Marketplace Listing Evaluator

A used-gear shop makes its margin when it buys. Much of its stock comes from private sellers on Facebook Marketplace and Craigslist, where listings are short, photos are poor, and the asking price says little about real value. Many listings are misidentified, overpriced, or quietly hiding a problem.

Checking each one meant researching the model, guessing the year from blurry photos, looking up recent sales, and remembering what tends to go wrong with that model, often for dozens of listings a week. Most weren't worth pursuing, but finding the few that were took hours.

We built a listing evaluator into the shop's admin system. Staff paste in one or more listing links. The tool collects each listing's title, description, price, location, and photos automatically, then AI evaluates it.

Each evaluation identifies the item: category, brand, model, finish, likely year, and condition, plus a serial decode if a serial is visible. Where something is a guess, it says so. Years come as a tight range when exact dating isn't possible, and inferred brands or models are marked "not definitive".

It then values the item as low, typical, and high private-party prices, with notes on what a pawn shop would pay and what selling online would cost in fees and risk. The asking price is shown for context but deliberately left out of the valuation, so a hopeful seller can't anchor the estimate.

Finally, it lists what matters before buying: known weak points and typical repairs for that model, what to check in person, what sellers commonly misrepresent, the original specs for pickups and tuners, and common modifications. Results land in a scored list the shop can sort, review, and archive.