Two cohorts start the same onboarding programme on the same Monday. The first cohort is made up of people from different backgrounds, and a quick check shows that half of them are missing the basic procedure vocabulary the course assumes. The second cohort knows the vocabulary but freezes in the practice cases, where the right action depends on a judgement call. One timetable of three live workshops serves neither group well.
This is an illustrative scenario, not a documented customer case. It is a common design problem in blended learning: the delivery format is chosen before anyone has asked what the task, the learners and the constraints need.
This guide is for instructional designers and training managers who have to choose between adaptive, interactive and live formats, and between synchronous and asynchronous learning, for a given task. It gives you five questions to ask, a decision matrix, and a worked example you can copy.
Quick facts
| Question | Short answer |
|---|---|
| What decides the format? | The job task, what feedback it needs, the learners' starting points and your delivery constraints. Not the tool. |
| When is adaptive learning a fit? | When learners start with different prerequisite knowledge and you can define and check it. |
| When is interactive practice a fit? | When learners must apply or decide, and benefit from committing to an answer and getting feedback. |
| When does live support add value? | When a choice is ambiguous, risky, or needs specialist feedback no answer key can give. |
| Can higher-order work be asynchronous? | Yes, with the right design. Live is not the only place for judgement. |
1. Clarify the target task and learner needs
Before choosing a format, write down three things.
- The task. What must the person do at work? Write it as something you could watch, for example "complete a patient record update" or "handle a guest complaint within agreed limits".
- Learner starting points. What do learners know about the prerequisites? A short diagnostic tells you more than a job title does.
- Delivery constraints. Time, location, shift patterns, access to specialists, and whether people can be in the same room.
Then decide what the task needs from the learning: how much practice, and what kind of feedback. Our guide to matching learning activities to Bloom's Taxonomy shows how to describe the thinking a task demands, which is what the format has to support. Carnegie Mellon's Eberly Center makes the underlying point: objectives, assessments and instructional activities should reinforce one another, and misalignment between them hurts learning (Eberly Center, alignment of objectives, assessments and instruction).
2. Compare feedback and practice needs
Practice is only as useful as the feedback that follows it. There are three main sources, and they suit different situations.
| Feedback source | Best when | Watch out for |
|---|---|---|
| Automated (immediate, from a pre-written key or rule) | The answer is clear, the risk of a wrong answer is low, and learners need fast repetition | It cannot judge reasoning it was not built to recognise |
| Instructor or specialist | The choice is ambiguous, the stakes are high, or the reasoning matters more than the answer | It takes scarce time, so reserve it for where it adds most |
| Peer | Learners can compare approaches and explain their choices to each other | It needs clear criteria, or peers reinforce each other's misconceptions |
Ask three questions of each practice activity. How ambiguous is the right answer? How risky is a wrong one? Is there a credible answer key at all? If the answer is clear, low-risk and keyable, automated feedback may be enough. If it is ambiguous, risky or has no key, plan for human feedback somewhere in the journey.
3. Select adaptive or interactive experiences
These two terms are often blurred, so separate them.
Adaptive learning changes what a learner gets based on how they perform. In this guide it means targeted preparation: a short diagnostic, then different support for learners who start with different gaps. It suits prerequisite knowledge, where a defined concept either is or is not in place. It does not need generative AI, and a simple rule such as "if you miss the procedure vocabulary check, you get this refresher before the practice" is adaptive in the sense that matters here.
Interactive learning asks learners to respond: choose, rank, explain, vote, commit to a decision. It suits application, where the aim is to use a procedure or make a judgement in a new case. When each person commits before the group discusses, you see where understanding diverges, which is what the debrief needs.
Match them to the cohorts in the opening scenario. The first cohort, with prerequisite gaps, needs adaptive preparation so people reach the same starting line. The second needs interactive decision practice with feedback on reasoning.
AhaSlides supports adaptive learning, interactive learning and moving between live and self-paced delivery. Check any specific diagnostic, authoring, scoring or reporting function you plan to rely on against your own workflow, and treat AI, simulation and integration features as external or unconfirmed until you have seen them work.
4. Decide when live support adds value
Synchronous and asynchronous learning are both legitimate. A useful rule: use live sessions where the interaction or specialist feedback is the point, and asynchronous delivery where learners can work alone at their own pace.
Live adds the most value when:
- A choice is ambiguous. Two defensible actions exist, and hearing why people chose differently is the learning.
- The reasoning needs a specialist. Someone who knows the work can explain why an answer is defensible or incomplete.
- Risk or emotion is high. Practising a difficult conversation with a person is different from reading about one.
Asynchronous is often enough when:
- The content is rules and concepts. Self-paced study with a short recall check does the job.
- The case is clear-cut. A learner can submit a response and receive written feedback against agreed criteria.
- Schedules do not line up. Shift workers and distributed teams cannot all attend at once.
Higher-order work can also happen asynchronously. A learner can analyse a case, write a justified response and receive feedback on their reasoning without a meeting. Live suits ambiguity. Thinking itself does not need it. For the definition of the terms themselves, see our explainer on what asynchronous class means.
The research on blended formats is a useful caution here. A widely cited US Department of Education meta-analysis of online learning studies found that blended conditions showed larger differences from face-to-face teaching than fully online ones, and its authors noted that blended conditions often included extra learning time and instructional elements the comparison groups did not receive, so the benefit should not be put down to the medium itself (Means and colleagues, 2010, via ERIC). The studies cover 1996 to 2008 and come mostly from higher education and medical training, not the workplace. Treat them as a reason to design the learning carefully, not as proof that blending any set of formats will work.
