In April 2026, Josh Blyskal of the digital marketing team Profound analyzed billions of AI query logs and tracked roughly 900 newly published pages, setting a clear timing benchmark for Answer Engine Optimization (AEO): a median of 6.81 days, with 90% of pages cited at least once by ChatGPT or Claude within 37 days.

Seeing that data, this team did one thing: pulled up its own website for inspection.

SEO was about where you rank on Google. AEO is about whether you're in the citation list when ChatGPT answers.

01SEO is about ranking; AEO is about citation

For the past decade the search battleground was Google ranking. SEO engineers fought for "which blue link in organic search," optimizing for click-through rate and bounce rate.

From 2025, user search behavior began to shift. A question no longer goes to Google for ten blue links — it goes to ChatGPT for one assembled answer, with three to five citation URLs at the end. Whether you appear in those three to five slots decides the next wave of traffic.

This isn't SEO 2.0. SEO optimizes for "can you be found"; AEO optimizes for "can you be the answer." Different metrics, different structural needs, different ways to measure.

  • SEO measures: organic rank, CTR, bounce rate, time on page
  • AEO measures: number of ChatGPT / Claude citations, speed to first citation, number of queries covered

A page can rank third on Google yet never get cited in ChatGPT. It can also rank twentieth on Google yet be the top citation when ChatGPT answers a given query. Ranking and citation are two different curves.

02The 6.81-day median, the 37-day deadline

That Profound study gave three key percentile figures:

  • 50th percentile (median): 6.81 days — half of new pages earn their first AI citation within about a week
  • 75th percentile: 18.68 days — three-quarters are cited within three weeks
  • 90th percentile: 37.10 days — 90% are picked up within just over a month

The number that matters most to a marketing team is 37 days. It's the "something's wrong" health line — if a new page has been live for more than 37 days and still isn't in any ChatGPT or Claude citation list, it's usually a technical setup problem, not a "just wait longer" problem.

This benchmark turns AEO from "a feeling" into "a manageable timing metric." Before, when a client asked "how long until AEO shows results," all we could say was "usually a few weeks." Now there's a clear answer: first citation around day 6.81; if nothing by day 37, investigate the technical setup.

03We pulled ourselves up for inspection

We audited oakorc.com against the same standard. Overall verdict: medium-to-high citation probability, but three things to fix right away.

The audit runs across four layers — technical, structural, content, and entity:

  • Technical (9 / 10): whether robots.txt blocks OAI-SearchBot or ClaudeBot, sitemap integrity, canonical tags, hreflang setup
  • Structural (9 / 10): HTML semantics (H1 / H2 hierarchy), Schema.org structured data (Service / Article / FAQPage / Organization). All five service pages carry FAQPage schema with 6 Q&As — more than most Taiwanese agency sites
  • Content (7 / 10): whether titles and H2s match real user questions, whether paragraphs are easy for an LLM to extract. Service-page H2s are all questions, but the homepage over-indexes on brand narrative
  • Entity (5 / 10): whether the author is verified in structured data, whether sameAs links exist, whether an LLM can establish a person entity

The lowest score — the entity layer — revealed the real weakness: our journal articles named the author as the organization "OAKOrc," not the person "Oakley Lin." On citation probability, "a named viewpoint" far outranks "an anonymous company post." That an a16z partner's article gets cited 10x more than the same content posted anonymously is no accident.

04Three changes worth doing immediately

After the audit, three changes were the most worth doing right away. Each takes under 30 minutes and pays off for a full year.

Change one: add a one-line "what OAKOrc is" definition to the homepage. The homepage H1 was the brand slogan "Rooted in craft. Built for the new wild." — beautiful, but an LLM can't extract "what kind of company OAKOrc is" from it. The fix: between the hero and the services block, add a plain declarative definition section:

OAKOrc is a Taiwan-based digital consultancy offering four services — brand design, web development, SEO, and AI workflow integration — from NT$15,000/month and NT$45,000/project.

We also dropped this line into the Organization schema's description field, so "what you see" and "the structured data" say the same thing. When ChatGPT is asked "is there a company in Taiwan that does both branding and AI consulting," this is exactly the kind of sentence it wants to lift straight into the answer.

Change two: remove the duplicate sitemap entry. The audit found that sitemap.xml listed en/services/seo-content twice. It looks trivial, but it lowers the credibility of the whole sitemap — one error can deprioritize every entry in the file for AI crawlers. The fix was deleting 8 lines, but it's a non-negotiable hygiene step.

Change three: switch the journal author from Organization to Person. The Article schema's author was an Organization — "OAKOrc" the company writing the article. Change it to a Person:

"author": {
  "@type": "Person",
  "name": "Oakley Lin",
  "jobTitle": "Principal",
  "worksFor": { "@type": "Organization", "name": "OAKOrc" }
}

This turns "OAKOrc wrote an article" into "Oakley Lin wrote an article at OAKOrc." LLMs start tracking Oakley as an independent person entity, and citation rates rise over time.

05How to verify over the next 37 days

Finishing the changes isn't the end — it's the start of the timer. We split the verification cycle into three checkpoints:

  • Day 7: use an AEO audit prompt to query ChatGPT and see whether the homepage and service pages get cited
  • Day 14: scan all 8 site URLs page by page
  • Day 37: any page still not cited = red flag; go back and investigate the technical setup

The Neuron newsletter demonstrated a handy set of AEO audit prompts — anyone can use them to scan their own site:

You are an AEO review expert. I'll paste a URL. Assess the likelihood (high / medium / low) of that page being cited by ChatGPT or Claude within 7 days, and explain why; list the 3 fixes most likely to speed up citation; and infer the 5 query scenarios where it has the strongest citation advantage.

Paste that prompt into ChatGPT, then add your URL — it'll tell you the page's AEO structural issues, their priority, and the queries you should compete for. 30 seconds to audit, half a year of payoff.

06The bigger lesson: entity thinking

Fixing the website is only the entry ticket. The real long-term investment is something else — turning a person into an entity the LLM recognizes.

LLMs don't just cite pages; behind the scenes they build an entity graph: "what company is OAKOrc," "who is Oakley Lin," "what services do they offer," "which topics are their views tied to." Every journal article, every media interview, every LinkedIn post draws one more line on that graph.

Before: SEO was "optimize the page."
Now: AEO is "optimize the page + grow the entity."

A page can be fixed in a week; an entity takes a year or two to grow. But time eventually pays out — once an LLM establishes "Oakley Lin is one of the principals talking about AI consulting in Taiwan," citations on all related queries start tilting toward our articles.

This isn't a new marketing channel. It's a new search entrance being built. The biggest lever for the next year isn't fixing another 100 pages — it's pushing this team's viewpoints, judgment, cases, and positions out onto the web in every form, so the LLM has something to assemble its answers from.

Rooted in craft. Built for the new wild. — In the AEO era, what we put down roots for has gained one more object: not just the brand, but the people.