In this episode of Product Unplugged, host Mike Belsito sits down with Nimit Bhardwaj, a data product manager at Booking.com, to discuss how his product team faced a high-stakes data quality issue and walked us through the process they used to solve it.
Key Takeaways
- “Treat data like a product.” Ownership doesn’t end at building features—it extends to the data those features create.
- Short-term fixes matter. A tactical “quarantine” service helped buy time while driving long-term system change.
- Don’t underestimate 1%. A 1% error rate in financial data still led to 40,000+ manual corrections each month.
- Data quality isn’t one team’s job. True success came when upstream teams took responsibility for their data’s accuracy.
- Measurement is meaningless without messaging. 99% sounds fine—until you translate it into delayed payments and angry suppliers.
Chapters
00:00 – Intro & Meet Nimtt Bardwage from Booking.com
01:45 – The big data quality problem that couldn’t be ignored
04:00 – Root causes: legacy systems, knowledge gaps & migration chaos
06:00 – When 99% data accuracy is a massive risk
08:00 – The two-track approach: tactical + strategic
10:00 – Building a quarantine service to protect financial systems
12:00 – From patch to platform: scaling into strategic change
14:30 – The results: automation, ownership, and 12x improvement
16:00 – What we’d do differently next time
18:00 – Common pitfalls to avoid when dealing with data quality
20:00 – Everyone owns the data
Keep reading
The future OS: Why data will define the next decade
How to use churn data to build stickier products
How to deal with information overload in product management
How to quantify the value of UX with stakeholders