5. Combine formats in one coherent journey
A decision matrix helps you pick per stage. Fill it in for your own task.
| Need at this stage | Format | Why |
|---|---|---|
| Learners start with different prerequisite knowledge | Adaptive, self-paced preparation | Targets each learner's gap and leaves live time for discussion |
| Learners must apply a procedure to a new case | Interactive practice, self-paced or live | Each person commits to a decision and gets feedback |
| The right action is ambiguous or high-risk | Live facilitated discussion | Specialist feedback and comparison of reasoning |
| Rules and concepts need confirming | Self-paced study with a recall check | Cheap, repeatable, confirms prerequisites only |
| Learners must show they can do the task | New case or supervised work task | Checks transfer, not memory of the training case |
Then connect the stages with clear handoffs. For example:
- Prepare (adaptive, self-paced). A short diagnostic routes each learner to the refresher they need. Everyone who passes the check enters practice.
- Practise (interactive). Learners work through realistic cases, commit to a decision and a reason, and get feedback.
- Debrief (live). The facilitator takes the cases where answers diverged and asks why.
- Show it (new case or supervised task). Learners tackle a case they have not seen, or a real task with a reviewer.
Each handoff has a trigger: pass the check to move to practice, flag the ambiguous cases for the live session, and book a coaching slot for anyone whose reasoning was weak.
A worked example: choosing formats for a complaint-handling programme
Illustrative. A practitioner should validate the case and the answer criteria before use.
The case. A hospitality group is training front-of-house staff to resolve guest complaints within defined authority. A quick diagnostic shows two groups: new starters who do not know the authority limits, and experienced staff who know them but escalate everything.
The task for the designer. Decide the format for each stage, cite the evidence for the choice, and say what further information is needed.
A weak response. "Run a half-day workshop for everyone." It cites nothing and treats the two groups the same.
A stronger response. "The diagnostic shows a prerequisite gap in group one, so they get self-paced preparation on the authority limits with a short check before practice. Group two does not need that. Both groups then do interactive case practice where each person commits to a response and reason, because the task is to apply a policy. I would run a live debrief only on the cases where answers diverged, because those are ambiguous and the reasoning matters. I would need to know whether supervisors can attend and whether the authority limits are final, because unclear limits are a conditions problem that no format fixes."
The retry. The designer gets a second programme, this time for a compliance topic with clear answers, low risk and a dispersed shift pattern, and chooses a mostly asynchronous design with one optional live Q&A.
A rubric you can adapt
Score each criterion 0 to 2. This is a proposal to adapt with your own specialists, not a validated readiness standard.
| Criterion | Observable evidence |
|---|---|
| Task execution | Chooses a format per stage based on the task, the learners' starting points and the constraints. Records the choice and the evidence. |
| Reasoning | Explains the alternatives considered, the feedback each stage needs and any uncertainty. |
| Boundaries | Uses approved procedures, confirms what the platform can do, and asks for help when information or authority is missing. |
Anchors: 0 = absent or unsupported. 1 = partial, with relevant omissions. 2 = complete and justified against agreed criteria. Define critical errors separately, for example assigning an ambiguous, high-risk task to purely automated feedback. A total score must not hide a critical failure.
What evidence would show it is working
The US Centers for Disease Control and Prevention separates evaluating learning from evaluating how well learners apply it at work, and recommends planning for both (CDC, evaluate training: measuring effectiveness). It is measurement guidance, not a forecast of results.
For a format decision, that means:
- Learning: compare performance on a new, unseen case with a baseline taken before the programme.
- Transfer: review a sample of real work, or a supervised task, with the same rubric, after learners have had time to use the skill.
- Fit: check whether the handoff triggers worked. Did the diagnostic route people correctly? Did the live session spend its time on the ambiguous cases?
Define who is included, the denominator and the timing. If outcomes move, check what else changed in the same period before crediting the format choice.
Constraints that may remain
Choosing the right format does not remove every constraint:
- Scarce specialists. A live debrief needs someone who can give credible feedback, and they may not be available often.
- Time and shifts. Learners who cannot attend live sessions need an asynchronous route that does not feel second-class.
- Unclear policy. If the authority limits or procedures are vague, no format can teach them clearly.
- Tool limits. The platform may not support a diagnostic, a feedback type or a report you assumed. Check before you design around it.
Plan separate fixes for these, and do not promise business results from the format choice alone.
Frequently asked questions
When should I use synchronous learning instead of asynchronous learning?
Use synchronous sessions when the choice is ambiguous, the stakes are high or a specialist's feedback is the point. Use asynchronous delivery for rules, concepts, clear-cut cases and schedules that do not line up.
What is blended learning design?
It is the process of combining formats, such as self-paced study, interactive practice and live sessions, into one journey, with each stage chosen for what the task and the learners need and with clear handoffs between stages.
Is adaptive learning the same as AI?
No. Adaptive learning means the experience changes with how the learner performs. A diagnostic that routes learners to different refreshers is adaptive, with no generative AI involved.
Where to start
Take one programme you run now. Write the task as an observable action, run a five-question diagnostic on the prerequisites, and fill in the decision matrix above for each stage. Mark any stage where the answer is ambiguous or risky: that is where live time is best spent.
To see how adaptive preparation, interactive practice and live debriefs could fit together for your own programme, explore AhaSlides.
