Let search intent shape the site

Search shows you what people want before they ever reach a site, and most sitemaps get planned without looking at it. Then someone sprinkles keywords on at the end, and the pages still don't match what people were trying to do.

Intent is what the person is trying to get done. Keywords are the words they happened to type. Two people can type nearly the same words and want different things. Two others can type completely different words and want the same thing. So you plan pages around intent, and you don't need a page for every way of phrasing it.

Before you plan anything, search the queries yourself, in a private window. Read the results for what kind of thing comes up, more than for who ranks. Here's 'sourdough near me', drawn, with made-up results. A map of nearby bakeries with their hours means people want somewhere near them, open now. 'Bakery open Sunday' brings the same map, with today's hours. That person wants one fact: is it open, and when. 'How to make sourdough' brings recipes, guides and videos. That person wants to bake it themselves. And the AI-written answer at the top already walks through the steps, so read it carefully: it shows you what a generic answer covers. 'Order bread for pickup' brings ordering pages, with times and prices. That person is ready to choose.

Here's the bakery's query map. 'Sourdough near me' and 'bakery open Sunday' are people looking for somewhere nearby: that's the Visit page, with the hours, and Today's bread. 'Order bread for pickup' is someone ready to act, and it points straight at Order ahead. 'How to make sourdough' gets no page at all. The bakery sells bread; it isn't a recipe site. A query map is also a list of what not to build.

Now the AI answer. For 'sourdough near me', the summary at the top may already name a few bakeries and tell people to check the hours. A page that repeats that adds nothing. The page worth making is the one only this bakery can write: today's bread and what's left, the sourdough ready at 7, orders closing at 6 pm the day before. Google's own page on its AI features says there are no special requirements for appearing in them. The usual fundamentals apply, and useful, specific content is one of them.

Where do the queries come from? Start with the customers' words you collected, because people search in their own words: 'I order the rye for Saturday' is a query waiting to happen. Add what callers ask the owner on the phone. Type the first few words into a search box and read what it suggests, and note the related questions the results page shows. A model can help you sort a long list by intent, but check its sorting against real results pages. I'm not giving you search volumes here. For a small local site, I'd rather answer ten real questions well than chase one big number.

Then turn the map into structure. Every intent you're keeping becomes a page type, or a section of one. Finding the shop becomes Visit. Ordering becomes Order ahead, one tap from Today's bread. Group related pages together, so the site is organized the way the questions are. That's the hand-off to the sitemap.

Titles, descriptions, readable addresses, how a site gets found and crawled: that's all in Google Search Central's SEO Starter Guide, and I'd trust it over any summary, mine included. This lesson is the part that comes before it: deciding which pages should exist at all.

Plan pages from intent. Build answers only this business can give.