Mar 15, 2022

7 Lessons From Running 22,000 AB Tests to Optimize your Conversion Rate

Bethany Cowan

The average success rate of most AB testing programs hovers around 12%.

Yep, for every 100 experiments you run, only 12 of them will generate an increase in conversions. That's worse than a coin toss. '

Too many companies start AB testing hoping that it will deliver a great increase in conversions. A few months later, they are bitterly disappointed.

At Affiliate Summit West 2021, Ayat Shukairy, Co-Founder of Invesp, shared the lessons she learned by analyzing over 22,000 AB tests on 700 CRO projects. These lessons paint the common patterns that helped her team deliver upwards of 50% success rates on AB testing programs.

The lessons provide a strategic roadmap for a CRO project as well as tactical details needed while running the day-to-day operations.


Lesson #1: Stop Focusing on the Small Stuff

Look at the big picture. CRO is not about simply changing the color of a CTA to make it "more attractive.' It's about qualitative and quantitative research. It's about knowing what to prioritize. It's about coming up with a hypothesis that you can test. Only then will you be on the path you need to be.


Lesson #2: Interview for Growth

Never forget that you are looking for long-term, sustainable growth as an organization. So, you need to SHIP - meaning Scrutinize, Hypothesize, Implement and Propagate. Focus on expert reviews. Focus on design priorities that you think will work. Double down on development and post test analysis.


Lesson #3: Take a Stroll Through Your Site

How something appears is always a matter of perspective. If your perspective is different from that of your users, you might have a problem on your hands.

So, pay attention to qualities like the visibility of system status and the match between system and the real world.


Lesson #4: Align Business Objectives With Experiments

Simply put, nothing you attempt should be done in a vacuum. That is to say, you shouldn't be A/B testing purely because you're hoping something might work. Again, you need to be testing a hypothesis and that hypothesis must be aligned with the very real goals you're trying to accomplish.

Doing so for any other reason is simply wasting time and resources that can be better spent elsewhere.


Lesson #5: Don't Take Shortcuts

This is an important one, because it's a trap that a lot of businesses fall into. Case in point: too many incentives.

You're trying to create a rock-solid foundation upon which you can build from moving forward. You're not simply trying to incentivize people into a relationship with your brand. That will come later, yes - but there needs to be more "meat on the bone," so to speak, so that you can work up to that point.


Lesson #6: Stop Copying Competitors

Make no mistake: your brand is your own. You can learn from the mistakes of your competitors, but you should never copy their victories. That is purely a way to get lost in a sea of similar businesses, failing to differentiate yourself from them.


Lesson #7: What Comes First: Bugs, UX or Conversion?

Finally, you need to consider the order you're addressing the characteristics of your program for the best results moving forward. Consider these three elements like a pyramid: conversion issues are at the top, usability issues are in the middle and bugs and other implementation issues are at the bottom.


Put them in practice

If you implement these learnings, you'll see'what it really takes to run a successful CRO program (time, resource, management, etc). There's a'balance between strategy and tactics you have to strike when running an experimentation program, so do so carefully.