Early in my career, I built a feature that was supposed to change everything.
The stakeholder was convinced. “This will transform how the team works,” they said. I took it at face value. We spent weeks on it. Good code, solid architecture, all the edge cases handled. Shipped it. Nobody used it.
The feature fixed a symptom, not the actual problem. By the time I’d finished building, someone else had solved the root cause a different way. The feature just sat there, working perfectly, doing nothing.
That experience stuck with me. I’d built exactly what was asked for without questioning whether it was the right thing to build.
AI makes this mistake easier to make. AI can generate specs, scaffold code, write tests. The gap between “someone suggests a feature” and “feature is built” keeps shrinking. But speed doesn’t help if you’re heading in the wrong direction.
How common is the wrong direction? Pendo analyzed 615 SaaS subscriptions and found 80% of features are rarely or never used. An average of 12% of features generate 80% of daily usage volume. They measured click frequency, so a compliance tool clicked once a quarter could still matter. But the pattern is clear. Most of what gets built does not get pulled into anyone’s week. Pendo 2019 Feature Adoption Report
Read More »