What AI says when buyers ask about your category

The Signal Harbor Journal

Research and perspectives from Signal Harbor

By Sebastian Miller
Co-Founder, Signal Harbor

What AI Says When Buyers Ask About Your Category

A potential customer or lead describes a problem to an AI assistant like ChatGPT. It suggests a few companies, explains the differences, and points to sources. By the time that buyer visits a website, they may already have a shortlist.

So how do you find out whether your company makes it onto that list?

Signal Harbor starts with the questions a buyer might ask. We test the answers, find the gaps, and show marketing teams what to improve and how. The AI search industry has plenty of names for parts of that work. Here is what those names mean in practice.

Prompt research: Which buyer questions should you test?

A prompt is simply the question or instruction someone gives an AI assistant. “What is the best software for managing inventory across several warehouses?” is a prompt.

Prompt research is the work of identifying useful questions to test. Signal Harbor studies a company’s products, customers, marketing, sales conversations, and buying process. We use that information to build realistic questions for the people the company wants to reach, often called its ideal customer profile or ICP.

Someone exploring a problem asks differently from someone comparing two providers. That is why we test questions across the buyer journey, from early research to a purchasing decision.

You may hear these called persona-based prompts, buyer-intent prompts, prompt clusters, or a prompt library. Those terms describe ways to organize the same basic work: testing questions that reflect how buyers make decisions.

These are researched test questions. They are not a record of every question real buyers type into AI or how often each question is asked.

AI visibility: Does your company make the answer?

Signal Harbor runs those questions across AI platforms and examines the responses. Does your company appear? When does a competitor appear instead? Is your company described accurately? Does it show up for broad category questions, or only when the buyer already knows its name?

That is AI visibility. Prompt tracking means repeating those tests over time. Competitive benchmarking means comparing your results with those of companies a buyer is likely to consider alongside you.

Some platforms call the resulting percentage share of voice or use a visibility score. Such a number describes performance within the prompts tested. It is not a measured share of every AI conversation happening in the market. Different tools also calculate these scores differently.

The distinction matters. A company might score well on questions containing its name while disappearing from questions like “Which inventory platform works best for a growing distributor?” Signal Harbor tests both branded and unbranded questions because the second group reveals whether your company enters consideration before the buyer knows whom to search for.

Citations and query fan-out: What did the AI use to answer?

When an AI answer cites a webpage, that citation tells us which source appeared in the response. Looking across answers reveals which pages, publishers, and competitors the AI draws on. The market often calls this source mapping.

You may also hear query fan-out. That means an AI system breaks a question into related searches while preparing its answer. Where a platform exposes those searches, Signal Harbor captures the queries it reports alongside the response. That gives us a partial view of the searches behind an answer, not a complete search history. A citation alone does not tell us which searches the AI made.

This helps us investigate a practical question: Is your company missing because its information is hard to find, because the available sources favor competitors, or because the AI has an incomplete picture of what it offers?

Prompt volume: What is measured, and what is estimated?

You may see a tool advertise prompt volume or intent volume: a number suggesting how many people ask a particular AI question. Treat that number carefully. AI platforms generally do not provide a complete public count of the questions their users ask. A volume figure may be an estimate built from other signals, such as traditional search data.

Signal Harbor measures what the AI returned for the questions we ran. That does not establish how frequently all buyers ask those questions. We use the buyer journey and business context to decide which questions matter, rather than presenting an estimated volume as a count of observed AI searches.

SEO, AEO, and GEO: What can you actually improve?

SEO means making a website easier for traditional search engines to find and understand. AEO (answer engine optimization) and GEO (generative engine optimization) are names the market uses for improving how a company is represented in AI answers.

In practice, the work often overlaps: clear product pages, specific claims backed by evidence, useful answers to buyer questions, and credible sources beyond a company’s own site. Google says its existing SEO guidance remains relevant to its AI search features; there is no special markup that guarantees an appearance.

Signal Harbor identifies where the tested answers leave your company out, misstate its strengths, or rely on weak information. We turn those findings into a content strategy: priorities for the pages, claims, and evidence your marketing team or agency publishes. We help shape that material and test again after changes go live.

That is what influencing AI visibility means here: improving the information available to these systems and measuring whether their answers change. No one can promise that an AI platform will recommend a particular company.

The useful question: Did the change help?

An AI visibility report is a starting point. Signal Harbor establishes a baseline, identifies a focused improvement, and reruns the relevant buyer questions after the change. Did your company enter more answers? Did its description become more accurate? Do competitors still own the questions closest to a purchase?

When your team has AI referral data, lead source information, or sales opportunity data, we use those signals to connect visibility changes to business results. Visibility alone does not prove that a new lead or sale came from an AI answer.

The point of all this terminology is a simple decision: when buyers ask AI for help, where does your company show up, where does it get left out, and what should your marketing team improve first? From there, the question for executives is whether those improvements bring in more qualified leads, grow revenue, and make marketing and sales spending more effective.

A deeper layer: Use your customer and website data to sharpen the research

Once that baseline is in place, Signal Harbor brings more of your business data into the work. Your CRM shows customer segments, sales questions, objections, and won or lost opportunities. Google Search Console shows which Google searches surface your site and which pages receive clicks. Google Analytics shows how visitors arrive and what they do on your site, including conversions your team tracks.

Together, those signals help us focus the next round of prompts on your ideal customers and the decisions that lead to a sale. They also give us a stronger basis for examining attribution: how changes in AI visibility relate to referral traffic, leads, sales pipeline, and revenue.

This data does not reveal every question buyers asked an AI assistant, and no single report proves that an AI answer caused a sale. It helps us ask better questions, make a more grounded estimate of business impact, and focus the next round of work where it matters most.

At Signal Harbor, we use this process to help marketing teams identify where AI leaves their company out of buyer recommendations, what evidence is missing, and what to change first.

Sebastian Miller
Co-Founder, Signal Harbor
signalharborai.com
sebastian.miller@signalharborconsulting.com
402-306-2213