Published 9 September 2026 · Updated 10 October 2026 · By Marie Castelli, Partner, Mentivis
Measuring skills acquisition: beyond the completion rate
Measuring skills acquisition means checking what a person can actually do after training, against criteria set in advance, rather than recording that they finished a module. The completion rate measures exposure. Acquisition is measured through criterion-referenced assessments spread across the learning path, repeated after a delay and carried out in realistic situations, which corresponds to levels 2 and 3 of the Kirkpatrick model.
Pillar: Skills measurement: methods, tools and metrics
Why is the completion rate not enough?
The completion rate shows that a learning path was followed to the end. It says nothing about what was understood, what will be remembered in a month, or what will be applied on the job. It became the standard because it is automatic, free to produce and easy to consolidate, not because it reveals the effect of training.
An employee can finish a 45-minute module on data protection and pass the final quiz, then not know what to say two weeks later when a customer asks for access to their data. Completion and the immediate quiz measure short-term recognition. The useful question is different: is the skill available when the situation arises?
What does the forgetting curve really say?
Hermann Ebbinghaus's forgetting curve (1885) is often summarised with figures such as "50% forgotten in 24 hours, 80% in a week". That summary is inaccurate. Ebbinghaus memorised lists of meaningless syllables himself and measured relearning "savings", that is, the time saved when relearning a list compared with the first learning. This is not a percentage of content retained, and the material has nothing in common with meaningful professional content.
In 2015, Jaap Murre and Joeri Dros replicated the experiment with the same method, on a single subject, and published their results in PLOS ONE. The resulting curve closely resembles Ebbinghaus's. Relearning savings were 47% after 20 minutes, 32% after one day and 17% after six days for the replication subject (compared with 58%, 34% and 25% for Ebbinghaus). The authors also observe a possible upturn at the 24-hour point, which they present as a trend still to be confirmed, possibly linked to sleep.
What training can take from this is modest and solid: without reactivation, forgetting is fast in the first hours and then slows down. The precise figures do not transfer to a professional skill.
What research says about retrieval and spacing
Two findings are more directly useful. Roediger and Karpicke (2006, Psychological Science) showed that students who tested themselves on a text retained the content much better at two days and one week than those who reread it, whereas rereading did better at five minutes and increased participants' confidence. Testing is therefore not only measuring: it is also learning.
The meta-analysis by Cepeda and colleagues (2006, Psychological Bulletin), covering 317 experiments, shows that spacing sessions improves retention, and that the optimal interval between sessions increases with how long the information needs to be retained.
What to measure, according to the Kirkpatrick model?
The Kirkpatrick model distinguishes four levels: reaction (level 1), learning (level 2), behaviour at work (level 3) and organisational results (level 4). Measuring acquisition sits at levels 2 and 3.
| Level | What is measured | Example tool |
|---|---|---|
| 2. Learning | Knowledge, skills, confidence acquired | Criterion-referenced pre-test and post-test, end-of-path work simulation |
| 2 (delayed) | Retention | Same assessment repeated 2 to 8 weeks later |
| 3. Behaviour | Application on the job | Manager observation, observation grid, review of real examples |
What is criterion-referenced assessment?
Criterion-referenced assessment compares a person's performance with criteria set in advance, not with other participants. It answers the question "has the skill been reached at the required level?" rather than "who got the best score?". This is the logic of the assessment framework of French vocational certifications, which, under Article L6113-1 of the French Labour Code, defines the criteria and methods for assessing learning outcomes.
In practice, each target skill gets one or two observable criteria (for example: "handles a data access request within the legal deadline, identifying the information to be provided"). An overall module score does not tell you which skill has been acquired; a criterion-referenced assessment does.
What does reliable measurement of acquisition require?
- Objectives expressed as observable skills, linked to the job's competency framework.
- A measured starting point, to calculate progress rather than an isolated final level.
- Assessments spread across the path, which require retrieving information and also serve as practice.
- A delayed assessment, a few weeks after the end, to distinguish retention from immediate memorisation.
- A check in the work situation, by the manager or through a realistic simulation, for level 3.
- A result per skill, usable in the skills map and in L&D metrics.
What this changes for the organisation and the employee
With acquisition data, L&D can identify the skills that resist, compare programmes and rework those that have no effect. This data is also the starting point for a credible training ROI calculation.
For the employee, a record of acquired skills is a concrete basis for the professional review (a mandatory career interview in France) and internal mobility. These individual results are personal data: their purpose, recipients and retention period must be defined in line with the GDPR (see GDPR and training).
Frequently asked questions
Sources
- Replication and Analysis of Ebbinghaus' Forgetting Curve, Murre J. M. J. et Dros J., PLOS ONE, 2015, DOI 10.1371/journal.pone.0120644
- Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention, Roediger H. L. et Karpicke J. D., Psychological Science, 2006, DOI 10.1111/j.1467-9280.2006.01693.x
- Distributed practice in verbal recall tasks: A review and quantitative synthesis, Cepeda N. J. et al., Psychological Bulletin, 2006, DOI 10.1037/0033-2909.132.3.354
- What is The Kirkpatrick Model?, Kirkpatrick Partners
- Article L6113-1 du Code du travail (référentiel d'évaluation), Code du travail numérique, ministère du Travail