Who This Helps
This is for founder operators who are drowning in data but starving for decisions. You have a hundred ideas for the next experiment, but only time for one. You need a way to cut through the noise and pick the move that will actually move your metric.
Mini Case
Meet Mei. She runs a 12-person startup. Trust in her numbers is broken. Every week, the team debates which experiment to run next. Last month, they ran three experiments that showed no impact. Turns out, the data feeding those experiments was wrong. Mei spent 7 days fixing a data pipeline instead of shipping a feature. After she defined a data contract for her key metric, she cut experiment waste by 40%. Now she picks the next experiment in 3 steps, not 3 meetings.
Do This Now (5 Steps)
- Pick one metric that matters most. Not all metrics. Just one. That's your north star for this week.
- Write a data contract for that metric. Define exactly where it comes from, how it's calculated, and who owns it. This is from the "Data Contracts" mission in the Data Reliability Leadership course.
- List your top 3 experiment ideas. Write them down. No judgment. Just get them out of your head.
- Score each idea against your metric. Ask: "If this works, will it move my north star metric by at least 10%?" If not, drop it.
- Run the highest-scoring experiment first. Commit to one. No multitasking. Ship it by Friday.
Avoid These Traps
- Don't pick a metric that's easy to measure but doesn't matter. That's busywork, not progress.
- Don't write a data contract alone. Get the person who owns the source data to sign off. Otherwise, it's just a wish.
- Don't run two experiments at once. You'll learn nothing. One at a time.
- Don't skip the scoring step. Gut feelings are great for dinner, not for experiments.
- Don't wait for perfect data. Good enough is better than perfect. Ship it.
Your Win by Friday
By Friday, you will have one data contract written for your key metric. You will have picked one experiment to run. And you will have started that experiment. That's it. One metric. One contract. One experiment. You'll move faster than you have in weeks. And you'll actually know if it worked.