon. A site with eight years of blog posts is a week, and skipping it is how businesses lose rankings they spent years earning.
- Deciding what not to keep. This is a judgement call, it needs the owner in the room, and it is the step people underestimate most.
- Whatever the old site was quietly doing. A form wired into a spreadsheet, a booking widget, a tracking script somebody added in 2019. Finding these takes an hour. Finding them after launch takes a fortnight.
- The DNS and email check. Moving hosting can take the email with it if nobody looks first. Rare, and unforgettable when it happens.
Add one to two weeks to any published figure for a rebuild, and ask a prospective supplier what their plan is for the redirects specifically. The answer separates people who have done this from people who have only built new sites. What a rebuild typically costs, with prices from 87 agencies, is in what a website redesign actually costs.
What a quoted timeline should contain
A date on its own is not a timeline. If a quote says “live in six weeks” and nothing else, there is no way to tell whether it slipped because of you or them. Ask for these five, and the vague quotes separate from the real ones immediately.
| Ask for | Why it matters |
|---|---|
| What starts the clock | Signature, deposit, or assets received. These can be weeks apart and only one is in your control. |
| What they need from you, and by when | A dated list. This is the single biggest cause of slippage and it should be written down. |
| How many review rounds, and how long each window is | “Unlimited revisions” has no end date, which means no launch date either. |
| What happens if you miss a deadline | Does the launch date move, or does the slot move? A slot moving can cost a month. |
| What is excluded from the estimate | Content writing, photography and migration are the three usually assumed and rarely stated. |
None of that is adversarial. It is the difference between a timeline you can hold somebody to and a number that made you feel better on the call.
Where the estimate usually goes wrong
Four patterns, in rough order of how often we see them.
- The content nobody costed. The quote assumed you would supply copy. You assumed they would write it. Two weeks vanish while that gets resolved, and the resolution is usually a change order.
- The approval chain that appeared late. A business partner, a spouse, a board. Not a problem if it is known on day one, and a fortnight if it surfaces at review two.
- The one small extra thing. A booking form, a members area, a calculator. Each is reasonable alone and each resets the estimate.
- The migration nobody looked at. Old sites have surprises: a form wired into something odd, a subdomain, four hundred URLs that need redirecting. Worth twenty minutes of inspection before a number is given.
“Has anything like this gone over for you before, and what caused it?” A supplier who answers that specifically has learned something. A supplier who says it has never happened is either very new or not listening.
Why we say seven days, and what it costs you
Our own promise is seven days from your assets arriving, and it is worth explaining the mechanism rather than just asserting it, because a fast number with no explanation should make you suspicious.
| What we do | What it buys | What you give up |
|---|---|---|
| One kind of site only | No discovery phase deciding what to build | No shops, booking engines, member areas or portals. We decline all of them. |
| Our own theme, no page builder | The build is assembly, not construction | The layout system is ours. You are not getting a blank canvas. |
| The clock starts on assets, not signature | The waiting happens before the promise, not inside it | The seven days genuinely do not start until everything has arrived. |
| Three review rounds in a fixed window | Feedback cannot sprawl across weeks | One consolidated round of feedback counts as one revision. |
| AI assisted drafting, human direction | Copy and structure in hours rather than days | Nothing, in our view, but we disclose it rather than hide it. |
The full process, including what is out of scope, is on how it works.
Read that right hand column properly. Seven days is not a better version of the six week project. It is a narrower product. If your list of exceptions is long, the six week agency is the correct choice and you should take it.
Speed is the least interesting thing about a website and it is the thing people ask about first. A site built in seven days that nobody finds is worth exactly the same as one built in seven weeks that nobody finds.
The reason I care about the seven days is not the speed itself. It is that a build which takes two months of somebody’s attention has to be invoiced as a project, and once it is invoiced as a project the relationship tends to end at launch. That is the actual problem with long builds.
Common questions
How long does a small business website take to build?
Four to eight weeks is the standard professional quote for a five to fifteen page site. One to two weeks if you build it yourself on a site builder. Most of that calendar is waiting rather than working.
Why do some agencies quote six weeks and others quote one?
Usually scope and queue rather than skill. A shorter quote generally means a narrower product, a template system, and a slot already reserved for you. Ask what is excluded, not just how long it takes.
What slows a website build down the most?
Content and assets from the client, by a wide margin. The published guides agree on this and so does our experience. Nothing can start until the logo, photos and service list arrive.
Is a faster build a worse website?
Not inherently, but speed always comes from somewhere. Usually a narrower scope or a reused system. Find out which, and decide whether the thing being given up matters to you.
Can I keep my existing content to speed things up?
Yes, and it helps, though it is worth being honest about whether the old copy was working. Reusing content that never generated an enquiry saves a week and costs you the reason you rebuilt.
Before you start any of it
Find out whether a machine can currently read the site you already have. Nine checks, ten seconds, no email.