A manufacturing business ran a communication skills day for forty supervisors on a Saturday. Decent trainer, good room, lunch that people mentioned afterwards. The feedback scores averaged well above four out of five. Six weeks later the plant head asked what had changed on the floor, and nobody could point to anything.
The reflex was to blame the trainer, then the content, then the supervisors. All three were fine. The day was the only part of the intervention that existed.
What the feedback form actually measured
The form asked whether the session was useful, whether the trainer explained clearly, and whether people would recommend it. Every one of those is a question about the experience of the day. None is a question about behaviour.
This matters because a high score on those questions is achievable by a good speaker with no lasting effect whatsoever, and a genuinely difficult session that changes how someone runs a shift handover can score badly on the day because it was uncomfortable. Businesses that optimise for the form get more of the first kind.
1 REACTION Did they like it?
Cheap. Measured on the day. Tells you about the
trainer and the logistics, not about the outcome.
2 LEARNING Can they do it in the room?
A short exercise at the end. Still cheap. Tells you
the content landed.
3 BEHAVIOUR Are they doing it six weeks later?
Requires someone to look. This is where nearly all
training evaluation stops before it starts.
4 RESULT Did the thing you cared about move?
Rework, escalations, attrition on that shift, repeat
complaints. Hard to attribute cleanly. Worth trying.Nobody needs to measure all four for every session. But a business that only ever measures the first has no basis for deciding whether to run the programme again, and usually decides on whether the feedback scores were good — which is a decision about catering.
The two weeks after are the intervention
New behaviour is fragile. A supervisor who learnt a different way to open a difficult conversation on Saturday returns on Monday to the same pressure, the same shift, and colleagues who were not there. The default is overwhelmingly strong and the new behaviour has one attempt to survive it.
Almost everything that makes training stick happens in this window, and almost none of it is the trainer's job.
- 1
One thing, not twelve
Every participant leaves having written down one specific behaviour they will do differently, in their own words, with a date. A person attempting one change has a reasonable chance; a person holding a list of twelve does none of them.
- 2
The manager knows what it was
Their line manager sees that one commitment and asks about it within a fortnight. This single step does more for transfer than doubling the length of the training day, and it costs one conversation.
- 3
A visible first opportunity
Point at the next real occasion to use it — the Thursday handover, the next customer escalation. Skills that wait a month for an opportunity have already decayed.
- 4
A short return
Ninety minutes, four to six weeks later, in which people describe what they tried and what failed. The failures are the content. This is the session businesses cut for budget reasons and it is the one carrying most of the value.
The manager problem
The most reliable predictor of whether training changes anything is whether the participant's manager was involved before it happened. Not attending — involved.
In the manufacturing example nobody had told the shift managers what the supervisors were being taught. So when a supervisor tried a slower, more careful conversation on Monday, their manager saw someone taking too long over something that used to take a minute and told them to get on with it. The training was undone by a person who was never told it existed.
- Brief managers before, in fifteen minutes. What the session covers, what they should expect to see differently, and what to say when they see it.
- Tell them what it will look like when it is going well — often slower, more awkward, visibly less fluent than the old habit. Without that warning, early attempts read as incompetence.
- Ask them what they need it to fix. Managers who helped define the problem defend the solution. Managers who were informed afterwards treat it as HR's programme.
- Do not train the team without training the manager where the behaviour depends on the manager's response. Feedback skills, difficult conversations and escalation all fall into this category.
Saturday was part of the problem
Worth saying plainly, because it is so common. Training scheduled on a rest day carries a message before the trainer opens their mouth: this matters, but not enough to give up production time for it.
People attend, because attendance is not really optional. They are also tired, resentful about the weekend, and conscious that the business valued their Saturday at zero. None of that shows up on the feedback form, which they complete politely in front of the person who taught them.
Where operations genuinely cannot spare a weekday, run two half-days instead of one full Saturday, and compensate the time. The cost is real and small, and it buys a room that believes the topic matters — which is the precondition for everything else in this article.
Design backwards
Most training is commissioned forwards: somebody identifies a topic, a provider proposes a day on that topic, and the day is delivered. The question that is skipped is what specifically would be different afterwards.
Designing backwards means starting from an observable behaviour — supervisors raise quality concerns to the shift manager the same day rather than at the weekly meeting — and working back to what has to be true for that to happen. Sometimes the answer is a skill, and training is right. Often the answer is that the escalation route is unclear, or that raising concerns has historically gone badly for the person raising them, and no amount of communication training will touch it.
That diagnosis is the first half of training consulting work, and it regularly concludes that the requested programme is not the answer. A business that has run three communication courses in two years usually has a structural problem being treated as a skills problem.
What to do with the Saturday you already ran
If the session has happened and nothing changed, the value is not entirely lost — but recovering it requires acting within weeks rather than commissioning a replacement.
Go back to the participants. Ask what they tried, what worked once, and what they abandoned. Ask their managers what they noticed. That conversation costs an afternoon and tells you whether the content was wrong, the follow-through was missing, or the behaviour was blocked by something else — three very different problems with three different fixes, and no way to tell them apart without asking.
Then fix the surrounding structure before buying another day. A second session into the same conditions produces the same result, with a smaller budget and a more cynical room. Where the pattern has repeated more than twice, the honest next step is not another programme but an HR process audit of how training gets commissioned — and a serious conversation about what training consulting should be asked to diagnose before anything is scheduled. The broader question of which programmes a business genuinely needs is worth settling once rather than annually, which is what a structured soft skills and management training plan is for.
