Generative engine optimization is becoming an operating discipline, not a slide-deck buzzword. On August 22, digital agency Volume Nine drew fresh attention to its free GEO Grader, a website audit that scores whether a brand is discoverable, understandable, and trustworthy to AI answer engines. The tool evaluates six areas, including structured data, AI readiness, mobile performance, reputation, and content built for extraction.
That matters because most companies still measure AI search with an occasional prompt: “Does ChatGPT mention us?” A serious GEO program asks a harder question: what repeatable signals make the answer engine confident enough to mention, describe, and cite the brand?
GEO is shifting from prediction to diagnosis
Volume Nine’s announcement is useful less because one agency released a free tool and more because of the model behind it. Its audit checks crawl directives and XML sitemaps, organization and author schema, visible authorship, page speed, reviews, awards, press mentions, brand consistency, freshness, sourcing, and answer-first structure. In other words, it treats AI visibility as a systems problem spanning technical SEO, content, reputation, and brand operations.
The agency’s own documentation says the grader checks more than 60 signals across ChatGPT, Perplexity, Grok, Claude, and Gemini. That is a meaningful correction to “optimize for the algorithm” thinking: there is no single AI ranking factor, and a score is not a promise of placement. A diagnostic is valuable when it turns an abstract visibility gap into a prioritized backlog that a marketing, web, and communications team can actually execute.
The data says owned content is only part of the answer
A June preprint, “Generative Engine Optimization at Scale”, provides a useful reality check. The study analyzed 102 brands, 102,025 prompt responses, and 149,912 source citations across ChatGPT, Gemini, Perplexity, Claude, and Grok using production tracking data from March through May 2026.
Only 2.9% of citations pointed to the tracked brand’s own domain. About 75.2% went to corporate or third-party brand pages, while YouTube accounted for 4.2%, technology and business media 3.8%, and Reddit or community forums 3.3%. The paper also found a steep maturity gap: global household names appeared in 72.9% of relevant unbranded answers on their first run, established mid-market brands in 43.6%, and small or niche brands in 11.4%.
The implication for executives is straightforward: publishing more pages on your own site is not the same as becoming more quotable across the web. A GEO audit should lead to an authority plan: credible third-party coverage, useful video, review and comparison presence, consistent entity information, and pages that make specific claims easy to verify.
Do not confuse a score with a business outcome
The same research shows why measurement needs discipline. Across repeated brand-prompt-engine cells, 63.2% were never mentioned and 14.3% were always mentioned; only 6.8% landed in the volatile middle. Sentiment was much noisier than mention rate, flipping between positive and negative framing in 45.5% of observed cells. The authors explicitly report a baseline, not a randomized test proving that any particular optimization causes lift.
That caveat should shape your dashboard. Track visibility rate, share of voice, citation rate, citation accuracy, source mix, and qualified AI referral traffic separately. Run a fixed prompt set across multiple engines, repeat it often enough to see a distribution, and record the date, engine, location, query wording, and cited URLs. Then connect changes to pipeline and revenue rather than celebrating a higher audit score in isolation.
What marketing leaders should do this week
- Baseline the buyer questions. Build 25 to 50 unbranded prompts across discovery, problem-solution, use case, comparison, and expert advice. Record whether your brand appears, how it is described, and which sources the engine cites.
- Fix machine clarity first. Confirm crawl access, indexation, canonical URLs, Organization and Article schema, author bios, contact details, and consistent descriptions of what you sell and who you serve.
- Rewrite for citation. Put a direct answer near the top of priority pages. Use descriptive H2s, short self-contained sections, named sources, original data, and explicit definitions. Make each section useful when extracted on its own.
- Build the off-site evidence layer. Earn expert mentions, credible reviews, comparison coverage, and demonstrative video. AI systems cannot cite authority they cannot find.
- Re-test before reallocating budget. Treat any free grader as triage, not truth. Validate its recommendations against repeated AI-engine results and downstream analytics before funding a large content program.
GEO is becoming measurable, but measurement is not the same as certainty. The winners will be the teams that combine technical hygiene, useful content, earned authority, and a repeatable testing loop.
Need a practical AI search visibility plan? Real Internet Sales helps businesses turn GEO from a vague concern into an executable growth system. Call 803-708-5514 or visit realinternetsales.com.
Sources: USA Today coverage of Volume Nine’s August 22 announcement; Volume Nine GEO Grader documentation; arXiv study, Generative Engine Optimization at Scale.