Trending PR Campaigns
Visa Agentic Commerce Benchmark
Visa tested 22 AI models across 15 shopping scenarios, including choosing a TV and a laptop. The benchmark found that models failed 42.8% of runs—and in more than half of those failures, they still bought something—turning a timely concern about AI shopping into a clear, newsworthy test.
Why it worked: A big, trusted payments brand published a simple, counterintuitive stat-based leaderboard and failure rate on a hot, high-stakes behaviour (AI buying wrong).
The lesson: Consider running a public, repeatable stress test with a clear pass/fail metric and surprising failure stats, then rank the players.
Coverage over time
How this story spread across the media