Skip to content
New: simulate any flow against real records before it goes live. See what's new →

Measure what a campaign or flow adds: holdouts and lift

Last updated 14 October 2026

Open rates say who looked. A holdout says what the campaign or flow actually caused.

1. Start a holdout

On a campaign (before it sends) or in a flow's Settings, find Measure what this adds. Choose how many to hold back (10% is a good start), what to measure (a meeting booked, an order paid, any event) and over how many days, then Start. The same people are held back every time, picked at random.

2. Read the result

The card shows both groups, how many did the thing, and the difference with an interval. Below 30 people in either group or 10 conversions in all it says not enough data yet instead of guessing. You can look as often as you like: the interval allows for it. Every number opens on the people behind it.

3. A control group for all marketing

Admins can hold back a small share of everyone from all marketing in Settings ▸ Email ▸ Marketing. Operational email (receipts, password resets, replies) still reaches them.

"Influenced" deals on a campaign report count deals linked to people who clicked, within 30 days. That is association, not proof; the holdout is the proof.