One of the strongest predictors of whether training changes behavior on the job is the learner’s direct manager.

Research on training transfer has pointed there for years. The environment the learner returns to is crucial, and within that environment support from the direct supervisor is associated with transfer more than twice as strongly as support from peers. A federally funded evidence review lands in the same place, listing supervisor support and coaching among the conditions most consistently associated with transfer.

That finding creates an awkward situation in most training programs, because the person with the most influence over whether training works is usually the person least involved in it.

Where the Loss Actually Happens

It’s often said that only 10% of formal training makes it into the work. Though the exact number is hard to trace, the direction is generally right and that’s why it has lasted. Most development happens on the job rather than in a classroom.

The more useful question is where the learning loss happens. Consider what typically follows a training session. The employee returns to a job where the work that was waiting is still waiting, the manager receives a completion report, and within a week or two nobody mentions the training again. Whatever that person intended to do differently now depends entirely on their own memory and initiative.

The manager will likely be blamed for this, but it’s closer to a design problem. Most programs are built to deliver content and record attendance, so they produce little a manager could observe, and they rarely ask the manager anyway. A manager can’t reinforce a behavior they never see the person attempt.

Practice Has to Happen in the Work

If the manager needs something to observe, the program has to extend into day-to-day work, which means building something into the weeks after the session. The session may be complete, but the behavior change isn’t. That takes time.

One way to support it is a series of short activities the learner completes on the job, aimed at a small number of specific behaviors. After the formal training ends, a program like that could include:

• Short activities done during real work rather than away from it, running over weeks rather than hours
• A focus on a few specific, observable behaviors tied to the business problem the program exists to solve
• Stages that get harder as the learner improves, so capability builds step by step
• A quick written record of what happened when the learner tried each activity, in their own words, from real situations
• An easy way for the manager to encourage the learner and check progress

The value for the employee is the added practice. Activities sustained over weeks give enough repetition to build a habit, and a habit is something a manager can be asked about.

PrimeFlight Aviation ran its time management program this way. Employees practiced during daily operations and wrote down what happened each time, and managers reported people handling the same workloads with less strain. More than 60% of participants showed measurable improvement in the targeted behaviors.

What the Manager Can Do

The manager’s part splits into two jobs: measuring whether change happened, and confirming whether someone is ready for more.

For the first, a manager can rate a team member on the targeted behaviors before the program starts and rate the same behaviors again at the end of the quarter, while the learner rates themselves on the same items. Comparing before and after shows whether anything changed.

For the second, the learner says when they think a behavior has taken hold and describes a specific instance, and the manager checks that claim against what they’ve seen. The answer is either yes, that’s happening now, or not yet, keep working at it, with a note on what’s missing.

Whether any of this survives a working week depends on the manager’s calendar, the constraint most programs underestimate. A request that takes 10 minutes gets ignored by exactly the managers whose teams need it most, while one that takes 90 seconds and arrives with the learner’s own example attached tends to get answered.

Why One Reading Isn’t Enough

Asking the learner is reasonable. Asking only the learner is a measurement problem.

A synthesis of self-evaluation research covering about 330,000 people found that self-insight is sharper for specific skills with a clear measure than for complex ones. That finding tells you what kind of question to ask: whether someone is a good coach is nearly unanswerable, while whether they ran a particular kind of conversation last week is checkable.

The manager’s rating, taken on the same items over the same window, gives a second, independent reading of the same behavior. That is what turns a self-report into evidence, and it changes what you can tell an executive: what managers observed in their own teams, alongside what people said about the program.

The most valuable output, though, is the case where the two readings disagree. A learner who reports a behavior their manager hasn’t seen is either not transferring it yet or working in a way their manager isn’t positioned to notice. Both are real problems, and a completion dashboard hides both. When the same gap turns up across a whole team, the finding usually isn’t about the learners at all. It points to a manager who needs support more than their people do.

Where to Start

Manager involvement is one of the strongest levers available, and most programs leave it unused because they hand the manager nothing to act on. Build the practice into the work and the manager’s role becomes a small, specific ask rather than a vague expectation. That is the difference between hoping a program worked and being able to show that it did.

So, the question worth taking back to your programs isn’t whether your managers are engaged enough. It’s whether anything in the design gives them something to engage with. If the honest answer is a completion report, that is where to start.