The one-sentence mental model
Think of it as a funnel the machines run: Google and AI crawlers must be able to find and rank you (SEO), their extraction layers must be able to lift clean answers from your pages (AEO), and the models' broader world-knowledge must associate your brand with your category (GEO). Break any layer and the layers above it starve.
This is why 'SEO is dead' takes are wrong in a useful way: the foundation didn't die, it grew two new floors. Google's own 2026 guidance confirms AI features source from pages that rank in regular Search.
What changed in practice?
Three shifts matter. First, the March 2026 core update re-weighted information gain — original data, first-hand results and named expert authorship now beat synthesis of what already ranks. Second, measurement moved: zero-click answers mean mentions and citations matter as much as sessions. Third, the buyer's journey compressed — AI assistants now assemble shortlists, so being absent from AI answers means being absent from consideration.
- Information gain beats keyword coverage post-March-2026
- Named, credentialed authors beat anonymous content
- Mention/citation tracking joins rank tracking as a core KPI
- Freshness cycles tightened: money pages need 7–14 day to quarterly updates
What should a founder actually do?
Keep the classic hygiene — fast pages (LCP ≤2.5s, INP ≤200ms, CLS ≤0.1), clean information architecture, server-rendered content, real backlinks. Add the AI layer: answer blocks on every commercial page, FAQPage and Organization schema, original stats worth citing, founder bylines with real credentials, and open doors for AI crawlers in robots.txt.
Then measure like it's 2026: a monthly prompt-set run across ChatGPT, Perplexity, Gemini and AI Overviews, tracking whether you appear, get cited, and lead the recommendation.
Questions we hear about this
No — and splitting them usually creates contradictions, because they share one content and entity foundation. One team should own the stack end to end, with one scoreboard covering rankings, citations and AI mentions. That's how Chalk Labs structures its retainers.
Not for being AI-generated — for being unoriginal. AI-assisted content edited and bylined by a named expert with verifiable credentials performs fine. Pure rehash of what already ranks is what the March 2026 update demoted, however it was produced.
Publish an honest pricing page for your category ('how much does X cost') with real ranges, an answer block and FAQ schema. Pricing queries are heavily asked to AI assistants, rarely answered by vendors, and disproportionately cited.
Classic metrics (rankings, organic sessions, conversions) plus AI-era metrics: mention rate, citation rate and recommendation position across a fixed monthly prompt set on ChatGPT, Perplexity, Gemini and AI Overviews.