AI Search Research

From Invisible to #1 in 6 Weeks: An AI Visibility Study

By the Ask and Be Found team 7 min read
Short answer

This AI visibility case study tracks a Seattle mortgage broker who went from completely invisible to the #1 AI-recommended broker in his market in six weeks, producing roughly 30 leads and four closed deals. The change came from answer-first content, schema markup, a corrected Google Business Profile, and trusted citations — the same Answer Engine Optimization playbook Ask and Be Found runs in every audit.

When buyers ask an AI assistant for a recommendation, they almost never get ten blue links. They get a name, sometimes two or three, and a short reason why. That is the whole game now: either you are the name AI gives, or you do not exist in that conversation. This study follows one business across the line from one state to the other, and documents what actually moved the needle.

The subject is Keith Akada, a mortgage broker in Seattle. When we started, he was invisible — ask ChatGPT, Perplexity, or Google AI for the best mortgage broker in his area and his name never came up. Six weeks later he was the first broker those same engines named, and the visibility turned into roughly 30 leads and four closed deals. Below is what we changed, why each change mattered, and what the pattern tells us about Answer Engine Optimization in general.

The starting point: invisible by default

Invisible is the normal state for most local professionals, and it is rarely about quality of service. Keith was a strong broker with happy clients. The problem was that nothing about his web presence was readable or verifiable to an AI assistant trying to make a confident recommendation.

When we ran the baseline audit, the pattern was familiar. His site spoke in marketing language — “your trusted lending partner” — but never answered the concrete questions buyers actually type. There was no structured data telling engines who he was, where he worked, or what he offered. His Google Business Profile was thin and partly out of date. And he had almost no presence on the directories and review platforms that AI leans on to confirm a business is real.

To an answer engine, that adds up to a source it cannot quote and cannot trust. So it does the safe thing: it names a competitor who is easier to verify. Across the audits we run, this is the most common reason a good business never enters the answer.

How we measured AI visibility

Before changing anything, we set a baseline we could re-test. AI visibility is not a single rank; it is whether you are named, where, and how you are described across the assistants buyers actually use.

Our method for this study was deliberately simple and repeatable:

  • We wrote a fixed set of buyer prompts — variations on “best mortgage broker in Seattle,” “who should I use to refinance in Seattle,” and similar.
  • We ran them weekly across ChatGPT, Perplexity, Google AI Overviews, and Gemini.
  • We recorded three things each time: whether Keith was named, his position in the answer, and the description the engine attached to him.
  • We tracked citations and direct inquiries that referenced an AI assistant, so visibility could be tied to real demand.

Week zero was clean: zero mentions across every engine. That gave us an honest line to measure against.

The six-week AEO playbook we ran

The work was not exotic. It was the core Answer Engine Optimization checklist applied carefully and in order. Four moves did the heavy lifting.

1. Answer-first content

We rewrote and added pages so each one opened by answering a real question in the first two sentences — “What credit score do you need for a mortgage in Washington?”, “How long does pre-approval take?”, “Is a broker cheaper than a bank?” AI assistants quote clear, self-contained answers. Pages that bury the answer under a sales pitch give the engine nothing to lift. This was the single biggest lever, and in our experience it usually is.

2. Structured data

We added schema markup so engines could read, without guessing, that Keith was a mortgage broker, where he operated, and what services he offered. Structured data does not write the answer for the AI, but it removes ambiguity about whether the source is the right kind of business in the right place.

3. A corrected Google Business Profile

We completed and cleaned up his Google Business Profile — accurate name, address, phone, hours, categories, and service area, all matching the site. AI engines treat the profile as a verification signal. When it conflicts with the website, confidence drops; when it agrees, the business reads as real.

4. Reviews and trusted citations

Finally, we built consistent citations on the directories AI tends to trust and put a simple system in place to earn fresh reviews. Reviews and citations are the social proof that tips an engine from “this business exists” to “this business is worth recommending.”

What changed, week by week

The movement was faster than many people expect, because the underlying pages were finally legible to machines. Here is the shape of the six weeks.

WindowWhat we observed
Weeks 1–2New answer-first pages and the corrected profile get indexed; first scattered mentions appear in Perplexity.
Weeks 3–4Keith is named across multiple engines, often mid-list, with descriptions pulled directly from the new pages.
Weeks 5–6Reviews and citations mature; he becomes the first broker named for the core prompts. Leads and closings follow.

By the end of week six, the same prompts that returned nothing at week zero returned Keith first — and the description engines attached to him was accurate, because it came from content he actually controlled. The visibility produced roughly 30 leads and four closed deals in that window. For more on the indexing speed behind weeks one and two, see our look at how fast AI engines pick up a new page.

