Analytics - the single biggest competitive advantage companies have but not many leverage

In TaxiForSure's early days, they were averaging 8-10 bookings a day when one customer stood out: 50 rides in a single month. That outlier led to an insight that changed how they thought about growth. It revealed why the customer was using the service, not just that she was using it.

The customer was a pregnant working woman. The taxi was her daily commute to office on weekdays and to yoga classes on weekends. No husband would question the spend when safety was at stake.

This single observation, from looking at individual user data, led them to market at yoga and aerobics classes across the city. By month's end, bookings had jumped to 40 per day. They secured funding by January and began scaling.

I had the chance to hear this story directly from Raghunandan G, TaxiForSure's co-founder and CEO. What stuck with me wasn't just the tactics. It was the underlying principle: while most competitive advantages can be copied, data compounds. The more you gather, the better your insights become over time.

Finding the Loyalty Threshold

Early on, TaxiForSure faced a classic allocation question: focus on acquiring new users or retaining existing ones? The team was divided.

The data revealed a threshold: users who completed more than 5 rides became significantly more tolerant of service issues. They'd stick around even if a few calls were missed. So the company shifted resources toward helping new users cross that 5-ride threshold as quickly as possible.

This pattern appears everywhere. Facebook famously discovered that getting users to 7 friends in 10 days was the key to long-term engagement. The specific number differs by product, but the principle holds: find your loyalty threshold and optimize ruthlessly for it.

Shaping Demand to Fix Supply

TaxiForSure had a driver supply problem. Most drivers worked the morning peak, went home during the afternoon lull, and didn't return for the evening rush. Inertia kept them away.

The insight came from booking data: most rides were booked 1-2 hours before pickup. So they invested heavily in morning Google ads to extend demand past the peak hours, keeping drivers engaged through the day. Evening ads captured bookings for the next morning.

The result: they flattened both supply and demand curves. Overall bookings doubled from 600 to 1,200 per day within a month. The drivers earned more, stayed motivated, and the supply problem solved itself.

Optimizing for Both Sides

Here's a marketplace optimization problem: two booking requests come in, one to Yeshwantpur, one to Whitefield, but only one cab is available. Which do you accept?

The data showed Whitefield was a favorable destination: drivers could easily find return rides. Yeshwantpur left them stranded. Accepting the Whitefield booking meant 2 happy riders and 1 happy driver. The other option meant 1 happy rider and a frustrated driver who'd cancel future bookings.

This led to a broader insight: studying cancellation patterns revealed which destinations were unfavorable by time of day and day of week. Drivers would accept bookings, then cancel upon learning the destination. The fix was simple: ask for the destination upfront. This feature is now standard across all ride-hailing apps.

Applying These Lessons

These principles have shaped how I think about data across my own work:

At Balance, we looked at individual user journeys to find outliers: people saving tens of thousands in months. Understanding their patterns helped us refine the product for everyone else.

At Paytm, we used similar individual-level analysis to identify high-ticket users whose issues predicted churn.

The enduring lesson: the best analytics doesn't just measure what happened. It exposes a behavioral truth you can build a strategy around. Once you find that truth, small teams can move like giants.