What Happens After an AI Visibility Score

The Signal Harbor Journal

Research and perspectives from Signal Harbor

By Sebastian Miller
Co-Founder, Signal Harbor

What Happens After an AI Visibility Score

When we started Signal Harbor, the first problem was clear: companies had little idea how ChatGPT, Gemini, Perplexity, Google AI Overviews, and other AI platforms described them to potential buyers.

So we built around measurement. We wanted to know where a company appeared, where it was left out, which competitors were recommended instead, and what information shaped the answer.

That is still our foundation. What has become clearer is what marketing teams need from the measurement.

The industry has different names for pieces of this work: prompt research, LLM visibility, competitive intelligence, citation analysis, AEO, GEO, and attribution.

At Signal Harbor, we call the complete process AI recommendation intelligence. In plain English, we find out who AI recommends to your buyers, why your company wins or loses consideration, what your marketing team should improve, and whether the answers change afterward.

Prompt research: We choose the buyer questions that matter

A prompt is a question given to an AI assistant. Prompt research, also called prompt discovery, is the work of deciding which questions to test.

Signal Harbor studies a company’s products, customers, and buying process. We build a prompt library, also called a prompt index, around its ideal customer profile, or ICP.

Consider a hypothetical Chicago manufacturer of precision metal components. A purchasing team starts by asking where to source a custom part. It then compares suppliers on specifications, capacity, quality, and reliability.

Those questions follow the buyer journey, from identifying a need to choosing a provider. We test branded prompts, which name the manufacturer, and unbranded prompts, which ask for options without naming it.

These are researched test questions. They are not a log of every question real buyers ask AI. Prompt volume is an estimate of demand, not a complete count of buyer conversations across AI platforms.

AI visibility: We measure who makes the shortlist

Signal Harbor runs those questions across AI platforms and examines the answers. Does your company appear? Is it recommended for work it actually does? Which competitors appear instead? Is its description accurate?

This is AI visibility, also called LLM visibility or answer engine visibility. Inclusion rate measures how often your company appears in the answers we test. Share of voice or share of model compares that presence with competitors. These figures describe a defined set of questions, platforms, and tests. They are not your share of every AI conversation in the market.

We examine brand sentiment, or the tone of an answer, separately from accuracy. AI can describe a company positively and still get an important capability wrong.

Comparing those answers is competitive benchmarking, also called competitor research or competitive intelligence. For our manufacturer, the test can expose a specific gap: AI knows the company when asked about it by name, but recommends a competitor when a purchasing team asks for suppliers with a capability the company also offers.

That gives marketing a specific buyer decision to investigate. We prioritize the gap according to the customers it affects, how close they are to a purchase, and whether it appears repeatedly.

Source intelligence: We find what shaped the answer

Signal Harbor investigates the evidence behind the response. Which webpages did AI cite? Which outside sources appeared? What information supports the competitor’s recommendation?

This is source mapping, also called source analysis, citation analysis, or source intelligence. A citation is a source shown in an AI answer. A citation is not the same as a recommendation: AI can cite your webpage while telling the buyer to choose a competitor.

We also inspect query fan-out where an AI platform reports it. Query fan-out refers to the related searches a platform makes while preparing an answer. The searches it reports give us a partial view of that process. A citation alone does not reveal them.

We use this evidence to identify whether the gap comes from unclear positioning, missing proof, buried capabilities, outdated information, or stronger sources supporting a competitor. The marketing team receives an explanation it can act on.

Content gaps: We tell marketing what to improve

The industry calls missing or weak information a content gap. When a competitor’s sources appear and yours do not, you will also hear citation gap or source gap.

Signal Harbor turns those findings into content and evidence priorities. For a manufacturer, the priority is specific: clarify a capability, make specifications easier to find, show credible proof of work in a target industry, or correct information AI repeatedly gets wrong. We help shape the work so an internal marketing team or agency knows what to publish or improve.

This work sits across SEO, AEO, and GEO. SEO focuses on traditional search. AEO, or answer engine optimization, focuses on direct answers. GEO, or generative engine optimization, focuses on how a company appears in AI-generated responses. Together, these efforts are often called AI search optimization.

The terms sound complicated. The work is clear: make the company’s capabilities easy to find, easy to understand, accurate, and supported by evidence.

Every gap does not require a new campaign. We focus the team on changes connected to buyer decisions the business wants to win.

Prompt tracking: We measure what changed

After a team publishes an improvement, Signal Harbor tests the relevant buyer questions again. This is prompt tracking, also called AI visibility monitoring.

We compare the new answers with the baseline, or first measurement. Did the company enter more relevant recommendations? Did its description become more accurate? Did the competitor gap narrow?

AI answers vary, so we examine patterns across questions, platforms, and measurement periods. We give the marketing team a continuing view of what changed and where to focus next.

Attribution: We connect visibility to business evidence

Marketing teams need clear direction. Executives need to know whether the work is helping the company reach more qualified buyers.

When a company has the data, Signal Harbor brings in context from its CRM, Google Analytics, and Search Console. Customer segments, sales questions, search queries, website activity, leads, and opportunities sharpen the research and help us assess the result.

This is attribution: examining how marketing work relates to business outcomes. AI referral traffic records identifiable visits from AI links. Conversions are tracked actions such as inquiries. Pipeline represents sales opportunities. We use these signals to assess how AI visibility relates to qualified leads and revenue.

An improved AI answer does not automatically prove it created a sale. We state what we measured, what changed, and what the business data supports.

The product we are building

Our primary focus is midmarket and upper-middle-market manufacturers, agencies managing many client brands, and private equity firms looking across portfolio companies.

The same question runs through each engagement: Where is a company winning or losing a buyer’s consideration, and what should the team do next?

Signal Harbor tests the buyer questions, measures the answers, compares competitors, investigates the sources, gives marketing teams clear priorities, and retests after changes are made.

That is AI recommendation intelligence: knowing what buyers hear from AI, improving the evidence behind the answer, and measuring whether the work helped.

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