· PM Editorial · Product Sense · 5 min read
Design a Carpooling App: Metrics and North Star
How to choose a north star metric and supporting KPIs for a carpooling app, covering rides matched, rider retention, driver utilization, and carbon offset.
Once you’ve designed a carpooling app in an interview, the near-universal follow-up is “how would you measure success?” This is where many candidates who nailed the design half lose points, because a two-sided marketplace needs a metrics tree that captures both sides of the market plus the operational reality of matching — not a single feel-good number.
Why a Single Metric Fails Here
Carpooling is a two-sided marketplace with an added logistics layer (routing and scheduling) and an added mission layer (many carpooling products are pitched partly on sustainability). A candidate who proposes “total rides per month” as the north star is only measuring volume, not health — a marketplace can have rising ride volume while driver supply quietly collapses in half its corridors, which won’t show up until it’s a crisis. The right approach names the tension between growth, marketplace balance, and match quality, then picks a north star that’s resistant to hiding an imbalance.
Step 1: Separate the Business Goal From the Product Goal
The business goal is sustainable transaction volume that scales without ballooning subsidy costs (many carpool products subsidize early rides to bootstrap supply). The product goal is successfully matching a rider with a compatible driver at an acceptable detour cost for both sides. A good north star should reflect successful matching, not just app opens or ride requests, because a ride request that never finds a match is a failure, not a data point toward growth.
Step 2: Candidate North Star Metrics
| Candidate metric | Definition | Pros | Cons |
|---|---|---|---|
| Rides matched | Total completed carpool trips per period | Simple, directly tied to revenue | Can rise while match rate (successful vs. requested) falls |
| Match rate | % of ride requests that find a compatible match within an acceptable time | Captures marketplace liquidity directly | Needs a clear time-window definition to be meaningful |
| Driver utilization | % of a driver’s available seats filled per trip | Reflects efficiency and cost-per-trip health | Can be gamed by restricting driver supply artificially |
| Rider retention (30/90-day) | % of riders still active after N days | Reflects real habit formation, the ultimate business win | Lagging indicator, slow to react to product changes |
Step 3: Choose the North Star
The strongest answer proposes match rate — the percentage of ride requests successfully paired with a compatible driver within an acceptable wait window (e.g., matched within 10 minutes for on-demand, or matched at all for recurring commuter requests) — as the north star, with this reasoning: match rate is the earliest, most direct signal of marketplace health, since a low match rate predicts churn on both sides before retention numbers even have time to move, and unlike raw ride volume, it can’t be inflated by throwing more marketing spend at demand while supply quietly lags behind.
State the caveat clearly: match rate says nothing about the match’s quality (was the detour reasonable, did both parties actually complete the trip happily), so it needs quality-focused metrics alongside it.
Step 4: Build the Supporting Metrics Tree
Tier 1 — North star Match rate (% of requests matched within the acceptable window, segmented by corridor)
Tier 2 — Diagnostic drivers
- Driver utilization (seat-fill rate per completed trip)
- Average detour time added for drivers (should stay under the tolerance threshold that keeps drivers participating)
- Rider retention at 30 and 90 days (does a successful match translate into a repeat habit)
- Corridor liquidity (matches per corridor per day — aggregate match rate can hide dead corridors)
Tier 3 — Guardrails
- Cancellation/no-show rate post-match (a “match” that falls through isn’t a real success)
- Safety incident rate per 10,000 completed rides
- Subsidy cost per matched ride (protect against growth that only exists because of unsustainable incentives)
- Carbon offset per matched ride (mission-alignment metric, useful for B2B/employer partnership pitches even if not the primary business lever)
Step 5: Map Metrics to Marketplace Sides
| Side | Best-fit metric | Why |
|---|---|---|
| Riders | Match rate + rider retention | Riders churn fast if they can’t get matched reliably |
| Drivers | Driver utilization + detour tolerance adherence | Drivers churn if the economics of a nearly-empty seat don’t work out |
| Marketplace overall | Corridor liquidity | Aggregate health metrics can mask corridor-level supply collapse |
| Business/mission | Carbon offset per ride, subsidy cost per ride | Needed for employer partnerships and long-term unit economics |
Step 6: Connect Metrics to the Chicken-and-Egg Problem
If your design answer proposed launching with employer or campus-verified pools to bootstrap trust (see the companion design post), the metric to watch is corridor-level match rate within those bootstrapped pools specifically — a healthy aggregate match rate that’s entirely propped up by two large corporate campuses, while every other corridor sits near zero, is a sign the launch strategy hasn’t actually solved the marketplace cold-start problem, just delayed noticing it.
A Simple Test for Any North Star You Propose
Ask three questions before committing to a metric in the interview: does it move when the core product loop (successful, high-quality matching) actually succeeds; can it be gamed in an obviously bad way (restricting driver supply to inflate utilization, for instance); and does moving it plausibly lead to the real business outcome (sustainable transaction volume without runaway subsidy cost). Match rate segmented by corridor passes all three; raw ride volume alone fails the second and third.
Takeaway
For a two-sided, logistics-heavy marketplace like carpooling, the strongest metrics answers pick a north star that captures successful matching rather than raw volume, then build a small tree of driver-side, rider-side, and guardrail metrics that would catch a marketplace imbalance before it becomes a churn crisis. This same pattern — pick the metric that captures a successful two-sided match, not just a top-line count — transfers directly to other marketplace products: food delivery, freelance marketplaces, and short-term rentals.
For more worked examples of marketplace metrics trees and north star selection, see The 100x Product Manager Interview Playbook (Amazon: https://www.amazon.com/dp/B0DBC1FQWH?tag=sirjohnnymai-20).