Gemini Skill-Tutoring Feature With a Love-or-Hate Reception: Diagnose and Fix It
Company: Google
Role: Product Manager
Category: Product Design & Strategy
Difficulty: medium
Interview Round: Other
Google has launched a Gemini feature that tutors people through complex skills, and reactions are sharply split: under 10% of users describe it as magical, while most call it useless. How would you fix it?
```hint Find what separates the two groups
Before proposing changes, look for what the delighted minority has in common: the skill they are learning, their starting level, and how they use the tutor. Whether you repair the experience for everyone or narrow it depends on that answer.
```
### Clarifying Questions
- Which complex skills does the tutor cover (for example coding, math, a new language, a musical instrument), and does the split look different by skill?
- Where does the feedback come from: in-product ratings, surveys, app reviews or social posts, and how much of the user base does it represent?
- What does the tutor do today: answer questions on request, follow a structured lesson plan, or review the learner's own work?
- What was the launch goal: engagement in the Gemini app, measurable learning outcomes, or a reason to upgrade to a paid plan?
### What a Strong Answer Covers
- Confirming the split is real, then segmenting delighted and dissatisfied users by skill, proficiency, intent and usage pattern
- Testable hypotheses for why most learners find the tutor useless, each paired with the evidence that would confirm it
- An explicit decision on whom to serve first, and why, without breaking the experience for the users who already love it
- Prioritized product changes tied to the confirmed causes
- Success metrics that track skill progress and return visits, with guardrails such as answer accuracy and cost
### Follow-up Questions
- Suppose the delighted users are mostly intermediate learners and the detractors are mostly beginners. Do you reposition the tutor for intermediates or rebuild it for beginners?
- How would you measure whether people actually got better at the skill, rather than whether they enjoyed the sessions?
- Leadership wants the sentiment split fixed within one quarter. What do you ship first, and what tells you the improvement is real and not cosmetic?
Overview: A product manager interview question about a Gemini feature that tutors people in complex skills and draws a split reaction, with a small minority calling it magical and most calling it useless. It tests how you check and segment polarized feedback, decide whom to serve, and measure whether changes improve learning.