A/B tests in adtribute
Connect ABlyft, Intelligems, Kameleoon or Varify, and compare orders, margin and new-customer LTV per variation.
Connect your testing tool
Section titled “Connect your testing tool”You need:
- Your domain connected.
- Your testing tool’s script on your site (Intelligems, Kameleoon).
- The Admin or Editor role.
- For ABlyft: JavaScript Events (ViewedExperiment, UserBucketed) turned on in Project › Integrations (ABlyft’s integrations).
- For Varify: Google Tag Manager set to Yes in Tracking Setup (Varify’s guide).
- In Sources › Connections, click Create Connection and select your tool.
- Click Create Connection (for ABlyft: I did that).
- Ask us at support@adtribute.io / Slack for the Experiments report in your Looker Studio report.
The connection shows Awaiting data until someone sees a test, then Connected.
How the Experiments report counts
Section titled “How the Experiments report counts”| How it counts | |
|---|---|
| Orders | An order counts for the variation the person saw before ordering. |
| Revenue and margin | From your shop’s orders, not from your testing tool. |
| LTV | Margin (CM2) of all orders, repeat orders included, from the new customers a variation won. Existing customers count in orders, revenue and margin, not in LTV. |
| Channel | Channel, campaign and ad of the person’s last touchpoint before they saw the test, such as an ad click. A visit with no ad, email link or referrer, such as a typed address, has the channel Direct. Direct / None holds people with no touchpoint before the test, and direct visits when your channel groups put them there. |
| People | One person counts once across visits, and across devices once adtribute recognizes them. |
| Several tests | An order counts in every test the person saw before ordering. |
| Several variations of one test | A person who saw more than one variation, for example on two devices, is left out of that test. |
| Tests | Only tests from the day you connect the tool. |
Why does my redirect test’s variation URL show few views or a low conversion rate? The browser loads the original page first, so landing page and funnel reports can credit the visit to the original URL. Judge the test in the Experiments report: it counts the people your tool assigned to each variation.
Why does my testing tool show different numbers? Your tool counts its own goals with its own tracking. See Data discrepancies. If the number of people per variation differs strongly, ask us at support@adtribute.io / Slack.