A/B test link titles
Try a second title on a link, show each to half your visitors, and keep the one that gets more clicks, automatically or by hand.
Updated
Not sure whether "Shop the sale" or "50% off this week" gets more clicks? An A/B test shows one title to half your visitors and the other title to the rest, counts the clicks on each, and keeps the better one.
Title A is the link's current title. Title B is the one you want to try.
Start a test
- Open the editor (
/app) and click the link you want to test. - Find A/B test the title in the link's settings.
- Type your second title in the box. This is title B.
- Click Save. The test starts when you save.
From then on, each visitor to your live page is shown title A or title B at random. A visitor keeps seeing the same title when they come back on the same device and browser, so their clicks are counted fairly.
You can test one title per link, and run tests on several links at the same time.
See the results
Open the link in the editor to see how the test is going. Under A/B test the title you'll see the clicks, views and click rate for each title, for example:
"Title A: 10 clicks from 150 views (6.7%). Title B: 28 from 150 (18.7%). Title B is clearly ahead."
While it's too early, the result says how many more views each title needs. If both titles have enough views but are close, it says Too close to call so far.
Views here count each time your live page loaded with that title on it. The preview in the dashboard doesn't count.
How a winner is picked
Linkdeck checks every running test once an hour. A title wins when both of these are true:
- Each title has been seen at least 100 times.
- The difference in click rate is big enough that it's very unlikely to be chance. Linkdeck uses a standard statistical test at 95% confidence.
When a title wins, it becomes the link's title for everyone, and the test ends. If title B won, the link's title changes to title B. The editor then shows A/B test finished: title A won or title B won, with both titles and the final numbers.
If the titles perform about the same, a test can keep running for a long time without a winner. That's a sign the title doesn't make much difference, and you can end the test yourself.
End a test by hand
You can stop a test at any time and choose which title to keep:
- Open the link in the editor.
- Under A/B test the title, click Keep title A now or Keep title B now.
The title you picked becomes the link's title and the test ends. These buttons appear once the test has started, after you've saved.
To stop a test without choosing, delete the text in the title B box and click Save. The link keeps its current title.
Start a new test
After a test ends, click Clear and start a new test, type a new title B and click Save.
If you change title B while a test is running, the test starts over with fresh numbers, so results for the old title B don't mix with the new one.
Tips for useful tests
- Change one thing at a time. "Shop now" against "Shop the autumn collection" tells you something; changing the title, the thumbnail and the position all at once doesn't.
- Test your busiest links first. A link with few views takes a long time to reach 100 views per title.
- Keep the rest of the link the same while the test runs. If you change title A or move the link, the earlier numbers are less useful.
Related
Still stuck? Send us a message, or go back to all help topics.