Gallup published its third-quarter US workforce figures on 28 October. Employee engagement fell to 29%, its lowest level since 2011 (Gallup, U.S. Employee Engagement Below 30% for First Time in 15 Years).
The figure I keep returning to is further down the same article. The share of employees who strongly agree that they know what is expected of them at work fell three points to 46%, matching the 2024 low for that question (Gallup, same article). Gallup surveyed 21,648 adults on the Gallup Panel in August.
Gallup does not say why clarity fell, and I will not pretend to know. What I can offer is a reading from more than 25 years of working on how people explain things at work.
An expectation has two ends
An expectation has a speaker and a listener. Managers tend to check it at their own end. "I covered that in the review." "It was in the kickoff deck." "I mentioned it on Monday's call."
Each of those statements can be true while the person on the other end still could not tell you what good looks like this month.
In my experience, the gap opens in a few predictable places:
The expectation describes an activity. "Work on the launch" leaves the finish line open. "The launch page live by Friday, with legal sign-off" gives a result someone can check.
The standard lives in the manager's head. The manager knows what a good client update looks like. Nobody has written it down or shown an example.
Priorities change and the update travels badly. A forwarded message or a line in a meeting replaces the conversation that set the original priority.
A ten-minute check
Ask each person who reports to you two questions:
What are the two things you think I most need from you this month?
How will you know you have done them well?
Write down what they say, in their words, without correcting them. Then compare it with the list you would have written for them.
Where the lists match, you have a shared expectation. Where they differ, you have found the gap, and closing it is the manager's job. Use new words when you explain it again: name the result, show an example of good work, and agree on when you will check.
The same discipline applies to AI
I teach the same habit in AI enablement work, where it goes by the name output definition. A tool given a vague instruction can return something plausible and wrong, for the same reason a person can. Describing the result you want, and what good looks like, does most of the work in both cases. More on how I approach that is on the AI enablement section of my site, and I wrote earlier about what poor communication costs leaders.
When did you last ask someone on your team to tell you, in their own words, what you expect of them?
Sources: Gallup, U.S. Employee Engagement Below 30% for First Time in 15 Years, 28 October 2026.
