SEO and Answer Engine Optimization
Ranking in Google and getting cited by an AI assistant stopped being two jobs. They run on one substrate, and most agencies still sell you half of it.
One substrate, two surfaces
A crawler and a retrieval model want the same three things from your site, and neither of them is a keyword density target.
Both surfaces start with an entity graph a machine can resolve without guessing. Who publishes this, what do they sell, which of the six people sharing that name is this one. Google settles that question before it decides what you rank for. A retrieval model settles it before it decides whether to name you at all. Get it wrong and you are not losing to a competitor, you are losing to ambiguity.
Both surfaces then want structure they can traverse. Clean canonicals, honest headings, internal links that state a hierarchy instead of scattering equity, and structured data that agrees with what the page actually says. The old argument for doing this was crawl budget. The new argument is that a model chunks your page before it reads it, and bad structure hands it garbage chunks.
The third thing is the writing itself, and it is where the two disciplines quietly diverge from what most SEO work still produces. A search snippet rewards a page that earns the click. An answer engine rewards a paragraph that stands alone, answers one question completely, and survives being lifted away from everything around it. Write only for the first and you rank without ever getting quoted.
That is the whole argument for treating this as one practice. The substrate is shared, the measurement is not, and nobody should be paying twice to fix the same foundation.
Three ways into the same problem
Measure first if you do not know where you stand. Build first if you already do. Read the reference either way.
Generative Engine Optimization
Getting retrieved and quoted is a different problem from getting ranked. This is the build work that makes a model reach for your page when someone asks a question you should own.
- 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 Audit
Measure where you actually stand before anyone touches a template. Crawl health, index coverage, structured data, and a live citation run against real answer engines.
- Technical crawl and index review
- Structured data and entity graph
- Citation rate on live AI engines
- A fix list ranked by effort and lift
SEO and AEO Best Practices
The working reference underneath both engagements. Metadata, schema, internal links, and the writing rules that decide whether a paragraph survives being quoted out of context.
- Metadata and canonical discipline
- JSON-LD patterns that validate
- Internal link architecture
- Content shaped for citation
The numbers come off this site
No client logos, no invented case studies. This site is the test bed, it runs the same tooling a client gets, and every figure below is something the instrumentation reported.
0.32% to 0.86%
Sitewide click-through rate across the August 2026 pass on this site, measured on flat impressions so the gain came from the work rather than from more traffic.
11.9 to 8.7
Average Google position over the same window and the same impression base. Two and a half positions of movement, from one pass over structure and snippets.
Two live engines
Citation rate on this site gets counted by querying real ChatGPT and Perplexity engines rather than a proxy score, which is the only way to know whether an answer names you.
One result worth stating plainly, because it runs against the usual assumption. On this site the AI engines cite the work more often than Google ranks it. The gap is not on-page quality. It is off-site entity density, which is slower and less glamorous to fix, and which almost nobody sells.
What people ask before they start
Straight answers on scope, measurement, price, and how long any of it takes to show up.
SEO earns a position in a list of ten blue links. Answer engine optimization earns a sentence inside a generated answer, where there is no list and usually one or two sources get named. 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.
You already have both, whether or not anyone is working on them. Every page you publish is being crawled by Google and by the retrieval systems behind ChatGPT, Perplexity, and Claude. The question is whether the substrate underneath serves both surfaces or only one. Most agencies still sell the first half and treat the second as a bonus, which is how sites end up ranking respectably and going uncited.
By asking the engines. I build a query set from the phrases your buyers actually type, run it against live ChatGPT and Perplexity engines through a paid search API, and count how often your domain gets named in the answer. That number is a rate, not a rank, and it swings between runs because the engines re-retrieve on every query. Watching the trend over repeated runs tells you something. Reading a single run tells you very little.
The audit starts at $6,000 and runs two to three weeks, ending in a ranked fix list you can hand to your own team. A generative engine optimization build starts at $14,000 and runs six to ten weeks, covering the entity work, the structural changes, and the measurement setup. Scope drives the number, so the exact figure comes after a call.
Technical fixes ship in days. Google takes weeks to recrawl a site and reprice it, and the August 2026 pass here moved sitewide click-through from 0.32% to 0.86% and average position from 11.9 to 8.7 inside roughly 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 and not on the first one.
Find out where you actually stand
Start with the audit if you want a number before a plan, or bring the build straight away if you already know what is broken.