The Engagement Survey That Bought Six Months and Not a Culture

The engagement score came in at seventy-eight. Up from seventy-two the year before. The CHRO presented the results at the leadership offsite. The CEO used the number in three subsequent board meetings.

Attrition in the highest-performing segment doubled over the following two quarters.

The CHRO stood in front of the board and could not explain the gap. Every engagement lever she had pulled had moved. Every retention lever had failed.

What the survey was measuring

The survey was measuring a specific thing well. It was measuring how employees felt about the company in the abstract, at the moment they filled out the survey, in language the company had chosen.

That measurement responds cleanly to specific interventions. Free lunch improves it. A learning stipend improves it. A better manager training program improves it. Anything that raises the average sentiment across the population will move the aggregate score.

The score does not respond to the specific thing that causes high performers to leave. High performers do not leave because they are unhappy in the abstract. They leave because a specific pattern of small interactions with a specific person, usually their manager or their manager's manager, has become emotionally expensive in a way the population average will never detect.

A CHRO who moves the aggregate score has done real work. She has also almost certainly not moved the number that matters, which is what one particular high performer's nervous system is doing in the leadership meetings she attends.

How much is one unresolved conflict really costing your company?

What the CHRO started measuring instead

She did not throw out the engagement survey. She kept it as a background metric and stopped using it to justify anything.

She built one new measurement. Every quarter, she asked each people manager to name the three highest performers on their team and to answer one question about each of them. "Compared to six months ago, is this person more or less likely to volunteer an unpopular opinion in your team meetings?"

The question was designed to catch a specific thing. Voice going quiet is the earliest signal a high performer's nervous system is beginning to protect itself from the room. It shows up months before disengagement scores move. It shows up before deliverables slip.

The first quarter of asking that question, three managers named eleven high performers between them. Six of the eleven had gone quieter. Two of the six had already begun interviewing. The CHRO had a list of specific conversations that needed to happen inside the following four weeks, and a specific set of manager behaviors that needed intervention.

She caught two of the two departures. Not because she had improved engagement. Because she had installed a signal the engagement survey was never designed to catch.

Doug Noll's new book Empathy Leadership: The Powerful Skill That Drives Winning Results lays out the signals people leaders use to see what a survey cannot see. See it here.

What she told the board the following year

The engagement score was stable. Attrition in the high-performer segment was down materially.

She told the board one sentence that she had not been able to say the year before. "The score is a lagging measurement of aggregate sentiment. It does not predict who will leave. We have installed a leading measurement that does."

The board asked what the leading measurement was. She explained the question and the process. One director, a former CEO, said, "You are asking managers to notice something they usually cannot articulate." She said, "Yes. And it is trainable."

She did not need to defend the engagement survey. She had stopped using it as a proxy for something it could not measure. She had also stopped promising the board that a higher score would produce lower attrition. It never had. The company had spent two years running on that assumption. It had cost, by her rough accounting, three of the seven senior people she wished had stayed.

‍