Why the order mattered

People often want to start with reviews, because reviews feel like the obvious trust signal. In our testing, reviews matter — but they amplify a page; they do not replace one. If the answer-first content and schema are not in place first, reviews are pushing a source the engine still cannot read clearly.

That is why we sequence the work the way we did: make the business legible, then make it trusted. Answer-first content and structured data are the foundation. Reviews and citations are the lift. Run them in the wrong order and the results come slower and shakier. Reviews specifically are worth their own look — we dig into the mechanics in our study on whether reviews move AI recommendations.

What the engines were actually doing

It helps to be precise about the mechanism, because “the AI started recommending him” can sound like luck. It was not. Each assistant we tested follows the same broad logic: it interprets the buyer’s question, retrieves sources it can read and trust, and then composes an answer that names the strongest match. Every step in that chain is something you can influence.

At the retrieval step, an engine has to find a page that plainly addresses the question. Our answer-first rewrites gave it exactly that — pages where the first sentence was the answer, not a preamble. At the trust step, the engine weighs whether the source is a real, verifiable business. The corrected Google Business Profile, the matching schema, and the directory citations all fed that judgment. And at the composition step, the engine pulls a description; because Keith’s pages now contained accurate, specific language, the description it produced was his, not a competitor’s framing of his category.

Put differently, we did not trick the engines. We removed every reason for them to skip a qualified local broker, and we gave them clean material to quote. When you do that, the recommendation is not a fluke — it is the predictable output of how these systems work. To see how that retrieval-and-trust logic plays out across professional services more broadly, our state of AI search for professional services in 2026 covers the wider pattern.

What this means for your business

The exact numbers in this study belong to one broker in one city, and we would not promise anyone the same lead count. What travels is the pattern. The conditions that made it work are not unusual:

  1. A defined local market where the buyer question is specific.
  2. A real service with genuine clients who can leave honest reviews.
  3. A willingness to publish plain, useful answers instead of marketing copy.

If those conditions describe you, the same four moves are available to you. The reason most competitors have not done them is not cost or difficulty — it is that the playbook is still new, and AI search is moving faster than most local marketing has caught up to. That gap is the opportunity, and it will not stay open indefinitely.

The headline of this study is the speed, but the real takeaway is simpler: invisibility is usually a fixable problem, not a verdict on the business. When a strong local operator becomes readable and verifiable to AI, the engines tend to do what they are designed to do and recommend the best available answer. Make yourself that answer, and the recommendations follow.

Want to see if AI is recommending you? Get a free AI visibility report.

Run My Report →

Frequently asked questions

How long does it take to show up in AI search?
In the audits we run, the first answer-first pages and a corrected Google Business Profile often get picked up by AI engines within one to three weeks. Becoming the consistently recommended name usually takes four to eight weeks, because reviews and citations need time to accumulate. Six weeks is a realistic window for a focused local business in a defined market.
What is an AI visibility case study?
An AI visibility case study documents how a specific business changed its answer engine optimization, then measures whether AI assistants like ChatGPT, Perplexity, and Gemini started naming it. A good one shows the starting point, the exact changes, and the measured result so the work can be repeated rather than guessed at.
Why was the business invisible in AI search to begin with?
Most invisible businesses have thin, sales-focused pages that never answer the questions buyers actually ask, no structured data, an incomplete Google Business Profile, and few citations on the directories AI trusts. AI assistants cannot recommend a source they cannot read or verify, so the business simply never enters the answer.
Which AEO change made the biggest difference?
Across our audits, the single biggest lever is answer-first content that directly answers a buyer’s real question in the first two sentences, backed by schema markup. Reviews and consistent citations amplify it, but they have little to recommend if the page itself does not contain a clear, quotable answer.
Can these results be repeated for other businesses?
The pattern repeats more reliably than the exact numbers. A defined local market, a real service, and a willingness to publish honest answer-first content are the conditions that travel. The timeline and lead volume vary with competition and review velocity, so we treat six weeks as a benchmark, not a guarantee.
How do you measure AI visibility?
We prompt the major assistants with the questions real buyers ask, such as best mortgage broker in a given city, and record whether the business is named, in what position, and with what description. We re-run the same prompts weekly to track movement and watch for changes in citations and direct inquiries from AI referrals.

Keep reading

Become the answer AI recommends

We make your business the name ChatGPT, Claude, Gemini, and Google AI give when buyers ask.

Book a call