Feedback at the end of a seminar has two audiences, and they want different things. You want to know what did not work. The client wants to know whether the investment paid off. As long as both go into the same form, the second one wins — and you never get to hear what you would need to know.
A story that stands for many
A colleague told it to me. In a workshop, a manager disagrees with him on a professional question — not a row, a matter of substance on which the two take different views. He sticks to his position.
At the end of the day the manager fills in the form. The rating is poor.
Two months later an email arrives: “You were right.”
By then the feedback had long been filed with the client.
The story is instructive because nobody in it did anything wrong. The trainer was professionally correct. The manager honestly wrote down what she felt at that moment. The client took the numbers he was given. And still, at the end, a verdict is in circulation that was demonstrably wrong — and that could no longer be corrected.
Why one voice is enough
This is not an isolated case, it is arithmetic.
With ten participants, one very critical person shifts the average by several tenths. With a return of four forms — and that happens more often than reports suggest — that person decides it.
On top of that comes a psychological amplifier. Baumeister and colleagues showed in 2001 why negative information systematically weighs more in an overall judgement than positive: it is processed more intensively and sticks around longer. That applies to you when you read the forms — and to the person in procurement skimming the report.
And within the group, a critical voice often has an effect before the questionnaire even appears. Felps, Mitchell and Byington described in 2006 how a single negative member can damage the functioning of a whole group out of all proportion — the others become defensive, withdraw, and the mood tips without anyone noticing.
False precision
A client I know has been producing feedback reports on his trainings for years: twenty-five pages, results to two decimal places, trainers compared side by side. Anyone who does not score well enough is weeded out.
Two decimal places on a base of sometimes just four questionnaires.
That is not an analysis, it is a formatting choice. A number like 4.17 looks like measurement and is rounding of noise. And it is not only the sample that is too small — it is the question that is wrong.
The research on training evaluation is unambiguous here: a meta-analysis by Sitzmann and colleagues (2008) shows that participant reactions correlate only moderately with immediate knowledge gain — and practically not at all with delayed knowledge. Blume and colleagues (2010) found that affective reactions do not correlate with transfer, that is, with whether what was learned is later applied.
What such reports compare to two decimal places does not measure what it is used for. It measures how pleasant a day felt. That is legitimate information — it is just not a performance measure, and certainly not one by which to weed people out.
The built-in perverse incentive
When ratings decide follow-up assignments, an incentive arises that works against the purpose: a day that pleases is professionally safer than a day that works.
The uncomfortable exercise gets cut. The objection gets phrased more softly. The manager who would have needed a critical response does not get it. Nobody decides this consciously — it happens over years, in small steps, and at the end there is a seminar that is consistently well rated and changes nothing any more.
Optimise for recommendation and you optimise for likeability. In professional development that is a dangerous target.
Five things you can do
The problem is structural and cannot be dissolved. But it can be defused considerably.
1. Feedback in the middle, so nothing surprises you at the end
Collect the group's goals and expectations at the start, visibly on the flipchart. Ask at least once in the middle of the day whether you are getting closer to those goals and what is still missing. Then you can adjust while it still helps — and at the end nobody can say something was missing without having had the chance to say so.
2. Supply the interpretation with the numbers
Whoever hands over only numbers leaves the interpretation to the spreadsheet. Alongside the score, supply two quotes from the open fields and the most frequently named obstacle to implementation. The obstacle is the most valuable part of the whole feedback — and it is the natural occasion for a follow-up conversation.
3. Write down the response rate
A number without a base is not a number. “4.17 (n = 4 of 12)” reads differently from “4.17”. Write it alongside, every time, even unasked. It is the one detail that exposes false precision by itself.
4. Agree beforehand what is being measured
Before the assignment, not after: which questions are asked? From what response rate does a result count? Is satisfaction distinguished from applicability? A client who compares trainers has usually never thought about it — and is often grateful for the question.
5. Take the professional disagreement out of the room
Stick to your position — but not as a contest in front of an audience. One sentence is enough: “We see that differently, and that is fine. I will send you afterwards what I am basing this on.” Then continue during the break. That keeps a question of substance a question of substance — instead of turning it into a question of status in front of the group, which somebody settles in the questionnaire that evening.
And staying honest
None of this solves the underlying problem. As long as the same survey serves both your improvement and the decision about your next assignment, it will do both badly.
What helps is separation: a form that goes to the client, and responses that are only for you — the spoken round at the end, a conversation with two participants in the break, the question about the obstacle to implementation. These sources are more honest because nothing hangs on them. How to run the round so it produces more than politeness is in The feedback round in a workshop. And which questions belong in the form is in Feedback forms for seminars.
In summary
- With a small response rate, one critical voice decides the result
- Two decimal places on four forms are not a measurement but a formatting choice
- What is compared there demonstrably measures neither learning nor transfer
- Feedback in the middle prevents surprises at the end
- Always note the response rate, always supply the interpretation
- Move professional disagreement out of the plenary and into the break
