Every reseller hits this moment eventually: you spot something everyone's buying, and your gut says stock up. Sometimes that instinct pays off. Sometimes it just means you're competing against ten thousand other sellers for pennies of margin.
This issue is about telling the difference, using actual sold data instead of vibes, and what happened when I brought AI along to help spot that difference in real time, safeguards and all. We've also got a proper callout for last week's webinar below.
The Find
Squishmallows are everywhere right now: high demand, easy to spot at estate sales and thrift stores. I passed anyway.
Pulled eBay's sold listings for the last 90 days: average sold price is $14.89, average shipping runs $6.14, and sell-through is just over 10%. With over 10,000 sellers currently competing on that search, it's not exactly an open lane either.
eBay sold comps for Squishmallows, last 90 days
Squishmallows are also big, and that eats into margin fast. Factor in packing time and shipping cost, and a $15 item with $6+ in shipping and a box big enough to need real packing material doesn't leave much profit, even before eBay fees.
None of this means Squishmallows are a bad category, just that "everyone wants it" and "it's worth your shelf space" are two different questions. Sold data, not asking prices, not hype, is what actually answers the second one.
Lesson: before you buy into a trending category, pull the comps.
The Prompt
Picture this: you're at an estate sale or thrift store and you spot a whole bin of Squishmallows. Demand is real, but so is the saturation we just covered in The Find. Instead of walking past, ask AI what else might be worth stocking, something like: "I know Squishmallows are too saturated, what plush alternatives should I stock, based only on sold comps."
That "sold comps" instruction is doing a lot of work. Skip it, and you tend to get results like the screenshot below. When I ran this, Gemini recommended a Valentino the Bear Beanie Baby, pointing to a $149.99 listing as proof. But that listing was a single active price for a rare "1994 error" mint variant, not a sold comp, and nowhere close to what the bear typically sells for. The same page showed the real range right next to it: $8 to $14.
Gemini's answer, with the outlier eBay listing circled against the actual typical price range
The lesson from the webinar holds here too. Telling AI to use sold data helps, but it can still grab the flashiest number instead of the representative one. Check its work before you trust it.
One More Thing
If you caught the "How to Build an eBay Reselling Side Hustle with AI in One Weekend" webinar last week, you already know the drill: pick one item, run it through an AI identification and pricing pass, and get it listed by Sunday. If you missed it, the replay's up.
We also covered Gemini Notebook (formerly NotebookLM), the tool to reach for once you've got a few listings under your belt, not something you need on day one.