The engagement survey came back at seventy-one per cent. Three focus areas were identified, an action plan was circulated with owners and dates, and a year later the same three areas appeared again with the same scores and slightly better wording. Nobody had lied. The plan had been delivered. The scores had not moved.
The average hides the only thing worth looking at
That seventy-one per cent is not a fact about the company. It is arithmetic performed on very different experiences. The same survey can contain a unit at eighty-eight and another at forty-nine, and the average will sit in the middle looking respectable.
What varies between those units is mostly the manager: whether the one-to-one happens, whether a problem raised goes anywhere, whether good work gets named in the week it happened. HR can set the conditions for those things. It cannot produce them, because they occur in a calendar between two people.
This is why engagement programmes so often produce communication rather than change. The part that can be procured gets funded. The part that only managers can do gets announced at a town hall and then quietly not done.
HR can deliver pay, policy, career architecture, a learning budget, the survey instrument, and the standard that says one-to-ones must happen. That is a substantial list and none of it is optional.
It cannot make a specific one-to-one happen next Tuesday. It cannot decide whether a manager answers an awkward question honestly or routes it to somebody else. Four things a survey will never tell you, and a team can:
- whether the one-to-one happened in each of the last three months
- whether the problem raised in June was closed, and how long it took
- whether the person who fixed Saturday’s outage was named on Monday
- whether anyone has asked them what they want to be doing in two years
The mistake is treating the second list as a communications problem. It is a management practice problem, and it responds to different things: a smaller span of control, a visible standard, and someone senior asking about it by name.
Four things that actually move it
- A one-to-one that happens. Quality matters, but it matters second. A one-to-one cancelled three weeks running is a message, and the message is not “I am busy”. It is “you are not a priority”, which the person will believe far more readily than anything said in the fourth week.
- Work with a visible end. People tolerate hard quarters. They do not tolerate work that never closes. If nothing they do is ever finished, the job becomes a treadmill, and a treadmill with better lighting is still a treadmill.
- Recognition from somebody who saw it. The annual award is not recognition. It is a lottery with a stage. The useful version is a named sentence, from a person who was in the room, within the week, describing what was done and why it mattered.
- A route for a problem that does not require courage. If raising an issue costs the person something, they will not raise it. You will hear about it in the exit interview, where it arrives expensive and roughly four months too late to act on.
Measuring it without the survey theatre
You do not need another instrument. Four of the five numbers below already exist in payroll, the HR system or a shared calendar.
| Measure | What it tells you | How often | Who owns it |
|---|---|---|---|
| Voluntary exits under twelve months | Whether hiring and onboarding match reality | Monthly | HR |
| Regretted attrition | Whether you are losing people you wanted to keep | Quarterly | Leadership |
| Internal move rate | Whether people can see a future here | Quarterly | HR and line |
| One-to-one completion | Whether the basic ritual is actually happening | Monthly | Line manager |
| Days to close a raised issue | Whether the route for problems is real | Monthly | Line manager |
The first row is the most honest number in the table. A new joiner has no loyalty to protect and no history to reinterpret, so their decision to leave within a year is close to a direct reading of what they met. Pull it by manager, not by function, or you will average away the only pattern worth seeing.
The load problem
There is a limit that engagement programmes routinely ignore. If a manager has eighteen direct reports and a delivery target built for eight, no amount of training will produce weekly conversations. Eighteen people at thirty minutes each is nine hours a month before any of the work they were hired for, which is illustrative arithmetic but the right order of magnitude.
The engagement problem is often a span-of-control problem wearing an engagement costume. Buying a programme in that situation is a way of asking a manager to do something the arithmetic does not permit, and the manager will correctly conclude that the request was not serious.
Engagement is not delivered to a team by a programme, it is built or eroded by a manager in ordinary weeks.
What to do in the next quarter
- Pull first-year voluntary exits by manager. Not to punish anyone, to see where the pattern sits. Two managers accounting for eight leavers between them is a fact worth understanding before you buy anything.
- Ask three teams one question, in person, with no survey. What makes your week harder than it needs to be? Write the answers down verbatim, because the wording is the diagnosis.
- Fix one thing within six weeks and tell them you fixed it. The telling is the intervention. It is the evidence that raising something works, and without it the next suggestion costs more than the last.
- Give the one-to-one the same calendar status as a client meeting. Move it, do not cancel it. A moved meeting is a scheduling problem. A cancelled one is a statement.
- Only then design the programme, scoped to what HR can actually deliver and what managers can actually sustain.
If the choice is between a new engagement platform and making sure forty managers hold a fifteen-minute conversation this month, take the second. It is cheaper, it is less impressive in a review, and it is the one that shows up in the exit data. Where the pattern sits across several units, a diagnostic piece of work is usually more useful than another survey, and an assessment of the management population will tell you whether the gap is skill, span or standard. The three have different fixes and the same symptom, which is why so many engagement plans treat all three as one problem and move none of them.
One framing is worth borrowing from the three-I approach to building a workforce: interesting work, individual growth and involvement. Those are the three things people actually describe when they explain why they stayed, and none of them can be bought as a platform licence.