ChatGPT Shopping Optimization

How ChatGPT builds its answers

ChatGPT shopping is built by OpenAI and answers questions by combining what it learned in training with live information it retrieves at query time. ChatGPT shopping blends merchant data, web retrieval through Bing and third-party review sites. The model then drafts a response, attributes facts to specific sources where possible, and ranks brands inside the answer based on a mix of source authority, recency and topical fit.

For AEO purposes the important thing to understand is that ChatGPT does not browse the way a person does. It runs targeted retrieval against a small set of trusted indexes, reads a handful of top results, and synthesises. That means the bar is not 'rank somewhere' — it is 'be in the small set of pages the model actually reads for your prompt'.

Where ChatGPT gets its sources from

• Bing's web and product index
• Product schema on retailer sites
• G2, Capterra, Trustpilot and category-specific review sites

What ChatGPT rewards

Product schema and review breadth matter more than backlink profile for shopping prompts. In practice that translates into a few repeatable patterns. Pages that answer the prompt in the first paragraph beat pages that warm up with context. Pages with consistent entity markup beat pages with strong prose but ambiguous identity. Pages backed by independent third-party coverage beat pages that only cite themselves.

Freshness matters more than most teams expect. ChatGPT weights recently updated content higher for any prompt with a time component, which is most commercial prompts. A page that was authoritative two years ago and has not been touched since will quietly fall out of the answer set.

A practical playbook for ranking in ChatGPT

• Map the prompts your buyers run in ChatGPT and measure where you currently appear.
• Audit your most important pages for entity clarity, structured data and direct-answer structure.
• Identify the third-party sources ChatGPT cites most often in your category and build a presence on them.
• Refresh your top pages on a fixed cadence so freshness signals stay green.
• Re-measure weekly and feed the gaps back into the next sprint.

Common mistakes when optimising for ChatGPT

• Treating ChatGPT like classic search and only optimising on-site content.
• Ignoring third-party sources that ChatGPT cites more often than your own domain.
• Measuring once a quarter when answers drift weekly.
• Optimising only for ChatGPT and assuming other engines will follow — they won't.

How to track progress and what to expect

A serious programme tracks ChatGPT alongside every other major engine, because what wins in one will not always win in another. The headline metrics are share of voice (the percentage of your tracked prompts in which ChatGPT cites you), citation rate (how often it links your domain specifically), sentiment, and prompt coverage across competitor sets.

Hertz runs prompt sets against ChatGPT and other engines on a continuous schedule and reports the deltas weekly, so teams can react in days rather than quarters.

Expect the first wins inside a month. Entity and schema fixes usually move citation rate first, then third-party source work compounds over the next quarter, then freshness routines hold the position. The brands that stall are almost always the ones who treat one good month as the finish line instead of the baseline.

It is also worth setting expectations internally. ChatGPT answers will keep drifting as the underlying model is retrained and the retrieval index refreshes. A weekly measurement cadence is not a luxury — it is the only way to tell signal from noise and to defend AEO budget against the inevitable quarter when a model update temporarily reshuffles the leaderboard.

Frequently asked questions