Hi everyone, this is Neo.

Lately, a lot of e-commerce friends have been asking me the same question: “Will AI search (like ChatGPT Search or Perplexity) completely kill our organic traffic?”

It’s a genuinely anxiety-inducing topic. We all know traditional SEO logic by heart: grab the keywords, crack the top 10, sit back and watch traffic roll in. But AI search feeds the answer straight to users’ mouths — who’s left to click links?

Recently, Perplexity AI’s head of communications, Jesse Dwyer, gave an in-depth interview that pulled back the curtain on AI search’s technical black box. His answers not only explained how AI “thinks,” but revealed a whole new concept: AEO (Answer Engine Optimization).

Today, drawing on that interview and the latest industry developments, I want to talk about how we can hold — or even expand — our traffic territory in the AI era.

Goodbye “Zero-Sum Game”: Personalization Is the Future

In the traditional Google era, SEO was a brutal zero-sum game. There are only 10 spots on page one — if you move up, I move down.

But Jesse Dwyer points out that AI search is breaking that rule.

“The biggest difference between AEO and SEO is that it’s no longer a zero-sum game. Two people searching the same question can get completely different answers.”

Why? Because AI search engines (like Perplexity and ChatGPT) load in the user’s personal memory and context.

It’s like asking an old friend who knows your taste to recommend a restaurant — he’ll pick based on what you like. Ask a random stranger, and you get the #1 spot from a review app.

Neo’s take: What does this mean for e-commerce sellers? It means “long-tail” is no longer long-tail — it’s “precision.” You don’t need one big keyword to please everyone. If your content precisely matches a specific slice of users’ specific needs (your ICP), you have a shot at becoming the “one answer” AI recommends to that person.

“Bing in a Trenchcoat” vs. Real AI Search

Jesse dropped a really interesting technical concept in the interview: whole-document processing vs. sub-document processing.

This also happens to be the difference between the two schools of AI search on the market today:

1. GEO (Generative Engine Optimization): Whole-Document Processing

This is essentially what ChatGPT’s web search does right now. Under the hood, it still runs a traditional Google/Bing search first, grabs the top 10-50 pages, and lets the AI summarize them. There’s a running joke in the industry that this is “4 Bing searches in a trenchcoat.” In this mode, traditional SEO still matters a lot — you have to get found before you can get summarized.

2. AEO (Answer Engine Optimization): Sub-Document Processing

This is the future Perplexity is pushing toward. Instead of indexing entire web pages, it shreds pages into countless tiny snippets. A snippet is only about 2-4 words (5-7 tokens). When you ask a question, the system instantly retrieves roughly 26,000 of the most relevant snippets (about 130,000 tokens), fits them together like puzzle pieces, and fills the AI’s context window.

Why do it this way? Jesse explains that when the AI’s context window is packed with highly relevant, real information, it has no room left to “hallucinate” (make stuff up). It’s no longer “creating” — it’s “assembling” the truth.

Neo’s take: This is the real disruption. If your article is mostly filler with two nuggets of substance, in “whole-document” mode you might still scrape a ranking. But in “sub-document” mode, the system will use a scalpel to cut out exactly those two nuggets and discard the rest. Information density will become the only metric that measures content value.

How Should E-commerce Sellers Respond? Neo’s 3 Recommendations

Facing this shift from SEO to AEO, what do we do? Lie down and give up? Of course not.

1. Don’t Abandon Traditional SEO Yet

Jesse was explicit: “Most traditional SEO best practices still apply.” Perplexity still uses PageRank-like algorithms to judge site authority. If your site doesn’t even have basic indexing and authority, AI won’t even put your content in the candidate pool. So backlinks, site architecture, and technical SEO are still your entry ticket.

2. Write Content Like You’re Feeding an AI

In the AEO era, we need to change how we write.

• Structured data: Schema markup, for example, makes it easier for machines to read your product specs, prices, and reviews.
• High information density: Since AI extracts content by “snippets,” every paragraph should say something real. Cut the hollow adjectives; use data, facts, and unique perspectives.
• Answer questions directly: Explicitly ask questions in your article (FAQ format) and give clear, direct answers. That’s the easiest content to get extracted as the “best answer.”

3. Build Brand Authority

Since AI recommends based on context, “trust” is the biggest context of all. If Perplexity decides your brand is the absolute authority in a niche (say, “outdoor camping”), it will prioritize pulling your data snippets when answering related questions. From now on, stop staring at “traffic” — start staring at “influence.”

Summary

Traffic may shrink, but it’s getting more valuable. Search may get harder, but it’s getting more precise.

Perplexity’s reveal tells us the future of search isn’t “finding web pages” — it’s “finding answers.” As sellers, our job is to become the most precise, most trustworthy answer.

Finally, one quote from Jesse:

“It’s no longer a zero-sum game about rankings — it’s about who can fill that context window best.”

Let’s keep pushing, together.

---

References: Perplexity AI Interview Explains How AI Search Works