Conversion Rate

In the early days of measuring ads, the only number within reach was clicks. More clicks read as things going well, and budgets went up on the strength of it.

There was a month at Pagedeck where that reading flipped. Ravi rewrote the ad copy to be more provocative and pushed click-through rate, the share of impressions that get clicked, from 1.2 percent to 3 percent. Two and a half times as many people arrived, and signups went down. The new copy had traded people looking for a meeting-notes tool for people who clicked on the promise of something free.

Conversion rate exists to measure that second stretch on its own. Up to the click, the ad is what's on trial; after it, the landing screen and the product are. Inside ad metrics, the set of numbers measuring ad performance, it pairs with click-through rate to form the response group.

Divide the number of people who took the target action by the number who arrived. In three steps:

1. Pick one target action. Write down whether it's a signup, a trial start or a payment. Mix two of them together and you can't explain what improved when the number rises.
2. Choose the denominator. Counting visits versus counting people splits the value widely. Somebody who stopped by three times and signed up on the fourth is 25 percent by visits and 100 percent by people. For products with long consideration periods, the per-person figure sits closer to reality.
3. Divide. A thousand people arriving and thirty signing up is 3 percent.

The thing to watch when reading it is the size of the denominator. Report 2 percent off fifty visits and one signup, and that same figure becomes 4 percent if a second person had signed up. It looks twice as good while one person is the entire difference. Wait until the denominator is in the hundreds before comparing.

The number doesn't get used alone. Multiply it by click-through rate for the odds that one impression leads to a signup, and attach the ad spend to get customer acquisition cost (CAC), what one customer cost to bring in. Whether to raise a budget gets decided on that final figure rather than on conversion rate by itself.

"Click-through went up and conversion halved. Let's roll the copy back."

All that happened was buying a lot of cheap clicks.

"Low conversion isn't an ad problem, it's a landing-screen problem."

Separating which stretch is responsible.

"Two percent off fifty visits isn't a number we can act on yet."

The denominator is too small for the ratio to hold still.

In Pagedeck's weekly ad review, Ravi always reads these two numbers side by side. In a week where click-through looks good and conversion looks bad, Mia brings the landing screen in so they can look at it together.