The DEI metric reported without context
Twenty-eight percent. The number went into every board deck. Nobody asked what it measured, against what benchmark, or whether the trajectory was right.
In one company's board deck, the same slide had sat for three quarters: 28% of engineering is female. The board nodded at it each time. Nobody asked a question. In one review I finally asked the founder what the trajectory was: up, down, or flat? He didn't know. I asked the benchmark for his industry and geography. He didn't know. I asked which of last year's hiring decisions had moved the number, and in which direction. He didn't know.
The metric was being reported, not managed. That figure was a true fact about the company and a number nobody owned. The board's silence wasn't indifference; it was the rational response to a figure with no context to discuss. No comparison. No trajectory. No named owner. No target. Nothing to grab onto.
This is the most common failure in DEI reporting at growth-stage companies. A single number, X percent of engineering or Y percent of leadership, carries almost no operational information. It could be climbing or sliding. It could beat benchmark or trail it. It could be the product of deliberate work or pure luck. None of that is visible in the headline, so the board can't ask a useful question and the company can't make a useful decision. The number shows that DEI is being reported, which isn't the same as being managed.
The real diagnostic means taking the funnel apart. Hire composition is the lagging indicator: by the time it moves, the decisions that moved it are months old. The leading indicators sit upstream. Pipeline composition, who applies. Interview composition, who clears the screen. Offer composition, who gets the offer. Accept composition, who says yes. Each one is measurable, and the shape of the funnel shows which stage is producing the gap.
A funnel that looks fine until the interview stage and then drops is pointing at the interview: the screen, the panel, or how decisions get made. A funnel that holds until offer acceptance is pointing at comp or culture, where candidates clear the process and pick somewhere else. A funnel that's thin at the application stage is a sourcing problem. Different stages, different fixes. One hire-composition number can't tell you which one you have. Screening for "fit" quietly does a lot of this damage, which is why I argue you should stop hiring for culture fit.
What produces movement is the same discipline as any operating metric. Named owner. Specific targets. Quarterly review. Accountability tied to the target. Most DEI work fails not because the company doesn't care, but because it's framed as everyone's responsibility, which operationally means no one's.
There's a defensive habit underneath all of this. A company wants to show progress without committing to accountability that might fail in public. A single number is the ideal artifact for that posture: celebrate it when it rises, wave it away when it falls, and rarely examine it closely enough to start an uncomfortable conversation. The same reflex shows up whenever a vanity metric lands on the homepage, and it's why a headline figure so often turns out to be measured on the wrong cohort.
Report the trajectory. Report the funnel. Name the owner. Set the targets. Today's DEI numbers are measuring last year's hiring decisions. If the number isn't where you want it, the lever is upstream, in the process that produced it.