Tested whether session-only behavior (no purchase history) can drive useful product recommendations for small stores

Been working on a side project testing this: can you personalize product recommendations for a small fashion store using only what a visitor does in their current session — no accounts, no purchase history, no prior data — versus the usual approach that needs thousands of past sessions to train on.

Early results have been promising enough that I want to test it against real store catalogs instead of synthetic data. If anyone runs a small fashion store and is curious what this would actually surface for your products, happy to show you the output on your real catalog — no cost, mainly want real-world test cases.

Also genuinely curious from people here: if you’ve tried recommendation apps before, what made you keep or cancel them?