Your Next Customer May Never Click Your Website: The AEO Citation Playbook, Explained

Hi everyone, this is Neo.

Let me start with a counterintuitive claim: your next customer may never click through to your website.

This isn’t fear-mongering — it’s the core argument of the AEO Playbook that Conductor’s team shared in a recent Search Engine Journal webinar. They put it bluntly: buyers already get their answers from ChatGPT, Perplexity, Claude, and AI Overviews before they ever see a search results page. Your brand has to be visible where the answer is formed, or you’re invisible for the entire customer journey.

Today I’ve pulled the meat of that webinar together — plus my own take and some independent data — to give independent store owners a practical map for “getting cited by AI.”

1. The data first: why “citations” matter more than “traffic”

Conductor’s Pat Reinhart (VP of Services & Thought Leadership) and Lindsay Boyajian Hagan (VP of Marketing) opened with two data points:

1. The way AI crawlers visit your site has fundamentally changed

  • The old model: for every visitor Google sent you, it crawled your site roughly 2 times
  • The new model: AI crawlers hit your site thousands or even tens of thousands of times, yet bring back very little — or zero — traffic

Conductor saw the same trend in their own Search Console data: impressions keep climbing while clicks keep falling. More people are searching for help, but more of them simply read the answer inside the LLM interface.

Pat Reinhart is blunt about reporting: “The goal isn’t traffic, because traffic will naturally decline as people educate themselves in the LLM interface.” If your weekly report still leads with “organic sessions,” it’s time to change the metric.

2. AI prompts vs. traditional keywords: an information-density gap

  • Average AI prompt: 23 words
  • Average Google keyword: 3-4 words

Those 23 words carry context keywords could never hold: “wide-footed runners running on rugged urban pavement” is a completely different question from “best running shoes” — and the LLM recommends the brand that answered the need behind that specific scenario.

Neo’s take:

What do 23 words mean? It means the granularity of “search intent” in AI search is an order of magnitude finer than in traditional SEO. With traditional SEO we optimize for terms like “best running shoes” and maybe tack on a few long-tail variants; in AI search, the potential question space is nearly infinite — no brand can chase every prompt. That’s exactly why Conductor built its “Prompt Index” methodology, which I’ll get into below.

2. Core method #1: build a “Prompt Index” before you touch any content

Conductor’s approach: before touching any content, build a “Prompt Index” first.

In plain terms: systematically collect every question your target buyers would ask an AI, then categorize and score them by buying-journey stage. Lindsay calls it the turning point for her team’s strategy: “It was a huge unlock for my entire marketing strategy.”

Why do this first? Because:

  1. Prompts are nearly infinite — no brand can cover them all
  2. You need to know which prompts are worth winning (the ones with purchase intent and budget behind them)
  3. Everything downstream — content, PR — revolves around this index, instead of writing articles by feel

Neo’s take:

I’m fully on board with this, and it takes AEO from “mysticism” to “a checklist job.” I’ve said it repeatedly in my posts on tracking AI search: define the question set you want to win first, then talk optimization. Without an index, every “make AI recommend me” action is shooting buckshot — you might hit occasionally, but there’s no compounding.

A down-to-earth way to start: take the 50 long-tail keywords that brought you inquiries in Google Search Console over the past year, and expand each one into a natural-language question (“Is my X good for Y scenario?”, “Which is better, X or Z?”, “Is X worth buying?”) — that’s the seed of your first Prompt Index.

3. Core method #2: citations live on a spectrum — aim for “disproportionately positive recommendations”

Conductor places every brand mention on a spectrum:

Passive mention → Neutral description → Positive recommendation → Disproportionately positive recommendation

At the top of the spectrum: when someone asks “what’s the best brand in your category,” ChatGPT names you as the authority — and it does so noticeably more often than your actual market share would suggest.

Lindsay’s reminder is key: not every citation is equal. Citations that drive orders and citations that do nothing are two completely different things.

Neo’s take:

At its core, this spectrum turns “brand reputation” into a “search question.” Before, we built brands by waiting for customers to praise you on Zhihu, forums, or social feeds; now the model does the reputation aggregation for users, so what you need to do is actively manage the part of your reputation the model can read. Whose mentions are high quality, where they appear, and what keywords they’re tied to — all of that has become an optimizable search problem.

