A post-treatment window (PTW) is the period after an experiment ends where we continue measuring results to capture any delayed impact of ad exposure. In other words, it accounts for lagging customer behavior—cases where people take time to convert after seeing an ad.

​
When should you use a post-treatment window?

Using a post-treatment window makes sense if:
  • You’re selling a high AOV product with a long decision-making cycle.
  • You’re measuring upper-funnel or brand-focused campaigns, such as YouTube ads.
  • The KPI you’re tracking naturally has a delay (e.g., a free-trial sign-up that takes time to convert into a paid subscription).
  • You want to see if there are additional conversions occurring after the main experiment period ends.

​
How do we handle post-treatment windows?

When setting up an experiment, you can define a post-treatment window, typically up to the length of the original test. Extending it further risks introducing noise that isn’t directly caused by the experiment.
We provide:
  • A standard analysis at the end of the main test period.
  • An additional analysis covering the post-treatment window, allowing you to see any delayed impact compared to control regions.
  • Results presented both with and without the post-treatment window, ensuring transparency on how long-term effects contribute to overall lift.

​
Why does post-treatment window length matter?

If the post-treatment window is too long, it can dilute the cumulative lift measurement. This happens because:
  • External factors start influencing conversions, making it harder to isolate the experiment’s true impact.
  • Background noise and unrelated customer behaviors get included, reducing clarity on how much lift was truly driven by the marketing activity.
To ensure accuracy, we recommend choosing a post-treatment window that balances capturing real lagging effects without introducing excessive noise.