Deciding what not to build, and why that is the hard half
The short version
The short version. Decide what to leave out of a website by putting four questions to every item on the wish list: who will use it in the first months, who will keep it up to date, what a buyer loses if it is missing at launch, and how hard it is to add later. Items that fail go on a written exclusions list with the reason and the event that would bring them back, and that list is signed with the plan before design starts.
In this piece7 sections
- The short version
- How does scope creep get into a website project?
- Which questions expose a page nobody will maintain?
- Why is a deferred feature cheaper than a half built one?
- What should never be on the exclusions list?
- How do you write the exclusions down so they survive the project?
- What happens when someone asks for an excluded item halfway through?
- Start your own list this week
- Five things to keep
- Questions we hear often
Deciding what to leave out of a website comes down to four questions asked of every item on the wish list: who will use it in the first months, who will keep it up to date, what a buyer loses if it is missing at launch, and how hard it would be to add later. An item that fails those questions goes on a written list of exclusions, with the reason and the event that would bring it back, and that list is signed before anyone designs a page.
Most owners arrive with the opposite of that list. The wish list has grown over months of conversations, competitor visits and well meant suggestions, and every line on it sounds reasonable on its own. That is exactly why the cut is the hard half. Adding a page takes one sentence. Removing one takes a reason somebody will accept.
How does scope creep get into a website project?
Scope creep is the slow growth of a project past what was agreed, one small addition at a time, without the budget or the timeline moving to match. On a website it almost never arrives as one large request. It arrives as a page that "only needs a paragraph", a form field someone in sales would like, a second language "for later", or a section copied from a competitor that nobody can explain.
Each addition is small, so nobody stops to decide on it. Together they change the project. The site needs more copy than anyone is writing, more photographs than exist, more review rounds than were planned, and a launch date that moves twice. The same pattern shows up on rebuilds, where every page on the old site is carried across by default because deleting one feels like a decision someone will be blamed for.
The cause is rarely a difficult client. It is a plan that listed what goes in and stayed silent about what stays out. Silence reads as maybe, and maybe gets built.
Which questions expose a page nobody will maintain?
The most useful question about any page is not what it should look like. It is who will keep it true. A page that goes out of date becomes a liability, because it tells buyers something that stopped being true, and they rarely tell you what they read.
We put the same short test to every item on the list, in this order, and an item has to pass it to be built at launch.
The leave out test, one item at a time
- Who uses it in the first three months, by name or role, not by segment
- Who keeps it up to date, and how often that has to happen
- What a buyer loses if it is missing on launch day
- How hard it is to add later, without redesigning what is already built
- What evidence asked for it: an enquiry, search data, or an opinion in a meeting
- What it pushes out of the budget if it stays in
The second question removes more items than any other. A news section with no one to write for it, a team page for a team that changes every quarter, an events calendar for events that are announced on WhatsApp anyway. Each looks complete on launch day and starts going stale the week after.
The fifth question matters on rebuilds. The site and search audit that comes before the plan shows which pages already earn visits and enquiries. Those have evidence behind them. A page that exists only because the old site had it does not.
Why is a deferred feature cheaper than a half built one?
A deferred feature costs nothing until you choose to build it, and by then you know more. A half built one costs its build, then its upkeep, and then the work of either finishing it properly or taking it down without breaking anything.
Online booking is a typical case. Built at launch with no clear rules about who confirms a booking and how fast, it produces bookings nobody answers, which is worse for a buyer than a clear "call us" button. Deferred with a trigger, such as "add booking once enquiries reach a level the team cannot answer by phone", it arrives when someone is ready to run it.
This is the same idea product teams use when they fix the time they are willing to spend and let the scope flex to fit, which the chapter on setting boundaries in Shape Up, Basecamp's book on product development (opens in a new tab) explains well. Government service teams go further. The UK government Service Manual's guidance on the discovery phase (opens in a new tab) treats stopping before anything is built as a legitimate result of research, not a failure. A website is smaller than a public service, but the principle carries: finding out early that something should not be built is a saving, not a setback.
Which items to defer and which to decide on now depends on two things, how much buyers need the item at launch and how hard it is to add later.
