2
1 Comment

We built the A/B testing tool we always wished existed.

I've been working in CRO and UX for 10 years — mostly helping e-commerce businesses figure out why visitors weren't converting and what to do about it. That meant running a lot of A/B tests, and using a lot of tools to do it.

The frustration built up slowly. The serious tools — Optimizely, VWO, AB Tasty — are priced for enterprise teams with dedicated developers and five-figure budgets. The cheaper alternatives cut corners on statistical correctness, or require you to hand your data to third-party servers with no real transparency. And none of them were built with European privacy law in mind — they treat GDPR compliance as an afterthought bolted on after the fact.

Meanwhile, the e-commerce stores I was working with — mostly small and mid-size businesses in France running on PrestaShop or WooCommerce — had real conversion problems worth solving, but no realistic way to test solutions properly.

So I called a developer friend and asked if he'd be up for building the tool we'd always wanted to use. He said yes. Between the two of us, we had what the project needed: I brought the CRO expertise and frontend experience, he brought the backend engineering and network architecture. A good split for something this complex.

The build was harder than expected. A visual editor that injects into third-party pages, distributes visitors statistically without third-party cookies, and doesn't tank your page performance is genuinely complex to get right. We hit walls regularly — statistical implementation, flicker elimination, compatibility across PrestaShop versions. What kept the project alive through the slow parts wasn't speed or team size. It was staying unblocked long enough to keep going.

14 months later, Jarstak exists. First-party variant assignment, a visual editor that works on any e-commerce stack, and reporting that shows statistical significance without burying you in jargon. Built in France, designed for the European market, priced for teams that aren't Walmart.

posted toAvatar for product Jarstak
Jarstak
  1. 1

    What stood out to me is that this isn’t really about A/B testing tools—it’s about removing the tradeoff between statistical rigor, privacy compliance, and accessibility for non-enterprise teams. Most tools force you to choose at least one of those compromises. If Jarstak genuinely holds all three without degrading performance or accuracy, that’s not just a better tool, it’s a different set of assumptions about who experimentation is actually for.