How to set up A/B test goals
The goal defines what counts as a conversion in your test. This guide shows how to configure each goal correctly.
Step 1: Decide what "success" meansβ
Ask: what should a visitor do on this page? Click checkout, click any CTA, reach the thank-you page, or click out to a payment link. That action is your goal.
Step 2: Pick the goal in the test formβ
When creating or editing the test, open Primary goal and choose one:
| Goal | Extra field needed |
|---|---|
| Traffic split only | None β no Atomicat conversion tracking |
| Checkout click | None |
| Any button click | None |
| Click selector | CSS selector |
| URL visit | Atomicat page or external URL text |
Traffic split onlyβ
Choose this when you only need Atomicat to send visitors to different pages and you will judge winners in your payment platform or CRM. In Analytics you will see visitors and traffic health β not a conversion rate or βbest variantβ from Atomicat.
Step 3: Fill the extra field (if any)β
URL visit β Atomicat pageβ
Select the published thank-you or next-step page. A page view on that page counts when the visit is still tied to the test.
Atomicat keeps experiment context after the visitor lands on the test (cookie + same-site links), so a normal click through to thank-you usually attributes without adding query params by hand.
URL visit β External URLβ
Enter text that appears in the outbound link on the test page (for example part of your Hotmart URL). Atomicat counts the click on the test page, not a visit on the external host.
Click selectorβ
Enter a CSS selector for the element you want clicked:
- By ID:
#buy-button - By class:
.cta-primary
To find a selector: open the live page, right-click the button β Inspect, and read the element's id or class. Prefer IDs β they survive design changes better.
Step 4: Verifyβ
- Activate the test and open the entry URL in a private window.
- Perform the goal action (click the button, reach the thank-you page, or click the outbound link).
- Open Analytics β A/B Tests after a few minutes and confirm the conversion appears.
Changing the goal laterβ
You can change the goal while the test is still a draft. Once the test is running, the goal (and variants / traffic split) are locked so results stay comparable under the same experiment. To use a different goal, archive the test and create a new one.
FAQβ
What happens if the selector matches nothing?
Conversions stay at zero. Re-check the selector against the live variant pages, especially after page edits.
Can URL visit track an external thank-you page?
Use URL visit β External URL to count the outbound click from the test page. Atomicat does not see page views on hosts that do not run Atomic Pixel.
Checkout click vs Any button click?
Checkout click only counts Atomicat Checkout buttons. Any button click counts buttons, links, and other important clickables on the page.
Why is there no Page view conversion goal?
Use Traffic split only when you only need to divide traffic. That avoids confusing conversion rates from refreshes. Pick a real conversion goal when you want Atomicat to compare winners.