•
15 April 2019
Analytics & Optimisation
1. Know why people (don’t) buy your products!
Karl Gilis was announced onto the main stage as fireworks from Belgium, as medicine for the lunch dip. Karl didn’t disappoint. Claiming most businesses don’t know their customer, they don’t know why people actually buy their product, only socio-demographic B*llsh*t. BANG!
How can you persuade visitors to buy your product if you don’t know why people buy it?
Instead of summing up product specifications or bragging about your company, you should talk about actual customer benefits.
3 ways to find out:
Ask existing customers; What would you miss most when our product would be taken away?
Ask why customers bought your product from you, right after they did it.
Ask why people didn’t buy your product. (Exit Enquete / Unconverted leads)
Another great tip: don’t only optimize your website based on learnings from user research but also change your Meta descriptions and Ad copy! Think about the full user journey.
2. Automate analysis of your A/B tests.
Tom van den Berg talked about the importance of test quantity: doing many tests. Or as he called it: “Gas op die Lollie!”. An efficient CRO process will increase the # of tests you can run, which will increase the ROI.
Every time an A/B test has run, you have to check the conversion, significance & impact on important segments and other goals. That’s a lot of manual work and should be automated!
What to include in the automated report:
General Test Result: Users, Conversions, CR and significance.
Sample Rate Mismatch checks: is there a segment over- or underrepresented in the test, possibly skewing the result.
Test Results most important segments: Desktop, Mobile, New- & Returning Users.
Overall evaluation Criteria: formula with the weighted average of all goals.
3. Use your A/B tests as insight for personalization.
Instead of declaring just 1 general winner of an A/B test: A or B is the winner.
Look at whether there are segments for which the winner differentiates from the general winner.
So if in general, B is the winner, find out if there is a segment where A is actually the winner.
A and B might both be winners, but for different people.
Check if it makes sense (based on statistical best practices), to retest for that segment, or even immediately personalize.