4. Three counterintuitive findings (the important part)

Finding #1: YouTube is the #1 citation source across all LLMs

“YouTube is the number one cited site across all major LLMs.” — Pat Reinhart

This isn’t one person’s opinion. Adweek’s independent reporting backs it up: analysis from data firm Bluefish shows that over the past six months, YouTube appeared in 16% of LLM answer citations, while Reddit got only 10% — the first time YouTube has overtaken Reddit as the most-cited social platform. The reason is pretty straightforward: video transcripts, descriptions, and narration make content easy for LLMs to read.

Neo’s take:

This is a strong signal for me: Chinese sellers running independent stores — if you haven’t started making YouTube videos yet (even just product walkthroughs plus subtitles), you’re basically invisible in AI search. English content + clean subtitles + brand messaging consistent with your site: that’s a route with excellent ROI. Stop treating YouTube as a “should we or shouldn’t we” option — it’s infrastructure in the AI citation system.

Finding #2: Seeing Reddit cited means a “content gap”

“LLMs only really cite Reddit when there’s no other good source out there. If you see Reddit, that’s a good indicator that the LLM is craving content.” — Lindsay Boyajian Hagan

Pat put it even more bluntly: “The model is never going to not give you an answer. If you see Reddit showing up there, it means no brand actually gave a good answer, and the model would rather cite discussions in a subreddit — which is crazy.”

Neo’s take:

For independent store owners, this is a golden signal. When you ask ChatGPT “best X for Y” and the answer cites a Reddit thread, congratulations: that spot is empty, and whoever fills it with high-quality content first wins. Conversely, if the model cites a competitor’s official page, that competitor has done their content homework, and you need to benchmark against them.

My suggestion: every independent seller should run a “Reddit check” once a month — run your 20 purchase-intent prompts for your core categories, and flag every answer that includes a Reddit link as a “content gap opportunity.” That’s your content calendar for next month.

Pat Reinhart’s exact words: “I don’t think backlinks matter much for AI visibility.” His personal track record: “I haven’t built a link in like 20 years, and I’ve never had a problem ranking a website.”

So what does he value more? The full webinar has the answer, but from the context the direction is obvious: entities, cross-platform consistency, and third-party mentions — these carry far more weight than traditional link assets.

Neo’s take:

This is the most controversial point, so let me say a fair word or two. Don’t read “backlinks don’t matter much” as “backlinks are useless” — in traditional Google rankings, links are still a core signal (I’ve written plenty about link building myself). Pat’s point is that in the new dimension of AI citations, the value links transmit is heavily diluted: the model cares more about “how others describe you” than “how many pages link to you.”

So the strategy should be: keep building links for traditional SEO, and focus on mentions for AI visibility — brand mentions in third-party media, review sites, and forum discussions. These are “brand citations” that don’t require links. Semrush’s research supports this too: brand citations are one of the core factors driving LLM visibility.

5. Practical additions: content freshness and “don’t mess with it”

The webinar also answered a few practical questions. I picked the three most useful:

  1. Should content be updated? Pat’s answer: look at performance. If a page keeps getting cited, “leave it be” — no matter how old it is. But don’t “just change the date to fake freshness” — the “freshness signals” LLMs recognize are updates that actually add something, not date refreshes.

  2. Content is missing vs. content exists but isn’t cited — what do you do? Lindsay’s answer: diagnose before you act. Missing content? Create it. Content exists but isn’t cited? The problem is usually technical — “the technical layer is really, really important now, or you will be invisible.”

  3. How many metrics should you actually track? Pat’s answer: don’t track everything. Teams that can “digest things into small chunks” tend to be more successful; monitoring too much leads to analysis paralysis.

Neo’s take:

These three points share one underlying logic: AEO isn’t about “doing more” — it’s about “doing the right things and staying consistent.” Of the failures I’ve seen, half were the wrong content direction, and half were people changing pages they shouldn’t have and losing their citations. Diagnose first, make small changes, watch performance — that’s far more reliable than sweeping grand gestures.

6. Action checklist: 5 things you can do this week

Here’s the webinar compressed into an executable checklist:

  1. Run the 20-batch test: write 5 purchase-intent prompts buyers would ask (“best X for Y”, “X vs Z”), then run them through ChatGPT, Perplexity, Gemini, and Google AI Mode; record each result as: named / cited a link / completely absent
  2. Do a Reddit content-gap check: every Reddit discussion that gets cited is a content opportunity for you
  3. Put YouTube on the calendar: even starting with an English walkthrough video of your product pages, with clean subtitles
  4. Build your first Prompt Index: expand your GSC inquiry keywords into natural-language questions and rank them by commercial value
  5. Swap your KPIs: move from “organic traffic” to “citation share” — of the purchase-intent prompts that matter to you, what percentage of answers mention you