The Follow-Up Questions After "What Do You Do?" (And Where Answers Break)

Why do interviewers always ask a follow-up here?

Because the first answer to "what do you do?" is almost always a claim without a receipt. People describe intentions. They say things like "I make sure we build the right thing" or "I solve customer pain points." An interviewer hears a claim like that the way an auditor hears "the books are fine." It is not that they think you are lying. It is that the claim, as stated, cannot be checked. So they check it.

In the session below, a product manager gave a fairly typical opening: they collect user interviews, align strategy, solve pain points, and cut through noise. Reasonable. Also completely unverifiable. Here is what got asked next, in order, with the exact moment each question found a crack.

"What specific metrics do you use to know a pain point is actually solved?"

What triggered it: the phrase "solving our customers' pain points." A pain point is invisible. Solved is a state you have to prove.

Where the answer cracked: the candidate reached for a grab bag of engagement metrics. Session length, drop-off, uninstall rates, onboarding completion. All real metrics, none attached to a specific pain point or a specific result. The debrief scored this turn as softened, and that is fair. Listing categories of measurement is not the same as measuring something.

How to answer it: pick one pain point, name the metric you watched, and give the before and after. One sentence. "Users were abandoning the booking page, so I tracked booking conversion, and we moved it from 2 percent to 4 percent." Nothing else required.

"What was the actual number you achieved, and what number were you targeting?"

What triggered it: the vague promise that users would stay "in the app for a longer period of time." Longer is not a number.

Where it held: this is the turn where the candidate found their footing. They told a real story. A flight comparison app where users browsed offers, then left to book on Skyscanner. Booking conversion sat around 2 percent. The team dug into why, fixed the underlying issue, and conversion doubled to 4 percent. The moment a real number entered the room, the temperature dropped. That is not a coincidence. Numbers end lines of questioning. Vagueness extends them.

"You said 2 percent. What was the rate after your team's work?"

This looks like a repeat, and in a sense it is. Interviewers do this deliberately. Asking for the same number twice tests whether your story is consistent or improvised. The candidate answered in seven words, "it went from 2 percent to 4 percent," and the debrief marked it held. Short, identical to the first telling, done. If your numbers shift between tellings, everything you have said becomes suspect. If they do not, you bank credibility.

"How did YOUR work drive that result? You keep saying we."

What triggered it: every result so far had been collective. "We managed to find the problem." "We managed to fix it."

Where it held: the candidate produced the single best moment of the whole session. They explained that by digging into the data, they discovered users were seeing cached flight prices. Stale prices meant that when a user cross-checked another platform, the other platform looked cheaper, and the booking happened there. Once the app served current prices, the price comparison stopped losing, and conversion doubled. That is a personal insight, mechanically connected to a business number. This is what the ownership question is fishing for. Not "I worked very hard," but "I found the specific thing."

How to answer it in your own interviews: name the insight only you had, then hand credit for the execution to the team. "I found that cached prices were killing us at the comparison moment. Engineering fixed the refresh, and conversion doubled." Ownership of the diagnosis, generosity with the delivery.

"In one sentence, convince me you're the right person."

Where it softened: asked for one sentence, the candidate delivered three, and all of them were trait claims. Good with data. Good with users. Able to dig through noise. The debrief flagged this as the weakest moment of the session, and the reason is worth sitting with. The candidate had just told a story that proved all three traits, then abandoned it for adjectives.

How to answer it: compress your best proof into the sentence. "I'm the person who found why a booking page was losing to Skyscanner and doubled its conversion, and I do that kind of digging everywhere I work." One sentence, one receipt, no adjectives doing unsupervised work.

How do you make your first answer follow-up proof?

• Include one number in your opening. It redirects every follow-up onto ground you have prepared.
• Claim one insight as personally yours before anyone asks whose it was.
• Never say a metric category when you can say a metric. Not "engagement," but "booking conversion."
• Keep your numbers identical every time you say them.
• Prepare one sentence that carries your best result, for the moment they demand brevity.

Editor's note: the session above is a real recorded practice session. The transcript has been lightly cleaned of speech-to-text artifacts. The rambling is original.