Likes dropped 10% overnight. What do you check first?

First check the drop is real, since a logging change or a broken dashboard can fake one. Then ask what shipped or changed outside, and split likes per active user by platform, app version, country and surface to find the cause before fixing anything.

How a 90-point answer starts

Before I fix anything, I need two answers. Is the drop real, and if so, where is it coming from?

See the full answer, the 50 and 70 versions, and what moves each score up.

Practice this question free

What the interviewer listens for

  • Do you check that the drop is real before explaining it?
  • Do you ask whether we shipped something or the world changed?
  • Do you split the metric until one group explains the drop?
  • Do you fix the cause, confirm the recovery and add a guardrail?

Common mistakes

  • Jumping to a fix, like a redesign or a campaign, before knowing whether the drop is real or where it comes from.
  • Trusting the dashboard.A changed logging event or a broken pipeline can produce a drop that no user ever felt.
  • Looking only at the total.A 10% drop overall can be a 40% drop in one platform or app version that the average hides.
  • Stopping once likes recover.Say what alert or release check would have caught it in an hour instead of overnight.

Likely follow-ups

  • What if the drop is spread evenly across every platform and country?Then a single release is less likely, so I'd look harder at logging changes and outside events like a holiday. I'd also check whether people are opening the app less or liking less once they're in.
  • How would you tell a logging bug from a real drop?Compare likes with numbers that come from a different source, like the likes stored on posts or comments and shares. If those hold steady while the logged event falls, it's the tracking.
  • Why watch likes per active user and not total likes?Total likes move with how many people open the app, so a slow day looks like a problem. Likes per active user shows whether the people who came are liking less.

More product management

    All interview questionsProduct manager interview practice

    Ready for an interviewer that talks back?

    Start the 7-day free trial and run your first PM mock with the AI interviewer today.

    Start your free week