Engagement scores are lying. Trends are honest.
Engagement scores are lying. Trends are honest.
Most engagement vendors hand you a number. 78. 7.4 out of 10. "Top quartile." Your CEO sees the number, files it under "we're doing well" or "we have work to do," and the conversation moves on.
The number is the worst part of the report.
What a single number can't tell you
Two teams can score the same and be in completely different shapes. A team scoring 78 because most people are quietly content looks the same as a team scoring 78 because half love it and half are about to quit. The score averages those out and hides the bimodal distribution that's actually predictive of attrition.
A score doesn't know what changed. If your number went from 81 to 78 over six months, the score doesn't tell you whether something specific happened in March, or whether a slow drift has been compounding since November. Same delta, different problems, different fixes.
A score doesn't know what's local. The number is an org-wide average. The actual problem is almost always inside one team, one manager, one office. The aggregate hides exactly the room where action is needed.
What trends do tell you
When you measure the same thing every day, you stop caring about absolute scores and start caring about movement. A team where "I felt heard this week" trends downward four weeks running is in trouble, regardless of what its number is. A team whose "workload feels manageable" climbs after a hire is telling you that hire is working.
Movement is what you can act on. A bad number doesn't tell you which week, which team, which intervention. Movement points at the date, the cohort, the cause.
The annual survey can't show you trends because it only has one data point. The biannual survey gives you two, which is two more than zero but still useless for spotting the slope. You need at least weekly cadence before "trend" means anything.
The benchmarking trap
Vendors love showing you a benchmark band. "You're in the 67th percentile for your industry." This is the part of the report executives quote in board meetings.
Industry benchmarks have two problems. The first is that they describe the average company in a category that includes companies with very different operating models, which means the comparable is nearly random. The second is that even if the benchmark were perfectly calibrated, beating it tells you nothing about what to do tomorrow morning. Your team's specific friction is almost never the kind of thing that comes out in a peer-quartile chart.
What you want isn't "are we above average." It's "is this getting better or worse, and where, and why."
What to ask
Stop asking yourself "what's our engagement score." Start asking "in which dimension, on which team, has the slope been negative for three or more weeks." That question has a fix attached. The score doesn't.
The score still has a use. It makes for a tidy slide in the board deck. But it shouldn't be how you decide where to spend your management attention. The trend should.