ChatGPT Changed How It Picks Products

Shopping-style questions — “best X under ₹5,000”, “which of these is good for beginners” — are increasingly answered inside the chat itself, pulling from product data rather than sending the user out to a search results page to compare tabs themselves.

What the assistant actually needs to consider a product

• Structured product data — Product schema with price, availability, and reviews, not just the same information written in page copy.
• Accurate, current pricing — a listing with stale or missing price data is a weak candidate for a direct recommendation.
• Enough descriptive content to match intent-based queries — “best for beginners” or “under ₹5,000” needs language on the page that actually supports that framing, not just a spec sheet.

The data hygiene checklist

1. Product schema present and passing validation, not just present.
2. Price and availability kept current — stale listings get dropped from consideration over time.
3. Review counts and ratings visible in markup, not only as on-page text a crawler has to interpret.
4. Descriptive copy that actually matches how people phrase shopping questions.

What this means practically for a small catalog

A handful of well-structured, accurate product pages beats a large catalog with thin or stale data every time — depth of data hygiene matters more than breadth of inventory here.

Read this with healthy skepticism

No platform has published the exact selection logic behind which products get surfaced. This is inferred from what the schema specification and public merchant guidelines reward, not a confirmed algorithm.