Build now, defer or decide now
Horizontal axis from Easy to add later to Hard to add later; vertical axis from Few buyers need it at launch to Most buyers need it at launch.
- Most buyers need it at launch, Easy to add later: Build now, in its simplest form
- Most buyers need it at launch, Hard to add later: Build now, and build it properly
- Few buyers need it at launch, Easy to add later: Defer, with a written trigger
- Few buyers need it at launch, Hard to add later: Decide now: drop it, or leave room for it
The bottom right corner is where most expensive surprises come from. A second language is the clearest example. It may not be needed on launch day, but if it will ever be right to left, the layouts have to allow for it from the first sketch. That item is not deferred quietly. It is decided now, either dropped or designed for, and the choice of stack follows from that decision.
What should never be on the exclusions list?
Some things look optional on a wish list and are not. Cutting them saves a little at the start and costs more later, often in ways that are hard to undo.
- Accessibility. We work to level AA of the W3C Web Content Accessibility Guidelines 2.2 (opens in a new tab) on every build. It shapes colours, type, forms and navigation from the first design, so it cannot be added as a final pass.
- Redirects for pages that already earn visits. On a rebuild, every old address that brings in buyers needs a new home. Leaving the redirect map out to save a day can remove pages that took years to rank.
- Legal and privacy basics. A privacy notice that matches what the site actually collects, and a cookie approach that matches what it actually loads. The detail differs between India, the UAE and the UK, so take advice on it, but do not leave it off the plan.
- The answer to the question buyers ask first. If buyers always ask about fees, process or timelines before they enquire, the page that answers it is part of the minimum, not an extra.
How do you write the exclusions down so they survive the project?
An exclusion that lives only in a meeting is forgotten by the second design review. One that is written down with a reason survives, because anyone who wants to reopen it has to argue with the reason rather than with a memory.
We write each exclusion as one line with four parts. Here is how a typical list reads for a service business site:
| Item left out | Why it is out | What would bring it back | Who decides |
|---|---|---|---|
| Blog | No one is named to write for it | A named writer and two months of topics agreed | Owner |
| Online booking | No rule yet for who confirms a booking | Enquiries the team can no longer answer by phone | Operations lead |
| Live chat | Nobody is free to answer it during the day | A person or a reception system assigned to it | Owner |
| Arabic version | Not needed for the first market | A decision to sell into the UAE, taken before design | Owner |
The third column is the one that does the work. It turns a no into a not yet, with a condition that anyone can check. When the condition is met, the item comes back as a planned piece of work with its own cost, not as a favour squeezed into a review round.
The list sits on the one-page brief beside the pages that will be built, and it is signed with them. On our own projects that signature comes before anyone opens a design file, and the studio page sets out the rest of how we work. If a feature will not earn back its cost, we would rather say so while it is still on paper.
What happens when someone asks for an excluded item halfway through?
They will, and that is normal. The request is a change to the plan, so it is handled as one: what it costs in time and money, what it would push out, and who has to agree. If the answer is yes, the plan is updated and signed again. If the answer is no, the request is logged against its trigger so the thinking is not lost.
What matters is that the request is recorded somewhere both sides can find it later, not settled in a chat thread. We keep requests and approvals in a client portal for that reason, and the case for it is in why every website request should be logged.
Start your own list this week
Take your current wish list and put each item through the test above. Mark every item that has no named user, no maintainer or no evidence behind it. Then write the exclusions as four column lines, with a reason and a trigger for each, and show the list to whoever will approve the finished site.
The items they argue for are the ones worth discussing before any money goes into design. Our services all begin at this step, and you can book a call once the list is on paper.
Five things to keep
- Scope rarely grows in one big request. It grows one reasonable sounding page at a time.
- Ask who will maintain a page before you ask how it should look.
- Defer what is easy to add later. Decide now on what is hard to add later.
- Never cut accessibility, redirects for pages that already earn visits, or legal basics.
- Write each exclusion down with its reason, its trigger and its owner, and sign it with the plan.
Built on Intent studio. A founder-led studio in India that plans and builds websites, search and AI visibility, and AI systems for businesses in India, the UAE and the UK. Reviewed by Srikar, founder. We do not publish invented numbers; where a figure appears, its source is named in the sentence.
How the studio works
