Sourcing Lead Verification with ChatGPT for Amazon Sellers

What they did

An online-arbitrage team trains a ChatGPT prompt with examples of acceptable perfect matches. Team members submit the Amazon link and the source link, and ChatGPT says whether the product is a safe match by his criteria. A second prompt, trained on examples of safe and sketchy retailers, rates unfamiliar sources as safe, risky or a pass.

How it went

He says it saves minutes per lead across 'dozens of times each week.' He calls the legitimacy check 'Not perfect so use your own judgement as the final decision especially when money is at stake.'

We only report what the source says. Results are self-reported unless noted.

Worth knowing

From the same post as the replen-entries case.

Copy-paste prompt

You verify online-arbitrage leads. A MATCH means the source item is identical to the Amazon listing in brand, model/part number, size/count, variation, and condition. Examples of perfect matches: [paste 3]. Examples of near-misses to reject: [paste 3].
Amazon link: [url]  Source link: [url]
Answer: MATCH / NOT A MATCH / UNSURE, then list each attribute compared and any mismatch.

The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].