The gaps you can't see: why search coverage decides who gets found

Here is a question I ask most new clients, and almost nobody can answer: what are your customers searching for that your website doesn't answer?

It isn't a trick. Most owners know roughly what their site says. Very few know what it doesn't say, because there is nowhere obvious to look. Your analytics show you the people who arrived. They have nothing to say about the people who didn't.

That missing half is what I mean by coverage.

What coverage means

Before anyone visits your website, they've asked a question. Sometimes they type it into Google, sometimes they ask an AI assistant, sometimes they ask a friend who then types it into Google. The questions exist whether or not you answer them. Together they make up the demand for what you do.

Coverage is how much of that demand your website actually answers, and how well.

A few things follow from putting it that way.

Coverage is measured against demand, not against your site. A site can be beautifully written, fast and complete in its own terms, and still cover very little, because it describes the business the way the owner thinks about it rather than the way customers ask about it. "Our services" is how a business organises itself. "How much does it cost to frame a football shirt?" is how a customer arrives.

Coverage is not the same as content. Fifty blog posts can cover less than five good pages if the posts answer questions nobody asked. One page can cover a whole group of related questions if it's built around them. The measure is the questions answered, not the pages published.

Full coverage is not the goal. Some questions aren't worth answering: too little demand, too far from a purchase, or already answered so well elsewhere that you'd never be the best result. Choosing not to cover something is a legitimate decision. Not knowing it exists is not.

Why nobody looks at the gaps

If coverage matters this much, why do so few websites manage it? In my experience it's rarely laziness. It's structural. Five things get in the way.

Launch feels like the finish line. A website project has a budget, a deadline and a launch date, and every part of it points at that date. Once the site is live, the project is over. But launch is the moment the site first meets real demand, which makes it the start of the useful information, not the end of the work.

Nobody owns it afterwards. The designer moves on to the next project. The owner goes back to running the business. Updating the site becomes something that happens when a phone number changes or a service is added: reactive, not planned against what people are asking.

Absence leaves no trace. This is the big one. Every tool you're likely to have measures what's there: visitors, pages viewed, enquiries, rankings for the terms you already rank for. A search you never appear for produces no visit, no bounce and no line in a report. A gap doesn't look like a problem. It doesn't look like anything.

Reports measure the wrong half. Even when someone does send a monthly report, it usually tracks positions for an agreed list of keywords. That list was chosen at the start, it reflects what the site already targets, and it rarely asks what's missing. A report can show steady improvement while the site covers a shrinking share of what customers actually ask.

"Content" has a bad reputation. Many owners have been told to blog, tried it, written a dozen posts that went nowhere, and concluded it doesn't work. They're half right. Writing without a map doesn't work. Coverage isn't blogging more; it's answering the specific questions your customers already ask, starting with the ones closest to a decision.

What happens when nobody covers the gaps

The cost of a gap isn't a single lost visitor. It compounds, and it compounds in someone else's favour.

Competitors take the questions you leave. If you don't answer "how long does custom framing take?", someone else does. Their page gets the visits, then the links, then the reviews that mention it. Each one makes their page harder to displace next year.

AI assistants describe you from other people's sources. When an assistant is asked about your kind of business, it assembles an answer from whatever it can find. If your site doesn't state your prices, your process or your specialisms plainly, the assistant uses a directory, a competitor or a forum thread instead, or leaves you out. I've written about this in Your Second Audience. In coverage terms, every gap is a place where someone else gets to speak for you.

Demand moves and the site doesn't. The questions change. New products appear, phrasing shifts, seasons come round, and AI answers absorb some questions completely. A site that covered its market reasonably well at launch covers less of it every month nobody looks.

The eventual fix is a redesign. After three or four years of drift, the site feels dated and underperforming, and the natural response is to rebuild it. Sometimes that's right. But a good share of what a redesign is asked to fix is really a coverage problem: the design is fine, the site just stopped answering the questions people were asking. A new design over the same gaps leaves you where you started, with a newer-looking site.

A worked example, small enough to see whole

Theory only gets you so far, so here is the whole process on a business small enough to fit on a page.

This business is invented. So are all the numbers. I've kept them small and rounded them the way keyword tools do, but they illustrate the method; they are not results from a client or from KOODOS.

Meet Ash Lane Framing: one framer, one workshop, a market town, a website with four pages: home, services, a gallery and a contact page. The services page lists everything they do in a paragraph each. Prices are "on request".

