Building Stuff Nobody Wanted (Because We Never Asked)
The Problem
Analytics team spends 6 months building a real-time customer segmentation engine. It's beautiful. Technically impressive. Launch day comes. Marketing uses it once and never again. "Too complicated." "Not what we needed." "We actually just wanted a simple email list." Six months of effort, zero ROI. This happens constantly—teams build what's technically cool instead of what's actually valuable. Nobody measured: Will this drive revenue? Will it save time? Will anyone use it? We just assumed it would be great. It's like a manufacturing plant making a product nobody ordered. It sits in the warehouse gathering dust while the factory burns cash.
The Principle
Before you build anything, define success: What's the ROI? Who will use it? How will it change their behavior or drive business outcomes? Get commitment from stakeholders upfront. Think manufacturing: you don't make widgets until you have orders. Same here—don't build data products until you have buyers. The magic happens when every project has a clear business case and you kill anything that doesn't deliver. Less building for the sake of building, more building what actually matters. That's how you become a powerhouse.
3 Steps to Take Action
- 1
Define Success Before Building: Write down: Who will use this? How will it change their behavior? What's the ROI? Get stakeholder signatures. No project starts without a committed user.
- 2
Show Prototypes Early: Build ugly mockups or wireframes. Show them to users in week 1. Get feedback. Iterate. Better to pivot early than deliver the wrong thing perfectly.
- 3
Measure Adoption from Day 1: Track usage from launch day. Set a target: "X% adoption in 90 days." If you're not hitting it, kill it or fix it. Don't let zombie projects linger.
In Practice
A data team at a logistics company built a machine learning model to predict delivery delays. Took 8 months. Accuracy: 91%. Presented it to operations. Response: "Cool, but we can't do anything about delays—our trucks are already committed." Model was technically perfect, operationally useless. Zero ROI. New data leader instituted a rule: "No project starts without a signed business case and committed user." Next project: simple dashboard showing which trucks were running late so dispatchers could reroute. Built in 3 weeks. Operations used it hourly. On-time delivery improved 12%. Revenue impact: $4M annually. The fancy model gathered dust. The simple, useful tool became mission-critical. Build what people will use, not what sounds impressive.
