What AI Visibility Scores Leave Out

The Signal Harbor Journal
Research and perspectives from Signal Harbor

By Sebastian Miller
Co-Founder, Signal Harbor

What AI Visibility Scores Leave Out

When we started Signal Harbor, we wanted to answer a basic question: What happens when a buyer asks AI about your market?

Which companies get recommended? Which ones get left out? Is your company described accurately? What information does the answer draw on?

We still measure those things. But the work has made the bigger problem clearer. A marketing team needs to know which gaps matter, what to change, and whether the change helped.

An AI Visibility Score Shows Where to Look

We start with prompt research: choosing realistic questions to test based on the buyers a company wants to reach and the decisions they make. Those questions form a prompt library. We run them across AI platforms and examine the answers.

An AI visibility score summarizes how often a company appears in a defined set of those answers. Inclusion rate is a similar measure of appearances; share of voice compares a brand’s presence with competitors under a tool’s particular method. The labels and calculations vary, so the useful first question is always: Which questions produced this number?

A score describes the tests behind it. It does not measure a company’s share of every AI conversation.

That distinction becomes practical when we separate branded prompts, which name a company, from unbranded prompts, which do not. Imagine a Chicago-area manufacturer that appears whenever a buyer asks about it by name. If it disappears when a procurement team asks, “Which Midwest suppliers make custom precision components for medical devices?”, it has a gap earlier in the buying journey.

The score points us toward that gap. The answer tells us what to investigate.

Competitive Benchmarking Needs a Business Question

Competitive benchmarking means comparing where your company and its competitors appear in the same tested answers. It can show which company enters a buyer’s shortlist for a particular need.

The next step is deciding whether that difference matters. A missing mention on a broad question may deserve monitoring. Being left out when a qualified buyer asks about a service you want to grow calls for closer attention.

That is where the research becomes useful to marketing leaders, whether they are responsible for one manufacturer, an agency’s client brands, or a private equity firm’s portfolio companies. The priority is the gap tied to a buyer and a business decision, not simply the largest number on a dashboard.

Source Analysis Turns a Gap Into a Plan

When an AI answer cites a webpage, citation analysis tells us which source appeared in that response. Source mapping looks across answers to identify the pages, publishers, and third-party sources that repeatedly appear. A citation is evidence the answer displayed a source; it is not the same thing as recommending a brand.

We also review accuracy: whether the answer gets a company’s capabilities, customers, and strengths right. A favorable description can still be factually wrong.

Signal Harbor brings those findings together to investigate why a company is missing or misunderstood. Is the relevant information difficult to find? Is the proof for an important claim thin? Are competitors explained more clearly by their own pages or by outside sources? Does the company already have the right evidence, but only in sales materials or customer stories that buyers cannot find?

The resulting content gaps become a short list of work: a clearer service page, a better explanation of who the product serves, stronger customer proof, or a credible outside source. Your marketing team or agency can publish the changes; we help shape the priorities and examine the answers afterward.

The market calls parts of this SEO, AEO, and GEO. SEO concerns visibility in traditional search. AEO and GEO concern how information is used in AI answers. For a marketing team, the practical question is simpler: What information would help a buyer understand why this company belongs in the conversation?

Remeasurement Tells Us What Changed

Prompt tracking means testing relevant questions again over time. We establish a baseline, identify what to improve, and rerun the questions after changes go live.

Did the company enter an answer where it was previously absent? Did the description become more accurate? Is a competitor still being recommended for a decision the company should be part of?

Those are measurable changes in the tested answers. They give a marketing team a way to judge its work and choose the next priority.

We call the full process AI recommendation intelligence: understanding whom AI recommends to buyers, investigating why a company wins or loses consideration, deciding what to improve, and checking whether the answers change.

The Business Question Comes Next

Better answers matter most when they help the right buyers find and understand a company. When CRM, Google Analytics, Search Console, and sales data are available, we bring that context into the research. It helps us choose more relevant questions and examine how AI visibility relates to referral traffic, qualified leads, pipeline, and revenue.

That is attribution: connecting marketing activity with business outcomes as far as the evidence allows. A changed AI answer alone does not prove it caused a lead or a sale.

The problem is getting clearer because the questions are getting better. We began by asking, “How visible is this company in AI?”

Now we ask: Where is it losing consideration with the buyers it wants to reach, what should the marketing team change first, and did that work help?

That is what Signal Harbor does: we identify where companies lose consideration in AI answers, give marketing teams clear priorities, and measure whether their changes improve the answers.

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