A/B testing is one of the most misused tools in e-commerce. Teams test button colours for weeks while ignoring the things that actually change revenue. Here is how we decide what is worth testing.
Test things that change the decision
The best tests target the moment a customer decides to buy: the offer, the value proposition, the hero section, pricing presentation and the checkout flow. These shape intent, so they move conversion rates meaningfully.
Worth testing
- Offer and positioning — how you frame the deal and the core promise above the fold.
- Product page structure — order of information, social proof placement, imagery.
- Checkout friction — number of steps, guest checkout, payment options.
- Pricing presentation — bundles, anchoring, shipping thresholds.
Usually a waste of time
- Button colours and micro-copy on low-traffic pages — the effect, if any, is tiny.
- Tests you can't reach significance on — without enough traffic, you're reading noise.
- Everything at once — change five things and you learn nothing about which one worked.
The traffic reality
Statistical significance needs volume. If a page gets a few hundred visits a week, most A/B tests will never conclude. In that case, make confident, evidence-based changes instead of waiting on a test that can't finish.
Test the decisions, not the decorations.
The takeaway
Prioritise tests by potential impact and your ability to actually measure them. Everything else is motion without progress.
