SEO and Visibility in AI: How to Set a Strategy Without Impossible Promises

The request almost always comes in the same form: “we want to be first on Google,” or “we want to show up when someone asks an assistant about us.” These aren’t goals: they’re pictures of a result, and no one can commit to producing them. A goal is written differently — which people need to find you, for which questions, and what they should do next — and verifiable activities follow from that. This guide exists to make that translation.

Three different ways someone finds you today

• Traditional search: someone types a question and gets a list of results. The searcher chooses, opens, compares.
• The synthesized answer inside the search page: a text summarizes, with a few links. The click may never come, even when your content helped build that answer.
• The conversation with an assistant: the person describes a problem and receives advice, sometimes with company names. Here there’s no ranking position: there’s either a mention, or there isn’t.

These are three behaviors of people, not three techniques of yours. What changes is what you can observe: in the first case, visits; in the second, appearing without a click; in the third, how you’re described when someone asks.

Four different things called “visibility”

Almost every misunderstanding starts here. There are four stages in sequence, and each one can fail on its own.

1. Accessibility: pages can be reached and read. A part of the site visible only after logging in, or built entirely inside the browser, can turn out unreadable to whoever needs to archive it.
2. Indexing: the page has been archived. Being there is the condition, not the result.
3. Appearance: the page is shown for a given question. It depends on the question, who’s asking, where they are, and when.
4. Contact: someone writes or calls you. It’s the only stage that matters to you economically.

Different questions for different stages. “Is the site readable?” has a technical, binary answer. “Why don’t we show up for this question?” has an answer about content and competition. “Why don’t visitors write to us?” has nothing to do with search engines, and continuing to look for the answer there wastes months.

The goal, activity, measure map

A framework like this is filled in during a meeting and supports all the work that follows. The rows are an example of how it’s written.

Questions come before keywords

1. Who you sell to: industry, company size, the decision-maker’s role, geographic area.
2. Which services really matter. Not the full list: the two or three you want to sell most.
3. What questions a person asks in the weeks before buying, including the uncomfortable ones — how much it costs, how long it takes, what happens if it doesn’t work.
4. Where your range ends, because showing up where you can’t serve anyone is a cost, not a result.

The answers to these questions are the pages to write. A keyword list built before having them produces text that answers no one, and that’s recognizable at a glance.

What no one can guarantee

How to set up verifiable work

1. Fix whatever prevents the site from being read. It’s the only part with a clear-cut answer: either it’s in order or it isn’t.
2. Choose a few priority questions, the ones close to the moment of decision.
3. Write the corresponding pages, one at a time, and link them together so a reader can keep going.
4. Keep company information consistent wherever it appears, on and off the site.
5. Set a point in time, not weekly, to look at what’s changed and decide whether to continue or change direction.

Timelines aren’t the same for everyone and can’t be announced at the start. What can be set in advance is when you’ll revisit it and which elements you’ll look at, so the next conversation doesn’t turn into an exchange of impressions.

What this guide doesn’t cover

This is the general setup. Telling apart a site that hasn’t been archived from one that’s archived but doesn’t show up, with the technical checks that requires, is a separate piece of work. The same goes for how to track over time how and when your company is mentioned inside AI systems’ answers.