Product Research · Podcast episode · source Jul 5, 2025
Product Idea Research with a Custom GPT in ChatGPT
Size up a product idea before sourcing it
Plain ChatGPT's answer was wrong: the links weren't even wooden egg trays, and one of Sutton's own top sellers in that niche wasn't on its list.
Per source From a vendor-affiliated source (see Worth knowing), not independently verified.
Copy-paste prompt
Research this product idea for Amazon US: [product]. I've attached [a Black Box / Product Opportunity Explorer export] - use it for sales and search figures and say which numbers come from it and which from the web. Cover: market size and trend, seasonality, the main competitors and their price points, the top 5 buyer complaints in reviews (with quotes), and what a better version would need. Don't name 'top sellers' unless they're in my file.
The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].
What they did
On Helium 10's Serious Sellers Podcast, Andrew Bell, a former Amazon listing manager who has built more than 350 custom GPTs, ran his product research GPT (on OpenAI's o3 reasoning model) live on 'wooden egg tray.' It works through 10 research dimensions, each with its own prompt and two or three web searches, and returns market size, search interest and seasonality, the competitor landscape, buyer complaints pulled from reviews and forums, and pricing. Before that, host Bradley Sutton asked plain ChatGPT for the top-selling wooden egg trays on Amazon.
How it went
Plain ChatGPT's answer was wrong: the links weren't even wooden egg trays, and one of Sutton's own top sellers in that niche wasn't on its list. The custom GPT's report surfaced complaints such as shallow holes that let eggs roll, splintering and finishes that aren't food safe. Its search volume came from the web, not Amazon, and Sutton's fix is to upload your own Helium 10 or Product Opportunity Explorer export so it works from Amazon data.
We only report what the source says. Results are self-reported unless noted.
Worth knowing
A live demo, not a launch result. The show is Helium 10's own podcast, so expect its tools to come up as the data source. The takeaway that holds without any tool: ChatGPT can't see Amazon sales data, so give it a file of real numbers before you ask what sells.