Odd Logic
AI Overviews and Perplexity now decide who gets seen before anyone clicks a link. Here's how both engines actually pick their sources, and the six-step playbook to become one of them.
How to Get Cited on Perplexity and Google AI Overviews
To get cited on Perplexity and Google AI Overviews, you need three things: a direct answer in the first 100 words of the page, evidence an AI can verify (named sources, real numbers, first-hand data), and enough third-party trust signals that the engine considers you a safe source to quote. Ranking number one is no longer the goal. Being the source inside the answer is.
This is the follow-up to our guide on How to Rank on ChatGPT: What Actually Works. ChatGPT is one engine. Google AI Overviews and Perplexity are the other two that matter for ecommerce, and they each pick sources differently. Here's how, and what to do about it.
Why this matters more than your rankings do
Google AI Overviews now show up on roughly 48% of tracked queries as of early 2026, per BrightEdge data. When one appears, organic clicks to the results below it drop hard. Ahrefs has measured click losses of over 50% to top-ranking pages on those queries.
But the traffic doesn't vanish. It concentrates on the handful of sources cited inside the answer. And that traffic is better: Semrush's 2026 data shows AI search visitors converting at more than 4x the rate of standard organic visitors. Fewer clicks, much warmer clicks.
For an ecommerce brand, the math is simple. Ranking third on a query with an AI Overview and no citation means you're invisible. Being cited inside the Overview, even from a lower ranking position, means you're one of two or three brands the buyer sees at the moment of decision.
How Google AI Overviews pick sources
Two findings should reframe how you think about this.
First, ranking and citation are now separate contests. Ahrefs analyzed millions of AI Overview citations and found only about 38% of cited pages ranked in the top 10 for the same query. Google uses a process called query fan-out: it breaks your search into multiple sub-queries, pulls results for each, and stitches an answer from across all of them. A page that ranks 40th for the main query can get cited because it perfectly answers one of the sub-queries.
Second, position on the page matters as much as position in the rankings. A CXL analysis of 100 AI Overview citations found 55% of cited snippets came from the top 30% of the source page. If your answer lives in paragraph twelve, it's probably not getting extracted.
What Google's AI is actually scanning for:
A self-contained answer block. A short passage that answers the question completely without needing surrounding context. Google confirms there's no special AI Overviews markup. Structure is the markup.
Evidence inside the passage. Claims paired with named sources and specific numbers get extracted more confidently than vague ones.
Question-formatted headings. H2s that match how people actually phrase queries make it trivial for the model to map your section to a sub-query.
Freshness. Recently updated content gets cited disproportionately, especially on pricing, product, and trend queries, which is most of ecommerce.
Definitive phrasing. "The best approach is X" gets cited. "There are many schools of thought" gets skipped.
How Perplexity picks sources
Perplexity searches the live web on every query, evaluates roughly ten pages, and cites three to five. It behaves differently from Google in ways worth knowing.
It's more literal. Where ChatGPT synthesizes across many pages, Perplexity is more likely to lift a discrete block of text from one page if that block directly mirrors the query. A study by Analyze of over 83,000 AI citations found Q&A and direct-answer formats hit top-3 citation positions at 55% versus a 31% average.
It has a stronger recency bias. In that same dataset, fresh pages routinely displaced older, higher-authority incumbents in fast-moving categories. For ecommerce, where products, prices, and comparisons change constantly, this is an advantage for brands that actually maintain their content.
It leans on third-party validation. Your own page saying you're the best carries little weight. Multiple 2026 analyses found that for commercial queries, the large majority of Perplexity citations come from third-party sources: industry publications, review platforms, comparison roundups, and Reddit. Reddit alone accounts for around a quarter of Perplexity citations in some studies, and more in product and software categories.
Schema helps here. Pages with FAQPage, Product, or ItemList markup outperformed near-identical pages without it in the Analyze study. Google says schema isn't required for AI Overviews, and that's true, but for Perplexity it acts as a confidence signal for extraction. Use both engines' preferences and you lose nothing.
The playbook, in order of impact
Rewrite your opening 100 words. Take your ten most important pages. Does each one answer its target question completely in the first paragraph, with a number or named source included? If not, that's today's work. This single change moves the needle on both engines.
2. Turn headings into questions and sections into standalone answers. Every H2 should read like a real query. Every section should make sense if it were the only thing someone read. This is what query fan-out rewards: your product care guide can get cited on a buying-intent query because one section answered one sub-question better than anyone.
3. Add evidence you own. Original data is the strongest differentiator in AI citations. For an ecommerce brand, that's return-rate data, fit data from thousands of orders, real durability testing, customer survey results. Nobody can replicate it, and AI engines treat verifiable first-hand numbers as citation-grade material. This is the same E-E-A-T principle we covered in What Is Answer Engine Optimization (AEO)?, now with harder data behind it.
4. Build your off-site citation surface. This is the part most brands skip and the part that decides Perplexity. Get into credible category roundups. Get reviewed on the platforms your buyers check. Show up genuinely in the subreddits where your category gets discussed. Not spam, actual participation. This is precisely why our service model pairs Search and AI Visibility with Review and Reputation and community presence: the engines are reading your reputation, not just your pages.
5. Add schema and a real last-updated date. FAQPage for Q&A sections, Product for product content, ItemList for comparisons. Include dateModified in your markup and a visible "last updated" line. Then earn it: update the substance on a real cadence. Cosmetic date bumps are a short-term trick that engines are already learning to discount.
6. Keep comparison and "best" pages fresh. These are the pages Perplexity cites most for buying queries and the pages that decay fastest. Refresh them monthly in fast-moving categories, quarterly at minimum.
What doesn't work
Keyword stuffing does nothing here. Neither does publishing more thin content, chasing exact-match domains, or hunting for a secret AI Overviews tag that doesn't exist. One well-designed study out of Canada that included a control group of ranked-but-never-cited pages found that a lot of popular AEO advice describes traits that cited and uncited pages share equally. Structure alone isn't enough. Structure plus evidence plus third-party trust is the combination.
Also worth knowing: AI citations are volatile. In that same study, nearly half of citations observed in a single check never appeared again. Don't judge your visibility, or an agency's report, off one snapshot. Track your key queries over time and optimize toward the citations that persist.
Measure it or you're guessing
Perplexity sends trackable referral traffic. Look for perplexity.ai as a referral source in your analytics. Google includes AI Overview activity inside Search Console's Web performance data, so watch for informational queries where impressions hold steady while clicks fall. That's the signature of an Overview answering the query without you.
Then test your money queries by hand, repeatedly, or use an AI visibility tracker. Which engines cite you, which cite competitors, and which sources beat you tells you exactly what to fix.
The honest summary
Getting cited in AI answers is not a separate discipline from good SEO. It's good SEO with a smaller unit of optimization: the passage instead of the page, the citation instead of the ranking. If you want the full picture of how these disciplines fit together, read AEO vs GEO vs SEO: What's the Difference? and What Is Generative Engine Optimization (GEO)?.
And if you'd rather have someone build this for you, that's what we do. Odd Logic handles the content structure, the review and reputation signals, and the AI visibility tracking as one system, because the engines evaluate them as one system. We're a newer agency and we won't pretend otherwise, but this is the exact work we run on our own site, including this post you're reading, which was built to be extracted.