What a forensic team taught me about yield

Every process that transforms raw material into a result is a funnel. A factory has one. A laboratory has one. A sales team has one. And, as I learned in twenty years inside a federal public safety agency, a crime scene also has one.

The number nobody wanted to look at

In a public forensic team I worked with, the question was simple: for every call that takes the team to a scene, how many end with an identified suspect?

Not "how many reports we deliver". Not "how many pieces of evidence we collect". Every call — with or without evidence, in good condition or not — went into the denominator. Measured with such honesty, the answer was single-digit, and low.

Most organizations never look at this number, because the intermediate numbers look much better. Reports were delivered. Evidence was collected. Everyone was busy. The waste was invisible because it was spread across all stages.

Measure each step, and then find the leak

So we measured each step: the call, the evidence collected, the evidence suitable for comparison, the database search, the identified suspect, and — the real goal — the conviction.

The funnel told a story no one expected. We weren't losing cases where everyone imagined — at the scene, during collection. Collection went well. We lost soon after, in quality: most of what we collected was not suitable for comparison.

This completely changed the conversation. Instead of "work harder at the scene", the question became "how do we know, still at the scene, if what we are collecting is usable?". This justified new equipment specifically for that stage, and training focused exactly on that stage. Money and effort went to the only place where they would pay off.

The second lesson: the average is a hiding place

Team numbers on a live dashboard helped. The rate went up.

Then we did something a little uncomfortable: each person got a bar with their own name. And the data showed that some of us — myself included — were hiding in the group's average. Not out of laziness. Out of ambiguity. When a result belongs to everyone, it belongs to no one, and no one really knows if they're doing well.

A year later, no one was below a level that had previously been the team's best. In about three years, the rate was approximately ten times the initial one. And whenever someone identified a suspect, the dashboard played the music chosen by that person for the entire room. Sounds silly. It worked.

People rarely hide from work. They hide from ambiguity.

Why I think this matters outside of policing

Today I read about labs trying to manufacture organs, or running thousands of experiments in parallel, and I see the same format. A yield funnel with losses at each stage, where intermediate numbers look healthy and the final number is what matters.

The lessons I would take are small and unglamorous:

None of this is new. What's rare is to do it honestly, in a place where the numbers are uncomfortable.

I'd love to find out if this transfers.