Diagnose a Drop in Zoom Meetings After the Launch of AI-Generated Meeting Notes
Company: Zoom
Role: Product Manager
Category: Analytics & Experimentation
Difficulty: medium
Interview Round: Other
Zoom has added AI features to its product, including notes that are generated automatically from each meeting. Since that launch, the number of meetings held on Zoom has gone down. As the product manager responsible, what would you do? Walk through how you would investigate the decline, what might be causing it, and how you would respond.
```hint Not every drop is a problem
Before hunting for something broken, ask whether the drop means users get less value from Zoom or the same value from fewer meetings, and decide which data would tell those two stories apart.
```
### Clarifying Questions
- How is "number of meetings" measured: meetings hosted, meetings per active user, or meeting minutes, and for which plans (free, paid, enterprise)?
- How large is the drop, did it start right at launch or build gradually, and is it the same across regions and platforms?
- Is the notes feature on by default, and who can switch it on or off: account admins, hosts or participants?
- Have active users and paid seats also fallen, or only the number of meetings per user?
### What a Strong Answer Covers
- Confirming the drop is real (instrumentation, seasonality, normal variance) and decomposing the metric before diagnosing
- A structured set of hypotheses: problems introduced with the AI launch (performance, interface, privacy and consent concerns), users needing fewer meetings because the notes replace attendance, and external causes such as seasonality or competitors
- The cuts and comparisons that separate those hypotheses, such as adopters versus similar non-adopters, meeting types, and support-ticket themes
- A judgment on whether fewer meetings hurts Zoom, tied to retention and paid seats, and on whether meeting count is still the right success metric
- A response for each likely cause, including an experiment design that does not damage customer trust
### Follow-up Questions
- Accounts that use AI notes hold fewer meetings but renew at higher rates. Would you still try to bring meeting counts back up?
- Many enterprise admins switch the notes feature off soon after launch. How does that change your plan?
- How would you measure the feature's causal effect on meetings without taking away a feature paying customers rely on?
Overview: A Zoom product manager interview question about a decline in the number of meetings after Zoom launched AI features such as automatically generated meeting notes. It tests metric validation and decomposition, structured root-cause hypotheses, separating product problems from healthy substitution, and experiment design.