Does borrowed registration page copy lift mobile homepage conversion?

A headline and button copy update had lifted conversion on our registration pages. We tested 2 versions of that copy on the mobile homepage. Over 14 days, the stronger version rose 15.8%, short of our confidence bar.

Not significant Mobile 14 days Not called

SURFACE

Homepage

DEVICE

Mobile

VERTICAL

Ecommerce (B2C)

PRIMARY METRIC

Conversion rate

VARIATIONS

2

The hypothesis

By bringing the headline and button copy from a winning registration page test to the mobile homepage, we will make the next step clearer, resulting in more visitors completing the form.

The registration page test rewrote the headline and button to match the wording visitors expect for a DMV task, and it lifted conversion on mobile and desktop. The mobile homepage still used its original header, which also ran long, so we tested whether the same idea would carry over.

What we changed

2 versions of new header copy, both with new button copy

Both versions used the updated button copy from the registration page test. Each took a different approach to the header: 1 removed the value proposition, and the other shortened a header that ran long. Everything else on the mobile homepage stayed the same.

VAR A: COPY UPDATE 1

New header copy, version 1
Button copy from the registration test

VAR B: COPY UPDATE 2

New header copy, version 2
Button copy from the registration test

The result

The stronger copy version rose 15.8%, short of our confidence bar

CONVERSION RATE

+15.8%

VAR A CONVERSION

-4.2%

Over 14 days, Var B converted 15.8% better than the original and lifted revenue per visitor 15.8%, but the gap did not reach our confidence bar. Var A converted 4.2% worse. Traffic to each version was modest, and the homepage did not show the clear lift the registration pages had.

What we would do next

Rerun the stronger header for longer before calling it

The 2 versions moved in opposite directions, so the header wording mattered more than the button copy they shared. We would rerun the stronger version against the original for longer, so a result this size has time to reach our bar or fade.