Scoping a project when the idea is still loose

How to scope a website project, when the idea is still loose, starts by naming what you do not know. Write a first slice that will teach the rest. A plan that pretends the whole product is already defined will slip the date the first time reality disagrees with it.

A loose idea is a normal place to start. Founders often have a name, a dissatisfaction with the current site, and a sense that the product is hard to explain. That is enough to begin a conversation. It is not enough to price “the platform.” The scope should match the knowledge you actually have.

Name the unknown

Make a short list with two columns: known, and not known yet. Known might be the audience you already sell to, the offer you can say in a sentence, and the page that is embarrassing you. Not known might be whether you need accounts, whether the shop has to launch with the marketing site, or which of three products is the one a stranger should see first.

Write unknowns as questions. “Do first-time buyers need an account before they pay?” is a question. “User accounts, phase one” is a feature pretending to be an answer. Features that sneak in from the question list become the scope you did not mean to buy. Leave them as questions until a real task cannot be finished without them.

Say which unknown blocks the launch, and which can be learned after something is live. A missing headline blocks a homepage. A future integration with a tool you might buy next year does not. The scope gets sharper when those two are not in the same sentence.

Google’s guide to creating helpful, people-first content includes a useful test for a slice, even though it was written for publishers. After someone uses the page, will they have learned enough to do the thing they came to do? A first version that only presents intentions fails that test. A first version that explains the offer and lets a person take the next step passes it. You can borrow the test without turning the project into a search program.

Scope a slice that can teach you

A slice has one audience, one path, and a place the result lives. For a company site, that is often the homepage, the offer, a proof point, and a way to talk to you. For a product, it is the single workflow a new user came to finish, including the empty state and the error, not a map of every screen you can imagine.

The slice should be buildable by a small team and readable by a stranger. If you cannot demo it without apologizing for the missing half, it is still a deck. Cut until the apology goes away. You can add the second audience when the first path has taught you what people ask.

Hanabi’s work for founders often looks like this kind of edge: a site that explains the company, a shop rebuilt around the catalog that matters, or a product interface with one job. Got Next and Leadly, in the public work, are examples of focused product UI. They are not a template for your scope. They are a reminder that a smaller surface can be the whole first engagement. Down the Cove’s ecommerce migration is the other reminder: even a store gets clearer when you decide what comes along and what stays behind, instead of moving every plugin because it was there.

What you postpone should be written down. A blog you will not write. A design system for a team of one. Localization. A second product. A members’ area. Call the list “later,” and keep it out of the estimate. Later is allowed to become next. It is not allowed to become “included” halfway through because it was mentioned once on a call.

What a change does to the date

A loose idea becomes expensive when every new thought is treated as already purchased. The scope should say what happens when the unknown resolves into more work. A second audience, a new checkout, or a product line that appears in week three is a change. The date moves, or something else leaves the slice. Both are legitimate. Pretending the date can hold is how the launch becomes a quarrel.

You do not need a complicated change-request ritual. You need a sentence you and the studio agree on: new pages or new flows are estimated before they are built. Verbal “while you’re in there” is the usual leak. Write the decision down in the channel you already use. A short note beats a memory of a call.

Price the slice, not the dream. If the dream is useful, attach a rough range and label it as a range. False precision on an unknown is a different kind of loose thinking. A range with a list of what would move it is honest. A single number for “the whole platform, we’ll figure it out” is a wish.

What the first document contains

Keep the scope to a few pages, or a single one.

• Who the slice is for, in one sentence
• The action it has to earn: a call, a purchase, a completed task
• The pages or screens in this slice, and the states that matter: empty, error, success
• What is explicitly out
• What you will own at the end: repository, domain, content logins, brand files
• How a change affects the date

References can sit in an appendix: two sites you respect, and one sentence on what you respect about them. A folder of fifty screenshots is not a scope. It is a mood, and moods are how extra pages sneak in.

Access belongs in the same document when you have it. The current site, the shop admin, the analytics, the person who can approve copy. Missing access is a common way the first week disappears. If you do not have a login yet, the scope should say who is getting it, and that the schedule assumes they succeed.

When the idea is too loose even for a slice

Sometimes the offer itself is the unknown. You cannot scope a homepage that explains a product you cannot yet describe. In that case the first slice might be a naming and message pass, or a single landing page that tests one sentence, not a full site. Ship the sentence. See if a stranger understands it. Then scope the site that can carry it.

Do not hire a build to avoid the sentence. A studio can help you find the sentence. It should not invent a large information architecture to hide that the sentence is missing. Say the gap on the first call. A good intro call will tell you whether a proposal for a slice is worth writing, or whether you should wait.

If you want to do that with Hanabi, book an intro call and bring the two columns: what you know, and what you do not. The useful outcome is a slice, not a performance of certainty.

When the work is specifically a marketing site and the pages are already clear, the next piece is more concrete. How a marketing site gets scoped so the date survives builds the scope from pages and states. Use this article until those pages are real. Use that one when they are.