# Formative Assessment in Training: Observe, Adapt, Check

Use formative assessment in training: collect task evidence, interpret it carefully, adapt instruction, and check again.

- Canonical URL: https://www.harmate.com/en/blog/formative-assessment-in-training-observe-adapt-check
- Author: Harmate Team
- Published: 2026-10-04
- Updated: 2026-10-04T00:20:30.555310+00:00
- Language: en

## Content

# Formative Assessment in Training: Observe, Adapt, Check

Formative assessment is not just an end-of-session quiz or a question such as “Did that make sense?” It happens during learning: gather evidence related to a task, interpret it cautiously, choose a teaching adjustment, and then check what happens next. The trainer owns that loop. A single response cannot grade a person, certify mastery, or explain a difficulty on its own.

France’s official education terminology defines formative assessment as taking place during learning or training and helping learners recognize what they have acquired, where they are struggling, and how they might progress ([Bulletin officiel](https://www.education.gouv.fr/bo/2007/33/CTNX0710380K.htm)). A Ministry guide on vocational training likewise describes assessment as part of the learning sequence, repeated over time, and grounded in dialogue rather than a one-way verdict ([technical guide 14](https://www.education.gouv.fr/media/195499/download)). For a trainer, the practical question is therefore not how many checks to add, but whether each check can inform a real decision.

## Start with the teaching decision

Before choosing a questionnaire or activity, write down what you might change. For example: “If several participants confuse the steps for preparing a meeting, I will revisit the distinction with a more concrete case at the start of the next session.” This decision is limited, situated in time, and open to revision. “Measure participants’ level,” by contrast, is too vague to guide the choice of evidence.

Next, name a learning task and an observable sign that would help you decide. If the aim is to prepare a usable agenda, ask participants to draft one for a realistic fictional situation. Share a few criteria before the exercise: an explicit meeting purpose, a sensible order of topics, time for a decision, and a follow-up item. That short piece of work is closer to the objective than a general statement of confidence.

Keep the check proportionate: a brief exercise, an explanation in the learner’s own words, a demonstration, an argued choice, or an open question may fit depending on the learning goal. A closed response (for example, “Does the agenda state the intended decision? Yes / No”) can support a quick check or simple comparison; an open response can reveal reasoning and unexpected contexts, but it requires interpretation. Neither format alone guarantees that you are measuring what you intended. For a question designed around a decision, see [training placement questionnaires](https://www.harmate.com/en/blog/training-placement-questionnaire-how-to-keep-it-short-without-losing-quality); for the difference between immediate feedback and later workplace application, see [post-training evaluation](https://www.harmate.com/en/blog/post-training-evaluation-questions-and-method-for-assessing-workplace-application).

## Gather evidence about a task, not a score about a person

A useful check describes what the participant tried, in what context, and what helped or got in the way. If participants then test their agenda in a short meeting simulation, you might ask: “What decision did the group need to make? What did you put on the agenda? What made it easier or harder to plan the follow-up?” The purpose is not to elicit a complete account or demand a confession of error. It is to make the relationship between the task, the conditions, and the reasoning visible.

Make it possible to respond without unnecessary exposure. An individual conversation, an anonymized draft, a pair discussion, or a short written response may be preferable to speaking in front of the whole group. The choice depends on the subject, the setting, and the people involved. Anonymity is not automatically desirable and does not make responses representative. Explain how responses will be used, and do not collect personal details that do not help with learning.

## Separate the trace, your interpretation, and what remains to check

A response is situated evidence, not a ready-made explanation. Imagine a participant writes: “I couldn’t cover follow-up because the meeting ran over.” The trace is that sentence. One possible interpretation is that time or the priority given to follow-up mattered. Other explanations remain possible: an unexpected issue, a postponed decision, an unclear role, or no agreement about next steps.

Keep the distinction visible in three columns:

| What was observed or said | Provisional reading | What still needs checking |
|---|---|---|
| “I couldn’t cover follow-up because the meeting ran over.” | Time or the priority assigned to follow-up may have mattered. | Was follow-up on the agenda? Did an urgent issue change the meeting? |
| The fictional agenda did not state an intended decision. | The task may have been understood as sharing information. | Were the instruction and starting example clear enough? |

Avoid turning a phrase into an individual diagnosis: “disorganized,” “unmotivated,” or “not capable” does not follow from one comment. Look for converging evidence, but keep accounts that do not fit. If someone succeeded in another context, that difference may point to an important condition rather than an exception to discard.

## Choose a limited, explicit adjustment

After reviewing the evidence, ask three questions: does the difficulty seem connected to the content, the instruction, practice time, or transfer context? What small change could be tried? What later observation might challenge the interpretation?

In the example, the trainer might spend ten minutes comparing two agendas—one organized around information, the other around a decision—and then ask participants to revise a third. The trainer does not conclude that the whole group had the same difficulty. They choose a short review because several clues converge and the exercise can still help participants who did not encounter the issue.

Before settling on an adjustment, examine the activity design as well: an apparent difficulty may depend on conditions that make it possible, or difficult, for a learner to express themselves ([technical guide 14](https://www.education.gouv.fr/media/195499/download)).

## Check again without mistaking progress for proof of impact

In the next session, use a similar but not identical situation. Ask learners to prepare an agenda for another meeting and compare their work with the announced criteria. If follow-up is now explicit, the work meets that criterion in this situation. It does not prove that the training caused lasting improvement, that each person has mastered the method, or that they will use it at work.

To discuss transfer, you need evidence from the work context and from a later point in time. That is the separate topic of [post-training evaluation](https://www.harmate.com/en/blog/post-training-evaluation-questions-and-method-for-assessing-workplace-application). For the formative loop, record the task checked, the feedback received, the adjustment tried, and the next evidence to look for. A short note is enough if it explains why you changed the session and what might make you revise the decision.

## Keep assessment in the service of learning

Formative assessment can use a quiz, case, conversation, demonstration, or questionnaire, but no tool is the method by itself. The loop is: explicit task, relevant trace, provisional interpretation, human adjustment, new check. If no adjustment is possible, ask whether collecting the response adds anything.

A questionnaire can gather responses during training, and a tool can help organize feedback, surface themes alongside source wording, and prepare a report. That does not replace the trainer: they must put the evidence in context, examine counterexamples, and choose what happens next. [Open-ended responses](https://www.harmate.com/en/blog/open-ended-questions-get-actionable-data-without-bias) can provide rich explanations, but do not by themselves prove mastery or representativeness.

### Sources

- [Education terminology — France’s Ministry of National Education](https://www.education.gouv.fr/bo/2007/33/CTNX0710380K.htm): definitions of diagnostic, formative, and summative assessment.
- [Examples of formative assessment in vocational training — Ministry of National Education](https://www.education.gouv.fr/media/195499/download): practices, dialogue, repeated checks, and adjustments in training.