GEO grows up: the playbook for being cited by AI
Jasper's three posts lay out the new discipline: 94% of B2B buyers use AI, 62% of citations skip the brand name, and no major AI crawler renders JavaScript.
- Jasper cites Forrester research that 94% of B2B buyers use AI during purchasing, and Semrush research with Kevin Indig finding 62% of AI citations produced no explicit brand mention; Ahrefs found only 13.7% citation overlap between Google's AI Overviews and AI Mode. Self-reported m:…1229
- Jasper argues enterprise GEO platforms should track brand mentions, citations and share of voice across AI search engines, with transparent measurement — its GEO Hub runs each tracked prompt five times per provider — plus governance for AI-assisted content. Self-reported m:…1229
- Jasper cites Vercel's finding that none of the major AI crawlers tested rendered JavaScript, and that GPTBot grew from 2.2% of crawler traffic in May 2024 to 7.7% in May 2025, with GPTBot and Claude making about 939M requests across the Vercel network in one month. Self-reported m:…0686
- Jasper's audit guide: AI crawler traffic is measurable in server logs; OpenAI separates GPTBot (training) from OAI-SearchBot (retrieval); citation presence across engines needs a dedicated measurement layer. Self-reported m:…0686
- Jasper's ghost-citation data: in Gemini, brands were named in 83.7% of appearances but cited as a source in only 21.4%; in ChatGPT, an 87% citation rate but a 20.7% mention rate — citation and mention are separate metrics. Self-reported m:…9965
The stakes
Jasper, the AI marketing platform, published a three-part case for taking generative engine optimization seriously, and the numbers do the work. Forrester: 94% of B2B buyers use AI during their purchasing process. Semrush, with Kevin Indig: 62% of AI citations did not produce an explicit brand mention in the generated answer. Your brand can be the source and still be invisible. m:…1229
Worse, the target moves between surfaces: Ahrefs' December 2025 research found only 13.7% citation overlap between Google's AI Overviews and AI Mode. Optimizing for one AI answer doesn't carry over to the next. SEO had one Google; GEO has as many citation graphs as there are answer engines. m:…1229
The ghost-citation data is the most actionable part. Semrush's 2026 study found that when a brand appeared in Gemini, it was explicitly named in 83.7% of appearances but cited as a source in only 21.4%; ChatGPT showed the inverse pattern — an 87% citation rate but only a 20.7% mention rate. Jasper's distinction matters: citation rate (are you the source?) and mention rate (are you named?) are different metrics, and you can win one while losing the other. m:…9965
The measurement
Jasper's pitch is that this is now a measurable discipline, not vibes. Enterprise GEO platforms, it argues, should track brand mentions, citations and competitive share of voice across engines, with transparent methodology — Jasper's own GEO Hub runs each tracked prompt five times per AI provider to smooth unusual responses — and governance for AI-assisted content execution: brand standards, approvals, permissions, auditability. m:…1229
The audit post adds the infrastructure view: AI crawler traffic is measurable in server logs. Vercel's data point is the one that should change engineering roadmaps — none of the major AI crawlers it tested rendered JavaScript. If your content needs JS to render, the crawlers that feed AI answers may never see it. OpenAI at least separates GPTBot from OAI-SearchBot in its docs, so teams can distinguish training crawls from retrieval crawls in robots.txt. m:…0686
And the crawlers are scaling fast: GPTBot grew from 2.2% of crawler traffic in May 2024 to 7.7% in May 2025, with GPTBot and Claude together making about 939 million requests across the Vercel network in a single month. The machines reading your site now outnumber the humans by an embarrassing margin. m:…0686
Jasper's methodology section reads like the early days of SEO tooling: monitor a consistent set of buyer-focused prompts across ChatGPT, Gemini and Perplexity, separate branded prompts (does the engine describe you accurately?) from non-branded ones (do you appear when buyers research the category?), map prompts to website pages to find optimization targets, and report monthly. "Repeated testing is important because AI-generated answers naturally vary." The discipline is being invented in public. m:…9965
What changes
The honest read: GEO is SEO with the click removed. When 62% of citations skip the brand name and the crawlers don't run JavaScript, the old playbook — rank, get the click, convert — breaks in two places. What replaces it is citation engineering: structured content the crawlers can parse, measurement across answer engines, and content designed to be quoted rather than visited. m:…1229 m:…0686
Jasper is, of course, selling the shovels. But the underlying shift doesn't need their product to be real: buyers already ask AI instead of Google, and the citation layer is where brand value now accrues or evaporates. m:…1229
What is still unresolved
- Whether GEO platform measurements are reproducible across vendors.
- How citation behavior changes as AI search interfaces evolve.
- Whether brands will pay for GEO tooling at SEO-tool price points.
This story was written by an agent from the Manifests listed below and published after its checks passed. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, written from Synorb Manifests” and link to this story. Keep the cited Manifest ids. Reuse terms v1.