Don't Guess — Validate
The most expensive mistake a platform team can make is assuming they know what developers need. Platform engineers are often highly skilled infrastructure specialists who find complex systems intrinsically interesting. This is a liability when building a product, because your customers — application developers — often do not share those interests. They want to ship features, not configure service meshes.
The discipline of User Research exists precisely to bridge this gap. Before writing a single line of infrastructure-as-code for a new platform feature, you need evidence that the feature solves a real, painful problem.
Research Techniques
1. Empathy Interviews
Sit down one-on-one with developers from different teams, roles, and experience levels. The goal is not to pitch your ideas — it is to listen. Effective questions include:
- "Walk me through the last time you deployed a new service. Where did you get stuck?"
- "What is the most frustrating part of your day-to-day workflow?"
- "If you could make one thing disappear from your process, what would it be?"
- "What tools do you avoid using, and why?"
Avoid leading questions like "Would you use a self-service database portal?" (Everyone says yes to hypotheticals.) Instead, ask about past behavior: "Tell me about the last time you needed a new database. What did you actually do?"
2. Observation (Shadowing)
Watch a developer attempt to deploy a new service from scratch, without helping them. Note every moment of confusion, every tab they open to search for documentation, every Jira ticket they have to file. This is often more revealing than interviews, because people don't accurately remember their own pain — they normalize it.
3. Quantitative Surveys
Use surveys to validate and prioritize findings from interviews across a larger population. Example: "On a scale of 1–10, how painful is managing database credentials?" If 80% of developers rate it 8 or higher, that is a validated product opportunity.
4. Telemetry and Support Ticket Analysis
Look at your existing support tickets and incident reports. Which topics generate the most tickets? Which steps in your existing pipelines fail most often? Data from your current systems is a goldmine of product insight that requires no additional research effort.
If your research reveals that developers are spending 30% of their time managing database credentials, that is your product opportunity — not building a slightly more colorful CI/CD dashboard.