Case study: Self-serve analytics for 20,000+ billers

Product owner, Biller Analytics Dashboard, NPCI Bharat Billpay (NBBL), 2022-2025. Adoption figures below are aggregate metrics; the linked demo reproduces no NBBL data or designs.

Context

Bharat Connect (formerly Bharat Billpay / BBPS) is India's interoperable bill-payments network: 20,000+ billers across 30 categories collecting through a central platform that sits between customer-side and biller-side operating units. The billers - electricity boards, telcos, insurers, lenders, universities - had no analytics tool at all for their collections through the network. A biller's operations team could not see its own volumes, decline patterns, or complaint load without asking someone.

The product problem

The hard part was not building charts. It was choosing metrics that drive action across 30 categories of wildly different sophistication - a state utility and a fintech-native lender cannot be judged on the same baseline, and a vanity dashboard would die quietly. The metric set shipped: transaction volume, collections value, technical decline rate, business decline rate, and complaint counts.

The decline split is the design decision I defend hardest. Technical declines (gateway failures, timeouts, connectivity) are the platform side's job to fix. Business declines (validation failures, fetch mismatches, customer-side issues) belong to the biller's own funnel. One metric split into two tells every viewer whose move it is - that is what turns a report into a routing mechanism.

Design choices

• Self-serve over reports-on-request. Each biller's SPOC logs in and sees their own numbers. The alternative - a central team fielding 20,000 billers' report requests - does not scale and was the status quo failing.
• Benchmark within category, not across the network. Mix differences would otherwise make every comparison misleading.
• Normalise complaints per transaction, not raw counts, so big categories do not automatically look worst.

Outcome

• ~50,000 monthly active users at ~30-minute average session length
• Analytics report downloads grew from ~100 in month 1 to ~15,000 monthly by month 12 - 150x adoption growth
• First biller-facing analytics surface in the network's history, replacing zero prior visibility

Interactive demo

A working demonstration of this product thinking - same metric architecture, fully synthetic data - is available as a single-file interactive dashboard: the interactive demo. Filters, category benchmarks, and the technical-vs-business decline split are all live in the demo.

Ruturaj Sahasrabudhe | ruturaj.mms@gmail.com | linkedin.com/in/ruturajsahasrabudhe