Passing the quiz on the day of training says little about what people can do six weeks later. Retention does not happen at the end of a course. It is designed in afterwards, with later practice that is spread out and checked. This guide is for heads of L&D, instructional designers and training managers. It covers five decisions: what must be retained, which retrieval and performance tasks to use, how to schedule them, how to give feedback and how to evaluate delayed retention and transfer. It also separates two things that are easy to confuse: keeping facts in memory, and keeping a whole task skill in working order. The case and rubric are illustrations: ask the manager of the role to review them first.
What the research says, and does not say
Two findings from the memory research are worth knowing. In a 2006 study by Roediger and Karpicke, students who were tested on a passage instead of re-reading it recalled more of it a week later, even though re-reading felt more effective in the short term. A 2006 review by Cepeda and colleagues of distributed practice studies found that spreading practice over time improves later recall, and that the best gap depends on how long the material must be kept.
Both are lab studies, mostly with students and mostly of recall. They support retrieval and spacing as sound design choices. They do not tell you the right interval for your workforce, and they say little about complex job tasks. Treat them as a reason to test your own schedule, not as a formula.
Quick facts
| Question | Short answer |
|---|---|
| What do you check? | Recall for facts that must be known, and fresh task cases for skills. |
| What sets the schedule? | How often the task is used, how risky an error is and what the learner's evidence shows. |
| What gives feedback? | Explanations after each retrieval, with extra practice aimed at observed gaps. |
| How do you test retention? | Delayed recall and delayed task performance, assessed separately. |
| Is there one right interval? | No. Do not copy a universal schedule. |
| What can it not fix? | A task nobody does on the job, and a process that has changed. |
1. Identify what must be retained
Separate what must be recalled from what must be performed. A dosage limit, a policy threshold or a safety rule is a fact that should come to mind quickly. Handling a difficult customer or diagnosing a fault is a skill that needs performance, not recall. Then ask how often the employee meets the task. Something done daily maintains itself. Something done twice a year, such as an emergency step, decays unless you rehearse it.
2. Select retrieval and performance tasks
Use retrieval for knowledge: short questions that ask people to produce the answer, not recognise it. Use fresh task cases for application: a new situation, not the one from training. The two give different evidence. A learner who recalls a rule has shown memory. A learner who applies it to a new case has shown more. Our guide to matching learning activities to Bloom's taxonomy explains how to match the activity to the demand.
3. Schedule practice around risk and use
Choose intervals based on task use, risk and learner evidence. A rarely used, high-risk task needs more rehearsal than a daily, low-risk one. Learners who answer correctly and quickly can be asked less often. Learners who struggle can be asked sooner. Do not present one schedule as suitable for every workforce: write down your starting intervals as a hypothesis, then adjust them with the delayed results.
A starting sketch: a short check a few days after training, another a few weeks later, then one at the point the task is likely to arise, such as before a seasonal peak. Treat these as placeholders to test, not recommended numbers.
4. Give feedback and adjust support
Explain the right answer, not only that the learner was wrong. Aim extra practice at the items missed, and escalate to an instructor or coach when the same misunderstanding persists across attempts. Persistent errors usually mean the original explanation was weak, not that the learner needs more repetition.
5. Evaluate delayed retention and transfer
Assess delayed recall and delayed task performance separately. A team can remember the rule and still misapply it. Then check whether better retention carries into real work: a manager observation, or a review of a recent task. Our guide to assessing judgement and problem solving in training covers scoring and calibration, and our guide to running a pilot that measures learning and workplace transfer shows how to test a design on a small group first.
A worked case with feedback and a retry
Illustrative. The instructional designer or training manager should validate the case and the answer criteria before use.
The case. Your team performs well on a quiz straight after a new returns-process training, then struggles with a returns case six weeks later.
The task. Plan retrieval and delayed performance checks. Explain your choices, cite the evidence you used and say what you still need to know.
A first attempt. "Send the same quiz again every Friday for the next ten weeks." The reviewer asks whether the quiz tests the same questions people already know, whether it checks handling a case or only recall, and how often anyone will meet a returns case. The learner has no answer to any of them.
Feedback and retry. The reviewer points out that repeating the same items trains people on the items, and that a quiz cannot show whether they can handle a case. The learner is asked for a new plan.
A stronger attempt. The learner separates five facts that must be recalled, such as the return window and refund approval limits, from one skill, handling an unusual return. Facts get short retrieval questions with explanations, a few days and a few weeks after training. The skill gets a new case each time, scored with a rubric. Frequency is raised for the rarely seen case types. Delayed recall and delayed case performance are reported separately, and a manager reviews two real returns at week eight. They say what they still need: how often each case type occurs and who owns the process rules if they change.
Why each response is defensible or incomplete. Judge against the approved criteria, not the reviewer's taste. The first attempt confuses repetition with retention and cannot show task performance, so it counts as a critical error. The second matches the activity to the demand, sets the schedule by use and risk, and measures both recall and performance.
Score the plan
Score each criterion from 0 to 2. Adapt it with your specialist: this rubric is a proposed starting point, not a validated readiness standard.
| Criterion | 0: absent or unsupported | 1: partial, with omissions | 2: complete and justified |
|---|---|---|---|
| Task execution | Repeats the original quiz. | Schedules checks but only for recall. | Plans retrieval for facts and fresh cases for skills. |
| Reasoning | Gives no reason for the intervals. | Gives a reason but not use or risk. | Sets intervals by use, risk and evidence, and states what is uncertain. |
| Boundaries | Claims one schedule suits everyone. | Notes limits without a plan to test them. | Treats the schedule as a hypothesis and defines how to adjust it. |
Define critical errors separately, for example reporting a recall score as proof of task competence. An overall score must never hide a critical failure.
To check the plan, compare a baseline with an unseen delayed task, and define the population, denominator and timing before you start. Measure task performance, the quality of the evidence collected and how suitable the resulting learning design proves. Look for concurrent changes, such as a process update or a new tool, before you credit the checks.
Where interactive delivery helps
Interactive formats help where people should commit before they see what others chose. In a live refresher, a facilitator can ask each team member to type their answer to a returns case, then show the spread before discussion. Self-paced delivery suits the spaced questions between sessions. AhaSlides, an all-in-one engagement platform for teams and classrooms, supports interactive learning and both live and self-paced delivery. Check how you would run the schedule against your own workflow before you rely on it.
What retention checks will not fix
- A task nobody does. Rehearsing something the job never asks for wastes time.
- A changed process. Better memory of an old rule is worse than none.
- Poor first teaching. Spacing helps people keep what they understood, not what they never grasped.
- Evidence limits. The research is mostly lab-based recall. Measure your own population.
Frequently asked questions
What is spaced learning in workplace training?
Spreading practice and checks over time, instead of delivering everything in one session, so that people keep what they learned.
How often should you run retention checks?
It depends on how often the task is used, how risky an error is and what learners show. Start with a hypothesis and adjust it with delayed results.
Is a quiz enough to show retention?
A quiz shows recall. It does not show that someone can handle a new case. Assess delayed recall and delayed performance separately.
Does spacing work for skills as well as facts?
The strongest evidence is for recall. For skills, use fresh cases and supervised observation, and test your own schedule.
Who should review the plan and rubric?
The manager of the role and an instructional designer or training manager, before anyone is scored.
Where to start
Pick one training that people complete well and apply poorly. List its facts and its skills, set a first check a few days out and a second a few weeks out, and score a fresh case at the second. Compare delayed recall with delayed performance, then adjust.
To see how a schedule like this could run live or self-paced, explore AhaSlides.
