Build a Pricing Research Routine With Product Research and a Chat AI
What it is
eBay's Product Research tool (formerly Terapeak) is free in Seller Hub and shows up to three years of eBay sold prices, sell-through and postage trends. A general chat AI is good at a different job: turning a messy set of sold results into a clear pricing decision. Put the two together and you have a pricing assistant that doesn't break eBay's rules.
Why it matters now
On 20 January 2026 eBay published an updated User Agreement, which applied to existing users from 20 February 2026. It explicitly bans robots, scrapers, data-mining tools, "LLM-driven bots" and buy-for-me agents unless eBay has given express permission. So "just have an AI scrape sold listings" isn't an option any more, if it ever was. Product Research is the approved route to sold data. eBay is also now putting pricing guidance into Magical Listing, and it helps to have your own independent check alongside it.
How to try it this week
- In Seller Hub, go to Research and then Product Research. Search a specific item, not a broad term: model number, edition, size.
- Set the filters that matter: condition, Buy it now or auction, date range (the last 90 days is a good default), and UK only if you post domestically.
- Write down or export (where the tool allows) a small summary by hand: median sold price, the lowest and highest real sales, how many sold, and postage charged. Leave out anything that clearly isn't the same item.
- Paste only those numbers and your notes (no buyer data, no account details) into your chat AI. Ask something like: "Given these sold prices and my item's condition (describe the faults), suggest a Buy it now price, a sensible Best Offer auto-decline level and whether an auction makes more sense. Explain your reasoning."
- Keep a simple spreadsheet: suggested price, what you listed at, what it sold for, days to sell. After 20 items you'll know whether the AI's reasoning helps you.
Human-in-the-loop checkpoint
The AI never sets a price by itself. Read its reasoning and check it against the evidence: did it treat a boxed example as if it were yours, which is loose? Did it average in a spare-parts sale? For anything worth more than, say, £50, look at the actual sold listings in Product Research yourself before you decide.
Cost and energy note
Product Research is free. A free or basic chat AI plan is enough for this, because you're sending a few lines of numbers, not photos. If you want to keep everything on your own computer, small open models run locally with free tools such as Ollama. They're slower and less capable, but nothing leaves your machine and the running cost is just your electricity.
Pitfalls
- Don't use browser extensions or scripts that pull sold data automatically unless you're sure eBay allows them. The User Agreement is explicit.
- Chat AIs will confidently make up "typical prices" if you ask without giving them data. Always give them your numbers.
- Sold prices include accepted offers. Check how your view shows them.
- UK and US prices are very different. Filter by marketplace.
Sources: ebay.co.uk/help/selling/selling-tools/terapeak-research?id=… | ebay.co.uk/help/policies/member-behaviour-policies/user-agr… | valueaddedresource.net/ebay-bans-ai-agents-updates-arbitrat… | ollama.com
#aiprojects #pricing
- Question
- Active discussion
- Answer emerging
- Community solution
- Knowledge article