How Tapper measures Vantage results

Tapper measures Vantage on your money, against a control, with the uncertainty stated.

The rules

1. Two halves at the same time. Current setup and Vantage run side by side on the same budget share, so season, prices and promotions hit both.
2. Your revenue is the yardstick. Store orders without tax and after refunds, or CRM outcomes, each tied to a half through its ad click. The conversion value Google reports is shown as context, never as the result.
3. Counting is fixed first. Before values start, each account counts each purchase once, in both halves. The result shows what Vantage adds, not what a counting fix adds.
4. Brand is read apart from non-brand. People who search your brand mostly buy anyway, so a pooled number can hide where the change is.
5. Tapper Block is recorded per half. If Block runs differently in the two halves, the result says Block and Vantage together.
6. The uncertainty is stated. Every result comes with a range and the probability Vantage is ahead, at 80% confidence unless you set your own level. The smallest change the test can detect is worked out before launch.

The views

What the result counts follows the source you connect:

After the pilot

Tapper does not hold back part of your traffic all year. To check the result again later, Tapper runs a new test, agreed with you.

Relay and Truth

To measure what a whole channel adds, rather than a bidding signal, Truth, part of Tapper Relay, runs holdout tests on your own channels. See Truth, part of Tapper Relay.

Related

• Your Vantage pilot readout
• How a Vantage pilot is set up
• Google's count, predicted value and real revenue