Signal › Contributors

Priya Sharma

Data & Analytics at Signal · Bangalore, India

I studied statistics at IIT Bombay and then did what every statistics graduate in India does: I ignored statistics and joined a management consulting firm. Two years at McKinsey taught me how to make PowerPoint decks look authoritative. It did not teach me how to think clearly about data.

I left consulting in 2019 for Swiggy, India's largest food delivery platform, where I joined the analytics team. Swiggy was processing millions of orders a day, and the data infrastructure was, let me be diplomatic, evolving. My first project was building a customer segmentation model that would inform marketing spend allocation. The model worked. The marketing team ignored it for six months. I learned that the hard part of analytics isn't the analysis. It's the organizational change management required to get anyone to act on the analysis.

Over three years at Swiggy, I built out the product analytics function from scratch. We went from "someone pulls a SQL query when the CEO asks a question" to a proper experimentation platform with automated significance testing, holdout groups, and pre-registered hypotheses. The experimentation platform alone saved the company an estimated $8M in the first year by killing features that A/B tests showed were actually hurting conversion.

In 2022, I moved to Razorpay, India's largest payment gateway, as Head of Product Analytics. Payments data is a different beast. The volumes are enormous, the stakes are high, and the regulatory requirements around data handling add a layer of complexity that most consumer companies never deal with. I built the analytics infrastructure for their lending product, which went from zero to $500M in disbursements in 18 months.

I started writing because I was tired of "data-driven" being used as a synonym for "we looked at a dashboard once." Most companies aren't data-driven. They're data-decorated. They have dashboards that nobody looks at, A/B tests that aren't properly powered, and "insights" that are really just confirmation bias wearing a lab coat.

I joined Signal in late 2024 after a Twitter thread about common A/B testing mistakes went viral (2M impressions, which is ironic because I know exactly how meaningless that metric is).

I live in Bangalore. I drink too much filter coffee and I will argue about Bayesian vs. frequentist statistics at parties. I am not invited to many parties.

Experience

• Head of Product Analytics, Razorpay
• Analytics Lead, Swiggy
• Associate, McKinsey & Company

Articles by Priya Sharma (14)