Blog 01 · data-driven design

We had the data. We still designed for the wrong customer.

Pranjal Chawan
Pranjal ChawanUX leader · LinkedIn

Here is an uncomfortable truth about data-driven design: you can have dashboards, stakeholders and a sensible plan, and still build the entire experience for someone who is not your customer.

I know, because I have done it.

A bank asked us to grow daily Fixed Deposit bookings from 5,000 to 10,000. The catch was that the backend was completely off-limits. Every gain had to come from the experience layer.

Stakeholder discussions pointed to one logical audience: people aged 35 to 50. We built the landing page, key screens and messaging around them. Then the redesign failed in testing.

That failure was the most useful moment of the project.

The audience we assumed was not the audience we had

Instead of polishing a broken assumption, we went back to real behaviour and data from people using the banking app. The customers most likely to book a Fixed Deposit were not 35 to 50. They were 25 to 30.

We rebuilt the landing page and every key screen around that audience, including its motivations, trust signals and pacing. We tested with a small group before rollout. Daily bookings rose from 5,000 to nearly 8,000 within three months, about a 60% increase, without a single backend change.

We did not reach 10,000. I tell you that because honest numbers are part of the job. You can read the full Fixed Deposit case study here.

Data does not make a team data-driven

Most teams already have data. What they lack is a habit of letting it disagree with them. McKinsey's Business Value of Design study found that just over half of surveyed companies had no objective way to assess or set targets for their design teams' output.

A dashboard answers the question you asked. It will not tell you that you asked the wrong one.

Five habits of real data-driven design

  1. Write the assumption down. "Our customer is 35 to 50" is a hypothesis until behaviour supports it.
  2. Look for evidence that could prove you wrong. Confirming data feels good. Contradicting data saves money.
  3. Pair the what with the why. Analytics show what happened. Conversations and usability tests explain why.
  4. Agree on the metric before the design. Otherwise, every result can be explained away later.
  5. Test small before you roll out. A failed test costs far less than a failed launch.

What this means for leaders

My first job, in 1997, was not in design at all. At Samsonite, I analysed sales figures and built MIS reports that sales teams used to hit their targets. That is where I learned that data only matters when someone can act on it. It is also why I now work to replace gut-feel design with data-driven frameworks.

The question I bring to any new product is simple: what do we believe that we have never tested?

That question can feel awkward in a room full of senior people. It is also the cheapest risk reduction you will ever get. My executive programme in Data Driven Product Management at IIT Guwahati gave me more structure for that conversation, but the lesson came from a failed test.

Questions people ask

What is data-driven design? It means using evidence about real user behaviour to shape and test design decisions, instead of relying on opinion or seniority.

Is data-driven design the same as A/B testing? No. A/B testing is one method. Data-driven design also includes analytics, interviews, usability testing and clear success metrics.

Let's compare notes

Is your team confident about a customer it has never properly tested? I would love to hear about it. Write to me at pranjal.productmind@gmail.com or say hello on LinkedIn. I read every message.

© 2026 Pranjal Chawan. All rights reserved.