By Tom Whitfield

Forbes published a piece titled "AI Visibility Numbers Are Unreliable. Measure Them Anyway" that argues marketers should track how often their brands appear inside AI-generated answers, even when the third-party tools producing those numbers disagree sharply. An SEO director who runs the same brand through three vendors can see share figures that diverge by a factor of ten — and still cannot skip the exercise.

The thesis names a tension SEO (search engine optimization) teams have wrestled with since generative answers started displacing the traditional ten-blue-link layout on Google results pages.

What does "AI visibility" actually measure?

The term refers to how frequently a brand or domain shows up inside answers produced by large language model (LLM)-based search products — Google's AI Overviews, ChatGPT, Perplexity, and Microsoft's Copilot among them.

Marketers have built dashboards around three signals:

• Citation frequency counts how often the model names or links to the domain.
• Answer share tracks the percentage of relevant prompts that mention the brand.
• Sentiment scores how the model characterizes the brand when it does appear.

Why the numbers disagree so widely

Every vendor in this space runs a different combination of prompts, models, and scoring rubrics. One tool queries the public ChatGPT interface with a fixed prompt library. Another scrapes AI Overviews through a headless browser. A third calls a model API and parses the response string. A fourth uses a retrieval-augmented generation (RAG) pipeline that retrieves web pages and asks the model to summarize.

The same brand can return a single-digit answer share in one dashboard and a much larger share in another on the same day, with no published error bars to explain the gap. Add in model version drift — and the silent updates that ship behind each new GPT, Gemini, or Claude release — and the target moves under the measurement.

The Forbes argument, reduced to one line, treats the inconsistency as a reason to measure more carefully — not as a reason to stop measuring. Pick one tool. Track it over time. Treat the absolute number as secondary to the trend.

What this changes for SEO teams

For practitioners, the practical implication is a shift away from headline benchmarks — "we appear in X% of AI answers" — and toward longitudinal tracking on a fixed methodology. The number on January 15 and the number on February 15, drawn from the same tool, gives you a defensible signal. The number on January 15 and the number on February 15 drawn from different tools gives you nothing.

Agencies that report AI visibility to clients should disclose which tool and which prompt set produced the number, the way analytics teams disclose sample size and confidence interval. "Appears in 15% of AI answers" without a footnote should fail review.

The argument also reframes the conversation inside organizations where marketing leaders have asked whether LLM citations "count" the way Google rankings do. The Forbes position implies they do, for planning purposes, even when the underlying measurement stays rough.

Which verticals feel this most acutely

Among the verticals that have invested most heavily in AI-visibility dashboards, B2B SaaS (software-as-a-service), financial services, and healthcare publishers appear first, because their buyers research through long, comparative queries that AI systems handle well. E-commerce brands follow closely, since product recommendation queries in ChatGPT now convert to traffic for a handful of retailers.

Smaller publishers and affiliate sites feel the same pressure with less infrastructure. They cannot afford enterprise AI-visibility subscriptions and rely on free or low-cost tools that sample less polish.

What to monitor next

Three signals will tell you whether the field is maturing. First, whether major tool vendors converge on shared prompt libraries, the way rank-tracking tools eventually adopted overlapping keyword sets. Second, whether Google or Microsoft publish first-party data on citation frequency inside AI Overviews, which would settle the calibration question once and for all. Third, whether brands start reporting AI visibility in earnings calls or SEC (Securities and Exchange Commission) filings alongside traditional search metrics, moving the number from marketing dashboard to board-level KPI (key performance indicator).

For now, the Forbes position matches the working consensus among senior search practitioners: track the metric, distrust the absolute number, and treat the trend line as the only honest read on whether your content surfaces where customers ask questions.