I barely get 500 visits a week. What should I A/B test first, and can I even trust the results?
At low traffic, your first move isn't a split test - it's a structured audit that fixes the obvious, high-confidence leaks (broken mobile add-to-cart, missing trust signals, slow load, a confusing checkout) without testing them; you don't need statistics to know a 6-second load time is bad. Save real A/B tests for genuinely ambiguous, high-impact bets, and don't call a winner early - CXL's rough rule of thumb is 350+ conversions per variation before you trust a result, which at 500 visits/week can take months. Most testing tools will happily flag 'significance' at 40 conversions; that's noise, not truth.
Go deeper
4 resources, 2 link-checked.
🛠️ Tool
✓ Link checkedFreemiumIntermediate
Instead of guessing what to test, GoodUI hands you patterns backed by hundreds of real A/B tests across ecommerce and SaaS sites - the fastest way to build a low-traffic-friendly test backlog with evidence behind it.
A free, structured audit across homepage, PDP, cart, checkout, mobile and speed - the checklist to run yourself in an afternoon before spending a rupee on tools or tests.
The most rigorous free explainer on running valid tests without fooling yourself with noise - essential reading before you're tempted to call a test 'significant' after 40 conversions.
Calculate sample size ahead of time rather than eyeballing a dashboard - a rough rule of thumb is 350+ conversions per variation before trusting a result.
Conversion research, not guessing, is what generates test ideas actually worth running.
Test the biggest friction point you can find, not the ten-minute button-color change.
The closest thing to a structured, practitioner-built curriculum in CRO - 20 courses across research, testing and psychology for a founder ready to move past ad-hoc fixes into an actual testing program.
