SEO and Answer Engine Optimization
Ranking in Google and getting cited by an AI assistant are now one job.
Two places to start
The explainer covers how answer engines pick a source. The reference covers the checks.
Generative Engine Optimization
Ranking and getting quoted are different problems. The explainer shows how AI models pick the page they quote.
- Entity clarity across the whole site
- Passages written to be lifted whole
- Agent-readable output and llms.txt
- Off-site density where engines look
SEO and AEO Best Practices
The reference behind both, with every claim linked to its source.
- Metadata and canonical discipline
- JSON-LD patterns that validate
- Internal link architecture
- Content shaped for citation
Want to test a site? Run the free AI visibility checker. It tests whether AI crawlers can read a site.
One foundation, two engines
Google's crawler and an AI answer engine want the same three things from your site. Neither wants a keyword density target.
Both engines start with an entity graph, a machine-readable record of who you are. Who publishes this, what do they sell, which of the six people sharing that name is this one. Google settles that before it decides what you rank for. An AI engine settles it before it decides whether to name you at all. Get it wrong and you lose to ambiguity, not to a competitor.
Both then want structure they can walk. That means clean canonical URLs, honest headings, internal links that show a hierarchy instead of scattering ranking value, and structured data that matches the page. The old reason was crawl budget. The new reason is that an AI model cuts your page into chunks before reading it, and bad structure hands it garbage chunks.
The third thing is the writing, and here the two part ways. A search snippet rewards a page that earns the click. An answer engine rewards a paragraph that stands alone and answers one question completely, even when lifted out of the page. Write only for the first and you rank without ever getting quoted.
That is why I treat this as one job. The foundation is shared, and the measurement is not.
The numbers come off this site
No client logos and no invented case studies. I test on this site and measure every figure below.
0.32% to 0.86%
Sitewide click-through rate after the August 2026 work. Impressions stayed flat, so the gain came from the work.
11.9 to 8.7
Average Google position over the same window, after one pass over structure and snippets.
Four live engines
I count citations in real ChatGPT, Gemini, Copilot and Google AI answers. No proxy score shows whether they name you.
One result runs against the usual assumption. On this site, AI engines cite the work more often than Google ranks it. The gap is not on-page quality. It is off-site entity density, meaning how often other sites name you. That is slower to fix.
SEO and AEO, answered
Measurement and how long results take.
Answer engine optimization (AEO) earns a named mention inside an AI-generated answer, while SEO earns a place among ten blue links. That answer has no list and names only a few sources. The mechanics overlap more than the labels suggest. Both reward a site whose entities are unambiguous, whose structure is crawlable, and whose paragraphs answer one question completely. AEO adds a retrieval layer on top, because a model has to find your passage, decide it is trustworthy, and then choose to quote it.
Yes, and every site already has both, whether or not anyone works on them. Google and the retrieval systems behind ChatGPT, Perplexity and Claude crawl every page you publish. The question is whether the foundation underneath serves both or only one. Many sites work on the first half and treat the second as a bonus. That is how sites end up ranking respectably and going uncited.
Pooya Golchian measures AI citations by asking the engines. He builds a query set from the phrases people type and runs it against live ChatGPT, Gemini and Copilot through a paid search API. He counts how often your domain gets named. That number is a rate, not a rank, and it swings between runs because the engines search again on every query. The trend over repeated runs tells you something. A single run tells you little.
Technical fixes ship in days, and Google takes weeks to recrawl and re-rank a site. On pooyagolchian.com, the August 2026 pass moved sitewide click-through from 0.32% to 0.86% and average position from 11.9 to 8.7 in about a month. AI citation rate can move faster, because a retrieval system re-reads your page on every query instead of waiting for a crawl cycle. It also moves less predictably, so judge it across several runs, not the first one.