Business Context

Asana launched a new onboarding experience in the workspace creation flow on 2024-04-01. In the same period, activation has historically shown weekly and quarterly seasonality, so the Growth team wants to know whether the observed lift is attributable to the release or to normal seasonal patterns.

Problem Statement

Use an interrupted time series approach with seasonal controls to test whether the product release caused a real change in activation rate.

Given Data

Daily activation rate is defined as the share of newly created workspaces that complete the activation event within 7 days. You are given 12 weeks of pre-period data and 4 weeks of post-period data.

Assume the fitted model is:

$Activation_t = \beta_0 + \beta_1 t + \beta_2 Post_t + \beta_3 (t \times Post_t) + \text{day-of-week fixed effects} + \varepsilon_t$

where $Post_t = 1$ after the release and 0 before.

Requirements

1. State the null and alternative hypotheses for the immediate release effect.
2. Compute the t-statistic and two-sided p-value for the immediate level change.
3. Construct a 95% confidence interval for the immediate release effect.
4. Test whether the post-release slope changed.
5. Compare the estimated release effect to the observed seasonal amplitude and explain whether seasonality alone is a plausible explanation.
6. Give a business recommendation for the Asana onboarding rollout.

Assumptions

• The release date is known and no other major onboarding changes launched during the window.
• Residuals are approximately mean-zero and the reported standard errors are valid.
• Day-of-week fixed effects adequately capture weekly seasonality.