Planning a website with AI
Planning a website with AI means deciding which pages should exist and how they connect, before any of them are written. It is the highest-value use of a tool on a website project and the one most people skip entirely.
Planning a website with AI means deciding which pages should exist and how they connect, before any of them are written. It is the highest-value use of a tool on a website project and the one most people skip entirely.
Short answer
Using AI to plan a site means producing a list of pages worth having, grouped into sections, with one clear purpose for each. The plan is where a site is won or lost. A good plan built slowly beats a large site of pages nobody decided on, and the plan is far cheaper to change than the pages.
A page written well in the wrong place still fails. It competes with a sibling, answers a question nobody asked, or sits where no reader or crawler will reach it. None of that is fixable by editing the page.
Structure decides what a site can ever rank for, and what a reader or an answer engine can ever find. Our page on site structure covers the shape, and the shape is decided at planning time.
Changing a plan costs an afternoon. Changing a built site costs redirects, lost links and weeks of reindexing. That asymmetry is the whole argument for spending real time before anything is written.
Producing the candidate list. Ask what questions somebody hiring your trade in your area would have, and you get a long list quickly, including several you would not have thought of yourself.
Grouping. Taking sixty candidate subjects and arranging them into sections with a parent page each is exactly the kind of organizing work that is tedious by hand and easily checked once done.
And spotting overlap. Asking which of these pages are really the same question phrased differently catches the duplication that our page on competing pages describes, before it is built.
Which questions are commercially real. A tool cannot tell which of sixty subjects brings in work, because that lives in your experience of which conversations turn into jobs.
What you will not cover. A plan is defined as much by exclusion as by inclusion, and a tool asked for pages will always produce more pages. Somebody has to say no to most of them.
And whether you can honestly write each one. A page you cannot fill with something specific should not be on the plan, however good the subject looks. Our page on thin content explains why.
The plan below is what we would want before writing a single page. Each line is a decision, and having them written down is what stops a site drifting into a collection of whatever seemed worth adding that month.
Read the page titles alone and ask whether any two are the same question. If they are, merge them. That single check prevents the most common structural fault on small sites.
Then check that every page has a route in. A page with nothing linking to it is invisible however good it is, which our page on orphan pages covers in detail.
Last, ask what a visitor is meant to do on each page. A page with no next step is a dead end, and a plan full of them produces a site that reads well and converts nothing.
Build the pages that earn money first, then the hub pages that hold them together, then the educational material that feeds them. That order keeps the site useful at every stage rather than only at the end.
Resist building the whole plan at once. Ten good pages linked properly outperform fifty thin ones, and the plan is not a commitment to build everything on it this year.
Leave the hub pages populated rather than empty. A section page listing three items reads as a section under construction, which is a poor impression to give a reader who arrives there first.
Treat it as a document that gets updated rather than a thing produced once. Every new page should be added to it deliberately, which is what stops the structure decaying as the site grows.
Revisit it when customer questions change. The plan was built from what people asked, and what people ask moves, particularly after a change in your services or your market.
The plan is also the answer when somebody asks what the site is for. Being able to point at a page list with a purpose beside each entry is a better answer than anything a homepage can say.
Plan it, then build a part of it. The plan should cover where the site is going, so structure decisions are made once. Building everything on the plan immediately is rarely wise and often produces thin pages.
As many as there are real questions worth answering well. Ten strong pages beat fifty thin ones. Add pages when a genuine question appears that no existing page answers properly, not on a schedule.
No. It can produce candidates and group them sensibly. Which subjects turn into paying jobs lives in your experience of customer conversations, and that judgment has to come from you.
Building several pages that answer the same question in different words. They compete with each other, split the links between them, and read as filler. Merging them at the planning stage costs nothing.
15-day free trial. Card required. Cancel before day 15 and you pay nothing.