If you haven't already, setup your first content A/B test.
Bayesian A/B tests
Last updated: 14.02.2025
Setting up your first AB test
Step 1: Install chrome extension
The extension is used to setup experiments which you can edit in the dashboard later.
After installation, do two quick things:
- Right-click anywhere on a new tab, select "Inspect" from the menu and access Wafrow.
- Generate an API token from dashboard settings and paste it into the extension.
Youtube video explaining this quickly:
Step 2: Retrieve copy variations and start your first experiment
Use the chrome extension to generate variants you like.
Once you're satisfied, start your first experiment. Don't worry, real customers will not see anything yet. Youtube video explaining this quickly:
Step 3: Install javascript snippet on your website
Place this javascript above the </head> tag in your HTML to expose real users to your experiments.
<script>
function wafrow_init(t,e){const n=document.createElement("script");n.type="text/javascript",n.async=!0,n.src=t,n.setAttribute("orgID",e),n.onload=function(){window.orgID=this.getAttribute("orgID")},document.head.appendChild(n)}
wafrow_init('https://wafrow.com/js/experiment.js', 'FILL_YOUR_ORGANIZATION_ID');
</script>
If you are logged in, visit a version of this in account settings which auto-fills your organization ID and lets you verify if all steps are working as expected.
Note: On adding this script to the page, any image with a class wafrow-personalized-images is dynamic even if no experiment is running on the page. The reason it's designed this way is that a lot of our customers want to run automated banners e.g., on an SEM campaign and are running the traffic split on google, not on the landing page.
Step 4: Monitor experiment results
From the dashboard, you can immediately start seeing the first results.
In the experiment list, you might see some experiment error states of the form:
Results Not Significant
It takes upto 5 mins for events to reflect in this dashboard
Please observe the exact error messages to troubleshoot the experiment setup. In the case above, there is no event yet in either the treatment or control group which could be because rollout % is set to zero.
We've invested a lot of effort in making these messages descriptive and cover all edge cases. In case you face any issues and need to talk to a human, [email protected] or slack for enterprise customers is always available to help.