# Secondary Action as a Link or a Button: A/B Test Idea

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/secondary-action-as-a-link-or-a-button-ab-test-idea

## Hypothesis

As a button, the secondary action may be easier to spot for visitors who need it, while a text link stays lighter and competes less with the primary action. Which format raises total action completion rate may depend on whether gains on the secondary path outweigh any loss on the primary one.

## When to run this test

Session recordings of visitors hunting for the link, support tickets asking where the secondary option is, or heatmaps showing the link almost never tapped all point to a secondary path that gets missed. If the button format works, secondary completions rise without a primary loss large enough to offset them, and total progress goes up. If the secondary option's wording is unclear or the option is not useful to visitors, deal with that first, because a format change will not make vague copy clearer.

## Primary metric

**Total action completion rate**: Count assigned visitors who complete either of the two predefined actions; deduplicate visitors who complete both and report each path separately.

## Other metrics

- **Primary action completion rate**: Compare eligible assigned visitors who complete the primary action.
- **Secondary action completion rate**: Compare eligible assigned visitors who complete the secondary action.
- **Path switching**: Report movement between the primary and secondary paths separately; it does not replace the overall outcome.

## What to check before and during the test

- **Format**: Compare the same secondary action as a text link and as a button while using the correct semantic element.
- **Action mix**: Review whether the button presentation changes completion of either path.
- **Visibility**: Measure interaction with the link or button.
- **Position**: Keep wording and position the same.
- **Device**: Check whether the link target is easy to tap on mobile.

## What to watch when reading the result

- Total action completion rate settles the test; a drop in the primary action may be offset by a rise on the secondary path. Still report each path separately, and do not let path switching stand in for the overall outcome.
- If the button version gives the secondary action as much weight as the primary one, the hierarchy disappears and the result measures that confusion more than the format. In the link version, make sure the link can be told apart without relying on color, for example with an underline.
- The conclusion is tied to this secondary action in this position, so test it again on any other action or page before reusing it there. Read device results separately, since the link's tap target may be small on mobile.
- Some visitors will only switch paths, so the change in the total rate may be small. Before launch, enter the current total action completion rate in the [sample size calculator](https://alidemirbas.com.tr/en/calculators/sample-size-calculator); it returns the visitors each arm needs. If the difference stays unclear, record that the test could not tell the formats apart rather than calling them equal.

## Frequently Asked Questions

### Does this test change the secondary action's function?

Only its visual presentation changes. Navigation must retain appropriate link semantics even when styled as a button. An operation must retain button behavior even when styled as a link. Verify keyboard interaction.

### Why use total action completion as the primary outcome?

The secondary route may draw users away from the primary route. Count each visitor completing either route once. If route values differ substantially, predefine a value-weighted metric. Report individual routes diagnostically.

### Does lower primary completion mean the button version failed?

Not automatically. More secondary completions may offset it. Assess total progress while preserving a clear primary-secondary hierarchy.

### What accessibility checks apply to the link version?

Do not distinguish links by color alone. Verify a usable mobile touch target. Keep placement unchanged when testing visual format.