Learn 80% of AEO in 19 Minutes (Ahrefs)

TL;DR

Winning in AI search isn't luck — it's a 5-piece system on top of solid SEO. (1) Understand query fan-out (one prompt explodes into dozens of sub-searches), and that AI cites for consensus + freshness + authority (76% of AI-Overview citations come from Google's top 10 — so SEO is the foundation). (2) Run a brand gap analysis (Brand Radar) to find missed-citation, competitor, and topic gaps. (3) Do keyword/prompt research and apply the "AI filter" — if an AI Overview fully answers the query, aim to be mentioned in the answer, not to earn a click. (4) Get cited via four levers — content, mentions, YouTube, technical — where branded web mentions and especially YouTube have the strongest correlation with AI visibility. (5) Measure with AI referral traffic, AI-bot activity, and (most important) a "how did you hear about us?" survey.

Key takeaways

• Query fan-out: ranking for one keyword misses most of what AI searches. (ChatGPT deep research ran 420 searches for a single "red phone case" prompt.) Be relevant across a whole topic.
• 3 citation drivers: consensus (same story in many places), freshness (AI-cited content is ~25.7% fresher), authority (76% of AIO citations are already top-10 in Google).
• The AI filter: AI Overviews show on ~21% of keywords but ~58% of question queries, and 99.9% of AIO-triggering keywords are informational. For those, don't fight for the click — get mentioned inside the answer.
• Content that gets cited: length doesn't matter (>half of cited pages are <1,000 words); freshness does (76% of ChatGPT's top-cited pages were refreshed within 30 days); format does (43.8% of ChatGPT-cited pages are lists / best-X / comparisons / reviews).
• Mentions are the highest-leverage move. In a 75,000-brand study, branded web mentions had the strongest correlation with AI visibility — beating backlinks, DR, and referring domains.
• YouTube is special: the most-cited domain in Google's AI Overviews, and YouTube mentions have a 0.737 correlation with ChatGPT visibility (GPT-4 trained on 1M+ hours of YouTube transcripts). Build search hits, not viral hits.
• Don't accidentally block AI crawlers — ~5.9% of sites block GPTBot, often via inherited robots.txt or Cloudflare's default. Check /robots.txt for GPTBot, OAI-SearchBot, ClaudeBot, Google-Extended.
• Attribution is the real scoreboard: at Ahrefs, AI search was 0.5% of visits but 12.1% of sign-ups — a 23× higher conversion rate than organic — invisible without a "how did you hear about us?" question.

---

1 · How AI search works (query fan-out)

One prompt → dozens of specific sub-searches → synthesized answer:

flowchart TD
    Q["'Plan a 5-day Japan trip in November'"] --> F["query fan-out"]
    F --> S1["best Tokyo neighborhoods"]
    F --> S2["Kyoto weather in November"]
    F --> S3["is the JR Pass worth it?"]
    S1 & S2 & S3 --> A["synthesized answer<br/>(cited from many sources)"]

AI cites on consensus + freshness + authority — and since 76% of AIO citations are already Google top-10, traditional SEO is the foundation of AEO.

2 · Brand gap analysis

Measure where you should appear vs. where you do, across Google, AI, and the web (Ahrefs Brand Radar → add competitors → filter). Three gaps to note: missed citations (mentioned, site not cited), competitor gaps (they're in, you're not), topic gaps (AI links a topic to someone else).

3 · Keyword/prompt research + the AI filter

Seeds + modifiers (ask your AI to generate them), then run them in a keyword tool. The extra step is the AI filter:

flowchart LR
    K["A keyword"] --> Q{"Can an AI Overview<br/>fully satisfy the searcher?"}
    Q -->|"Yes (informational)"| M["Aim to be MENTIONED in the answer<br/>(Brand Radar → 'others only' → best/top/vs/review)"]
    Q -->|"No (action/transactional)"| C["Rank to earn the click (traditional SEO)"]

4 · Get cited — four levers

flowchart TD
    CITE["Getting cited"] --> C["📝 Content (your site)"]
    CITE --> M["🌐 Mentions (off-site) — highest leverage"]
    CITE --> Y["▶️ YouTube — strongest single factor"]
    CITE --> T["⚙️ Technical (don't block crawlers)"]
    M --> T1["Tier 1: editorial (Wirecutter, listicles)"]
    M --> T2["Tier 2: UGC (Reddit, Quora, forums)"]
    M --> T3["Tier 3: your own (YouTube, podcast, LinkedIn)"]

Content that AI can use — focused, fresh, and in list/comparison formats:

Micro-structure — 4 content principles:

1. BLUF (bottom line up front) — answer first; AI (and humans) weight the start/end over the middle.
2. Atomic content — every section stands alone, because AI chunks your page and you can't control where the cuts fall.
3. Entity-rich — name specific entities/relationships ("Ahrefs Keywords Explorer finds low-difficulty, high-traffic keywords," not "this tool helps with SEO").
4. Simple & declarative — one idea per sentence, subject-verb-object; if it takes two reads, simplify.

Mentions & YouTube — the highest-correlation factors:

YouTube tactic: build search hits (evergreen topics people search) not viral hits. Find them via Ahrefs Site Explorer on YouTube's watch subfolder (top-3 organic keywords + niche terms), then optimize: keyword in title + first lines of description, chapter timestamps, and say the keyword out loud (Google reads audio).

Technical: confirm you're not blocking GPTBot, OAI-SearchBot, ClaudeBot, or Google-Extended in /robots.txt (Cloudflare enables AI-bot blocking by default now).

5 · Measure what's working

Since GSC doesn't cover AI search, track three things — the third being the most important:

flowchart LR
    A["1 · AI referral traffic<br/>(clicks from ChatGPT/Perplexity/Claude —<br/>always an undercount)"] --> P["Picture of AI impact"]
    B["2 · AI bot activity<br/>(citation bots hitting a page =<br/>it's being used as a source)"] --> P
    C["3 · Self-reported attribution<br/>'How did you hear about us?'<br/>(catches what analytics miss)"] --> P

The strategy (do-this-next)

SEO foundation, then layer AEO — content, mentions, YouTube, technical. Order of operations: 30-second robots.txt check → set up measurement + the 'how did you hear about us?' question from day one → attack your biggest gap by priority (topic gap = refresh content; competitor gap = get into listicles/reviews/Reddit; format gap = get on YouTube). AI traffic grew 9.7× in the past year and most businesses haven't started — so the head start is real.