Skip to main content
Not sure which heading, offer or layout brings more customers? An A/B test shows the page you have now to some visitors and a second version, Version B, to the rest. Hiveku counts the leads each version brings and tells you when one is clearly better. This guide tests a new home page heading on a test site: “Fresh coffee, roasted this week” against the original.
Tests run on your live site, so the site needs to be deployed to production before a test can start.

​
Set up the test

1

Pick the page

Open your site in the builder and click the flask icon, A/B tests, in the row of icons above the preview. You can also click the site’s name at the top, then A/B Tests. Under Choose a page to test, click Test this page next to the page, usually one your ads or links send people to.
The A/B tests icon, outlined; A/B tests: Show part of your visitors a new version of a page. Keep the one that gets more leads. Steps 1 Pick a page, 2 Make a new version, 3 Split your visitors; Choose a page to test, Home /, with Test this page, outlined

The site's pages, ready to test

2

Choose the second version

Page to test shows the page you picked. Under The new version, leave Copy this page, then change the copy: Hiveku saves a copy at a new address, such as /home-b, for you to change. If you’ve already built the second page, choose Use another page I already have instead.
New A/B test: 1 Page to test, Home /, outlined; 2 The new version: Copy this page, then change the copy, Recommended, Hiveku saves a copy on Main at a new address, like /home-b. Nothing goes live until you deploy, selected and outlined; and Use another page I already have

The home page, and a copy to change

3

Choose who's in the test and what counts as a win

Under Who is in the test, leave Everyone, or choose Only some visitors to test only people from one ad or on a phone. Under What counts as a win, Leads anywhere on the site is the usual choice: a form sent or a call made, on any page, after seeing the tested page. Other choices include a form on the tested page only, reaching a page such as a thank-you page, phone calls, time on the page and how far people scroll.
3 Who is in the test: Everyone, Every visitor to the page can be in the test, selected and outlined, and Only some visitors; 4 What counts as a win: Leads anywhere on the site, Most used, A form sent or a call made, on any page, after seeing the tested page, selected and outlined; then A form sent anywhere on the site and A form sent on the tested page

Everyone, and leads anywhere on the site

4

Split your visitors, then create the test

Under Who sees the new version, pick how to share visitors between the two pages, or drag the slider. 50/50 gets an answer soonest; 80/20 shows the new version to fewer people while you try it out. You can change the split while the test runs. Give the test a name if you like, then click Create test.
5 Who sees the new version: 80% of visitors see the original /, 20% the new version /home-b, with the slider and the choices 50/50, 60/40, 80/20 (outlined), 90/10 and 10/90; Name (optional) Home page heading, outlined; and Create test, outlined, with Nothing changes on your live site until you start the test

80% see the original, 20% the new version

Nothing changes on your live site yet. Hiveku copies the page and opens the new test as a Draft, with the original and Version B side by side.
Draft, Home page heading; Original: Home / with View, and Version B /home-b with View, outlined

The draft test and its two pages

​
Change Version B

The test page lists what’s left to do before the test can start.
The new version: /home-b, Change it to what you want to test. The original stays as it is, with Edit it in the preview and Ask the AI to change it, outlined; 2 Put both pages live, with Deploy and Check again, and Before the test can start: Version B does not answer on the live site yet, and Version B is still an exact copy of the original; 3 Start the test, with Start test greyed out

Three steps to start

1

Change the copy

Click Edit it in the preview to change Version B yourself, or Ask the AI to change it to describe the change. The preview opens on /home-b. To change text yourself, click Visual Edit, double-click the text, type, and press Enter to save.
The heading on /home-b being edited in place, now Fresh coffee, roasted this week, outlined, with Enter to save, Esc to cancel

Version B's new heading

Change only what you want to test, such as the heading or the offer. If you change several things at once, you won’t know which one made the difference.
2

Put both pages live

Click Deploy in the test’s list, then Deploy on the Production row. See Publish Your Changes.
The Deploy menu: Development, and Production with 1 file changed and its Deploy button, outlined; and Deploy All

Deploy to Production

When the deploy is live, go back to the test and click Check again. It reads Both pages are live. The test is ready to start.
The new version: /home-b and Put both pages live, both ticked; Both pages are live. The test is ready to start., outlined; 3 Start the test: Visitors are split within about a minute, with no deploy. Visitors who have not allowed analytics see the original and are not counted; and Start test, outlined

Ready to start

​
Start the test

Click Start test, and Start test again to confirm. Visitors are split within about a minute, with no new deploy.
Start this test? New visitors are split on your live site within about a minute. Visitors who have not allowed analytics see the original. Cancel, and Start test, outlined

Hiveku asks first

Visitors who haven’t accepted analytics cookies always see the original and aren’t counted. The test is now Running. What visitors see shows both live pages and how many visitors each has had. Pause stops the split until you resume the test; End test stops it for good.
Running, outlined, Home page heading, with Pause and End test, outlined; What visitors see: the original page with the heading Untitled site and Version B with Fresh coffee, roasted this week; 80% of visitors, 0 visitors, and 20% of visitors, 0 visitors

The running test, with both live pages

​
Read the results

Below the pages, Hiveku says whether there’s a winner yet, and if not, what it still needs: at least 7 days, 100 visitors on each version and 20 conversions in all. Each version’s card shows how many of its visitors converted. Recent history lists when the test was created, changed and started, and by whom.
Too early to tell: the test needs more data before it can call a winner. Still needed: 7 more days of data; 100 more visitors on Original; 100 more visitors on Version B; 20 more conversions, outlined; the cards Original / and Version B /home-b, each 0 of 0 visitors converted; Health: no visitor counted yet; and Recent history: Started, Edited: the second version, Created

Too early to tell, at the start

Let the test run until Hiveku calls a winner. Ending it early means guessing.

​
End the test

1

Choose what visitors see from now on

Click End test. Hiveku shows the result so far, then asks which page everyone should see: Keep the original, or Send everyone to Version B, which adds a redirect from the original address to Version B on your Redirects page. Click End test. The results stay.
End the test: Too early to tell; No winner yet: there is not enough data yet to tell the versions apart. You can still end the test; the results so far stay. What should visitors see from now on? Keep the original, Everyone sees / again, selected and outlined; Send everyone to Version B, Adds a redirect from / to /home-b on your Redirects page; and End test, outlined

Keeping the original

2

Remove the test copy

If you kept the original, the copy isn’t needed any more. Click Remove the test copy, then Remove the test copy again to confirm.
Ended. Everyone sees /.; The test copy /home-b is no longer needed. Remove it, as one version on Main, with Remove the test copy, outlined

The test has ended

The copy is removed in the builder at once. It stays on your live site until your next production deploy: click Deploy to do that now.
Waiting for your next production deploy: /home-b is removed on Main. It stays on your live site until that deploy goes live, with Deploy, outlined

Gone from the live site after the next deploy

​
Good to know

  • All your tests: All tests at the top of a test lists them, with tabs for Running, Drafts and Ended.
  • Extra goals: Add a goal tracks more than one thing, such as calls as well as forms. Only the main goal decides the winner.
  • Analytics: opening the pages from the test page counts as a visit in your site’s analytics and ad pixels, but not in the test.
  • Where visitors leave: to find the page worth testing, see See Where Visitors Drop Off.