Back to Hub
AI Visibility DiagnosticsCase Study 05

AEO: Eight High-Intent Opportunities

Executive Summary

AI answer engines now stand between buyers and brands: a prospect asks a question, a model names a few companies, and the rest are invisible. I ran two self-directed audits to find where the AI could be naming the company and is not yet. For an enterprise-technology company I mapped eight high-intent buyer prompts across four AI answer engines, each one an open opportunity, with one of the company's own subsidiary brands already surfacing in its place. A financial-services company was publishing strong material that machines could not yet read correctly: three of its homepage credibility figures rendered as the literal number 0 to any crawler that did not run JavaScript, and one basic fact about the company was stated four incompatible ways across its own pages. This is diagnostic work. Nothing was implemented and no result is claimed. What it shows is the method, and the discipline behind it.

An Enterprise-Technology Company

  • Eight open opportunities

    Across four AI answer engines, eight buyer-intent prompts came back without the company named, which makes each one a defined opening: a question its buyers are already asking that it could be the answer to. Competitors and third-party sources hold that space today.

  • Its older brand is already there

    One of the company's subsidiary brands surfaced in its place, so the engines already know the work. The brand consolidation just has not reached the AI layer yet.

  • Right answers, wrong shape

    Its best pages answered the buyer's questions, but at a length and position a machine does not extract from, with no crawler-facing index file or sitemap declaration to help an engine find them.

A Financial-Services Company

  • Headline numbers that read as zero

    Three homepage credibility statistics rendered correctly to a person and resolved to the literal number 0 for any reader that did not run JavaScript, the same figures already circulating from sources the company did not control.

  • One fact, four answers

    A single basic fact about the company was stated four incompatible ways across its structured data, homepage, about page, and pricing page, leaving an automated reader no consistent anchor.

The Method: Why the Findings Hold

Both audits ran read-only against public websites and public AI products, with no access to the companies' systems. I ran the citation baseline by hand, prompt by prompt, and read the raw HTML a crawler receives line by line against what the page shows a person. Every finding was labeled by whether it was directly observed, produced by a tool, inferred, or still needed checking by hand, and delivered as a list ranked by effort against impact. Hover any finding to see it.

Baseline
Ran eight buyer-intent prompts across four AI answer engines by hand, recording who the AI named and who it did not.
Read the machine's view
Read the raw HTML a crawler receives and compared it line by line to what a person sees in a browser.
Reconcile and label
Reconciled the company's own facts, ranked every fix by effort against impact, and labeled each claim by how it was known.
Diagnostic: delivered, not deployed

8 high-intent prompts, 8 opportunities

Across four AI answer engines, I identified eight buyer-intent prompts where an enterprise-technology company could be named and is not yet, each one a question its buyers are already asking. Competitors and third-party sources hold that space today, and one of the company's own subsidiary brands already surfaces in its place.

Headline numbers that read as zero

At a financial-services company, three homepage credibility figures rendered correctly to a person and resolved to the literal number 0 for any reader that did not run JavaScript, the same figures already circulating from sources the company did not control.

One fact, four answers

That same company stated one basic fact about itself four incompatible ways across its structured data, homepage, about page, and pricing page, leaving an automated reader no consistent anchor and pushing it toward third-party sources instead.

Buyer-intent prompts, 4 AI answer engines

open
open
open
open
open
open
open
open
8high-intent prompts identified as opportunities

Method

AI answer enginesCitation baselineCrawler vs browser renderingSchema and source of truthEvidence labeling