Most Shopify merchants know their conversion rate is weak, but very few can explain what is actually stopping them from fixing it. Beth Shero (CEO & Co-founder, Shero Commerce) sits down with Cande Rybecky (VP of Marketing and Growth, Shero Commerce) and Marc Adelman, VP of Growth at Visually, to unpack why testing two or three times a month will not move the needle, and what changes once agentic AI removes the dev bottleneck entirely.
In This Episode
- Why only a sliver of Shopify merchants run more than six tests a month, and what separates that group from everyone else
- How a six to twenty tests a month testing velocity becomes the difference between learnings and real revenue lift
- Why site segmentation gets skipped even by brands doing well past $100 million in revenue
- How agentic AI took a custom product finder quiz from first prompt to live data in a single day
Watch the conversaion to dive in deeper
Episode Highlights
Mark traced the testing gap back to a simple root cause: most brands assume CRO means pulling in dev resources they lack. Across roughly 4,000 brand conversations, Visually found that the overwhelming majority test two to three times a month at most, while an average successful program needs a 30% win rate and far more volume to move core business metrics in a measurable, repeatable way.
Cande shared the agency side of that resistance. Merchants often accept that a 1% conversion rate is bad, but push back hard on a phased roadmap, asking why they cannot just redesign the PDP and ship everything at once. Mark's comparison stuck: chasing one big swing instead of steady testing is a lot like going all in at a Vegas table, then wondering why the lift never showed up.
The clearest proof came from Laura Geller, which moved its product recommendations from a set it and forget it setup to an actively optimized one and saw a 103% lift, alongside roughly 40 tests running each month.
Mark also walked through a multi-step product finder quiz that would normally take months to scope, built with agentic AI and collecting live customer data within a single day, with no developers involved.