AI visibility · Answer

How should a CMO measure whether their brand shows up in AI search results?

Measure more than one platform, and pin the model, or the number will be stably wrong. Two major platforms named a different leading vendor on 72.7% of the same buyer questions and shared a median 3.8% of their sources. A measurement run on one platform describes that platform and nothing else.

The evidence

The platforms do not agree, and neither report says which one it ran on.

Across the same set of buyer questions, two platforms named a different leading vendor 72.7% of the time and shared a median 3.8% of cited sources. Both produce a clean, internally consistent number.

Citation sets are near-deterministic, which cuts both ways.

Asking the identical question three times returned identical citation sets — in identical order — on five of five questions tested on October 21, 2026. The prose varied by up to 1,400 characters; the sources did not move by a position. A single observation is therefore more reliable than this method usually permits, and a wrong number will be reproducibly wrong.

Often there is no citation to win at all.

Across 25 AI-generated answers to real enterprise-software buyer questions, 12 cited no source whatsoever. The four questions that cited nothing each consumed about 325 input tokens; the one that cited sources consumed 8,174. With web search enabled, the model retrieved for one question in five and answered the rest from its parameters.

Zero citations to the surfaces a vendor can buy.

In that same set: zero citations to G2, Capterra, SoftwareAdvice, Gartner or SelectHub. Two to Reddit, the one surface that cannot be bought, both on the same question.

Almost every published measurement runs on the cheap model.

Every vendor measurement we could find, and our own, ran at each platform's default budget tier. A buyer asking ChatGPT is not served that model. Pin the model explicitly or the measurement describes a tier nobody uses.

What would have to be true for this to be wrong

If your category's answers are dominated by sources you cannot place into — or carry no citations at all — then the measurement is sound and the budget behind it is not. Before funding a visibility programme, run ten of your own buyer questions and count how many answers cite anything. If most cite nothing, there is no surface to land on.

Sources: SEC EDGAR full-text search and company filings, with each figure's period stated above. AI-answer measurements are CULT+MATH primary research, criteria registered before retrieval. Published 2026-10-21 by Ipalibo Da-Wariboko, CULT+MATH LLC. Figures carry the date they were pulled; where a number could not be sourced, this page says so and gives none.

This is one of six decisions covered in the AI visibility decision brief — each with the evidence behind it, who owns it, who it is contested with, and what would have to be true for the answer to flip.

See the brief →