Landing page conversion rate: split it first

A landing page usually gets visitors from several places in one week: a search, a newsletter, a paid ad and a friend's link. The sign-up rate for the whole page averages them. The newsletter reader who already knows the product signs up at once, the paid visitor from a broad ad may leave at the headline, and the friend's link may land without any context. The blended rate looks like one problem, so owners rewrite the page for everyone. Split the rate by source first. One source may be fine and another the problem, and a change that helps one can hurt the other.

• Sign-up rate by source, for the same dates, with the source named in each row
• Sign-up rate by device: is the phone rate far below the desktop rate on the same page?
• The first line of the page against the ad or link text the visitor clicked
• The form: how many fields appear above the main button on a phone?
• The button label: does it name what the visitor gets, or only say Submit or Continue?

Read the rate one source at a time. Pull the sign-up rate for each source for the last several weeks, and mark the week of each change to the ad, the headline or the form. A source that held steady while another fell points to its traffic, not the page. A source that fell in the same week as a page change points to the page. New ad groups and placements start a new baseline, so do not compare them with the old rate.

The visitor decides in a few seconds whether the page matches the click. Put the ad headline or the link text beside the first line of the page, on a phone. If the words describe a different product, a different customer or a different offer, the visitor is in the wrong place, and a better form will not catch them. Then look at the form. A form that asks for a company size, a phone number and a role before the visitor has seen the product asks for more than a casual click agreed to give.

Landing page testing

A simulated visitor can be given the arrival of one source, such as a search for a specific phrase or the text of a paid ad, and sent through the page to the sign-up. The run shows whether the visitor saw the offer they were promised, which field or button they stopped at, and what they thought while they were there. Run each source separately, then compare the stop points. If search visitors reach the form and social visitors stop at the headline, the two sources need different first screens, not one better page.

A new ad group, a new audience or a different placement sets a new baseline, and an old rate can drop with no page change at all. Run each source through the page again after every change to the ad or the headline. Test a sign-up that sits behind a login with a test account, so the steps after the form are checked as well. Give each new audience its own run before you judge the page by its rate.