product sense · standard
How would you improve Google Maps?
How would you improve Google Maps?
The navigation layer is largely solved. Ask Maps (launched March 2026) brought conversational Gemini-powered search across 300 million places. Immersive Navigation replaced the flat map view with 3D turn-by-turn. Gemini replaced Google Assistant as the voice layer in November 2025. A candidate who proposes any of those features in 2026 signals they have not used the product recently. The strong answer starts by acknowledging what is shipped, then pivots to the open wound: local discovery trust.
Structure a strong answer
strong
"Before I dive in, can I clarify scope? I want to focus on the area with the highest unmet user need, not where Google is already investing. Google has done strong work on the navigation and discovery search layer this year with Ask Maps and Immersive Navigation, so I would stay out of that space and focus on local discovery trust, which I think is actively degrading."
"My user segment: local intent users. People who open Maps to find somewhere nearby to spend money or time. This is also Google's core monetization segment, so improvements here are strategically aligned with the business."
"The pain: the review system is compromised. Businesses buy ratings. Sponsored pins sit inside organic results with no clear visual distinction. Incorrect hours, duplicate listings, and review spikes are documented and widespread. Users cross-check Maps with Yelp or Reddit before acting, which means Maps is failing the one job that keeps local search defensible."
"Feature one: Verified Recency Score. Surface a confidence signal based on verified physical check-ins (via Maps Timeline or credit-card-linked visits) in the last 30 days, time-decayed. Not aggregate star rating but a freshness-weighted signal from people who were actually present. This attacks fake review gaming without requiring Google to adjudicate individual reviews."
"Feature two: Listing Integrity Alert. When Maps detects a spike in reviews inconsistent with visit volume from location data, surface a subtle label: 'Review pattern unusual, fewer verified visits than expected.' Treat it like a health warning, not an accusation."
"Primary metric: task completion rate on local discovery searches, tracked via Maps Timeline opt-ins. Did the user go to the place they tapped? Secondary: cross-app leakage rate, users who open Maps then open Yelp within 60 seconds. That is a direct trust-failure signal. Success condition: reduce cross-app leakage 20% among local intent sessions."
"The tradeoff I want to name explicitly: some of the businesses with the most suspicious review patterns are also paying for Google ads. A PM here has to argue that long-term trust cost outweighs short-term ad revenue, and frame this to finance as a retention play, not a revenue sacrifice."
weak
"I'd add an AI-powered itinerary planner and better AR navigation." Proposing features Google shipped in 2025-2026 signals the candidate does not follow the product. Picking commuters as the user segment without acknowledging that is the most over-covered segment in this question shows no preparation. Metrics like "I'd track engagement and satisfaction" with no definition of what engagement means for a mapping product are useless. Listing five features without prioritizing or naming trade-offs signals breadth over judgment. Treating Google Maps as a pure consumer utility, ignoring that the ad model is the root cause of the trust problem, misses the business framing an interviewer wants to see.
The PM judgment
The interviewer is testing whether you can do two things most candidates skip: acknowledge what the team has already built, and identify where the business model creates the user problem you are proposing to fix.
Navigation accuracy is a solved problem. The viable and lovable gap in Maps is now local discovery integrity. A friend who gives honest recommendations is a different product than a directory that sells placement. The strong answer makes that distinction explicit, names the business tension, and proposes a mechanism that works within the constraints rather than wishing them away.
One additional signal worth knowing: Ask Maps and Immersive Navigation launched only in the US and India. The other five billion Maps users have none of the new AI layer. Geographic expansion of the AI surface is a legitimate improvement angle if you want a second axis, but local discovery trust is the more differentiated answer because it is the one most candidates do not see.