The star rating finding mirrors what we see on the Google Shopping side: products with 4+ stars and a high review count outperform on Smart Bidding even when the price isn't the lowest, because the algorithm has learned to factor in quality signals at auction time. The actionable implication for Merchant Center is making sure your seller ratings and product ratings are properly mapped in the feed — we've seen accounts where rich review data existed but wasn't surfacing because the GTIN or brand matching was off. The agents.md piece is also worth watching closely for retailers with large catalogs, because the quality of that structured file will likely become a filter before any persuasion tactic even gets a chance to work.
think it'll become smarter esp at identifying the overly aggressive tactics eg The study found that GPT-5 didn't like the scarcity tactics, e.g., "only 2 left"
The star rating finding mirrors what we see on the Google Shopping side: products with 4+ stars and a high review count outperform on Smart Bidding even when the price isn't the lowest, because the algorithm has learned to factor in quality signals at auction time. The actionable implication for Merchant Center is making sure your seller ratings and product ratings are properly mapped in the feed — we've seen accounts where rich review data existed but wasn't surfacing because the GTIN or brand matching was off. The agents.md piece is also worth watching closely for retailers with large catalogs, because the quality of that structured file will likely become a filter before any persuasion tactic even gets a chance to work.
The tactics that moved humans (countdown timers, bundling) exploited shortcuts that AI models just don’t have. How would GPT-5 read these?
think it'll become smarter esp at identifying the overly aggressive tactics eg The study found that GPT-5 didn't like the scarcity tactics, e.g., "only 2 left"