Step one: the Demand Map

A Demand Map starts with what people actually ask, not with the site. For Ash Lane, research turns up a dozen questions worth looking at, which fall into four groups.

Three things stand out before we've looked at the site at all.

The biggest single number, "picture framer near me", is mostly decided by the map listing in Google, not by any page. It matters, but it's a different job.

Price questions add up to more than any other group, and every one of them comes from someone close to a decision.

And the "awkward objects" group is a set of specific, high-intent questions that a general page will never answer well. Nobody searching for how to frame a football shirt wants to read a paragraph about framing in general.

That's what a Demand Map is for. It isn't a list of keywords to rank for. It's a picture of the questions in the market, grouped by what people are trying to do, with an honest view of which ones matter.

Step two: the Coverage Map

Now hold the site up against the map. For each question, the Coverage Map records one of four states:

• Covered a page answers it directly and well.
• Thin the site mentions it, but wouldn't satisfy someone who asked.
• Missing the site doesn't address it at all.
• Answered elsewhere someone else owns the answer (a competitor, a national retailer or an AI summary) and has done for a while.

For Ash Lane, it looks like this:

Two covered out of twelve. That isn't unusual for a site built around what the business does rather than what customers ask, and nothing in Ash Lane's analytics would have told them.

Notice that the two maps do different jobs. The Demand Map is about the market and changes slowly. The Coverage Map is about the site against the market, and it's the one that moves as work gets done.

Step three: deciding what comes first

Ten gaps is too many to close at once, and some shouldn't be closed at all. Each one gets four questions:

1. How much demand? Searches a month, and whether AI assistants are being asked the same thing.
2. How close to a decision? Someone asking a price is nearer to an enquiry than someone asking how prints fade.
3. Can we realistically be the best answer? A museum's conservation guide is hard to beat on its own ground. A local framer's football shirt page is not.
4. What does it take? Rewriting a paragraph is cheaper than photographing new work.

For this month, Ash Lane takes three actions.

Improve: the prices page. Replace "on request" with honest ranges and an explanation of what moves the price: size, moulding, glass, mount. One page now speaks to all three price questions, the group with the most demand and the most intent. It also gives AI assistants a plain statement to repeat, which they can't do with "on request".

Write: framing sports shirts. A new page with photographs, the options, the price range and the turnaround. The competitor holds this question, but with a single page. It's winnable, and every searcher is close to a decision.

Leave on purpose: how to stop a print fading. Real demand, but early in the journey and owned by a museum. Instead, the conservation framing page gets written properly next month, and links to the museum guide rather than competing with it.

That third decision matters as much as the first two. Coverage is about choosing, not about answering everything.

Step four: improve, measure, map again

Here is how the next three months might go. This is still invented. It shows what the cycle looks for, not what it guarantees.

The prices page starts appearing for two of the three price questions. When an assistant is asked what framing costs in Ash Lane's town, its answer now quotes their range instead of a national retailer's.

The shirt page sits on the second page of results. The competitor's page has a year's head start and a few reviews linking to it. It might move with time and a couple of links from local sports clubs; it might not. The Coverage Map marks it thin rather than covered, and it stays on the list.

And the Demand Map has moved. In November, "framing a photo for a Christmas present" appears with real volume, a seasonal question nobody would have planned for in September. It goes on the map, and it competes for next month's three actions alongside everything else.

So the Coverage Map is redrawn: three covered now, a couple improved from missing to thin, one new question added. Then the next three actions are chosen.

That's the whole cycle: observe, prioritise, act, measure, repeat. Nothing in it is clever. What makes it work is that somebody does it every month, against a map, rather than once at launch and then whenever something breaks.

Why this is a decision, not a chore

The usual objection is that this sounds like a lot of work for a small business. Mostly it isn't. Ash Lane's first month was one rewritten page and one new one. What makes it different is that both were chosen against the market rather than guessed at.

The harder part is noticing the gaps at all, because nothing in the normal running of a website points at them. That's the reason to put a map in place: not to generate work, but to make the absent half of your market visible, so the decisions about it get made deliberately.

If you've had a site for a few years and you can't say what it doesn't answer, that's where I'd start. A Demand Map is what GroundWork produces. Keeping the Coverage Map current, choosing three actions a month and reporting what changed is Coverage. And if you'd rather start on your own, the first step costs nothing: write down ten questions your customers asked you this month, then check whether your website answers each of them.