Why One Metric Beats a Dashboard Full of Them
You're tracking DAU, retention, feature adoption, NPS, and churn. Your co-founder watches revenue. Your investor asks about engagement. You're optimizing for five different directions simultaneously—which means you're optimizing for nothing.
YC's North Star Metric framework solves this. It's the single number that moves when you're moving closer to product-market fit. Everything else is noise.
Sean Ellis (who coined the term at Dropbox) studied 100+ high-growth startups. The winners all had this: one metric they obsessed over, ruthlessly, for 6-18 months. Instagram: daily active users. Slack: message volume. Stripe: API calls processed. Not a mix. One.
How to Pick Your North Star (The Framework)
Three filters. Use all three.
Filter 1: Does it predict revenue in your model?
For Uber Eats, daily orders per city predict revenue. For a B2B analytics tool, monthly recurring active users predict MRR. For a creator platform, average posts per creator predict sponsorship demand.
Drop it if it doesn't cash-convert eventually. Day-7 retention doesn't matter if users never buy. Revenue doesn't matter if you're not sticky. Find the causal link.
Filter 2: Can you move it in 4 weeks?
If your metric moves every 6 months, you're flying blind. You need weekly signals. Slack measured daily active conversations. They could see impact in days. Airbnb measured bookings per 100 views—visible within weeks.
For Indian fintech startups, this might be: transactions per active user (weekly signal). For B2B SaaS: features used per account (daily signal). Not annual churn or quarterly revenue—too slow.
Filter 3: Can you isolate cause from noise?
If your metric depends on 50 variables, it's not actionable. You change something. Does the metric move? You need to know.
Bad North Star: "monthly engagement score" (undefined, 12 variables).
Good North Star: "messages sent per active user per day" (one lever, clear causation).
The Three Player Types: Pick Your Metric by Business Model
Revenue-First (B2B SaaS, marketplaces with take-rate)
Your North Star: MRR or bookings volume (weekly).
Why: Runway is limited. You need revenue proof before month 6. Stripe picks API calls → revenue correlation. A16z portfolio companies track ACV × deals closed weekly.
Indian B2B SaaS founder? Track: active subscriptions generating >$100/month. That's your real signal.
Retention-First (Consumer, communities, Creator platforms)
Your North Star: day-7 retention (%) or weekly active users.
Why: Acquisition is cheap in India. Retention is the filter. Instagram: daily active installs → day-7 retention >40%. If you fix that, monetization follows.
Indian social app founder? Don't track signups. Track: 7-day retained users who post or engage at least once. That number predicts viral growth.
Engagement-First (Startups in exploration mode)
Your North Star: core action frequency (daily/weekly).
Why: You don't know your revenue model yet. You're betting on engagement. Slack: messages sent per active workspace per day. LinkedIn: connection requests and profile edits per user per week.
Your metric: "aha moments" completed per user per week. That's the leading indicator for everything else.
The Non-Obvious Play: Pick What Your Competitors Ignore
This is the insight most founders miss.
If every competitor tracks "user signups," track "power user signups only." If everyone optimizes DAU, optimize for deeply engaged sessions (2+ features used per day). If the category watches "monthly revenue," watch "revenue per power user."
Why? You're smaller. You can't win on their metrics. But you can identify what they're leaving on the table.
BY locals (Delhi logistics) didn't track deliveries. They tracked deliveries per delivery partner per day. Smaller number, higher signal of operational health. That obsession led to their Series A.
The Messy Middle: What Happens When Your Metric Plateaus
Scott Belsky's concept applies here: you'll pick a metric, optimize it to 70%, then hit a wall. The curve flattens.
That's normal. When it happens:
1. Validate your metric is still predictive. Maybe you optimized a local maximum, not the true North Star.
2. Add a secondary metric. Only after you've maxed the first.
3. Don't abandon—evolve. Slack went from "daily messages" to "message threading depth" when DAU plateaued.
Indian founders often panic and switch metrics every month. Resist that. Three months minimum before switching. If you're jumping metrics weekly, you're not optimizing—you're scattered.
Your Action Plan
This week:
1. Write down your current top 3 metrics.
2. Which one predicts revenue in your model? (Kill the others.)
3. Can you move it in 4 weeks? If no, replace it.
4. Pick one. Set a 12-week optimization target.
First 4 weeks: Move it 5-10% minimum. If you can't move your own metric, it's wrong.
Weeks 4-12: Optimize ruthlessly. Build one feature. Measure. Iterate.
That's it. One metric. Twelve weeks. Everything else is distraction.