Running Honest A/B Tests With Modest Traffic
Sequential testing, MDE realism, and when to stop guessing and just ship.
Most sites lack the traffic to detect the small improvements they keep testing. Recognising that early is what separates useful experimentation from expensive theatre.
01Compute the detectable effect first
Before building a test, calculate the smallest effect your traffic can detect in a reasonable window. If that number is larger than the change plausibly delivers, do not run the test.
02Test bigger things
Low-traffic sites should test radically different pages rather than button colours. Large effects are detectable; small ones are not.
03When to just ship
For clear usability defects, ship the fix and monitor. Testing whether a broken form should be repaired wastes the traffic you need for real questions.
Want this applied to your account?
A senior strategist will review your funnel and share where the leverage sits, no junior hand-offs, no generic audit template.
Book a Discovery Call