Plenty of blended learning programmes are designed backwards. First it is settled that there will be a classroom day, because there always has been. Then someone looks for things that could happen digitally before and after it. The result is a learning path where the e-learning plays homework and the classroom day spends its first hour levelling out what people did or did not know.
The other way round works better. Every piece of content answers a few questions about itself, and the format follows from the answers. So this article is a set of forks. At each one you decide, for one piece of content, whether it belongs in the room, on screen or in a mix of the two.
The groundwork behind it, synchronous and asynchronous methods, design by timing and content, and six widely used models, is in the white paper “Blended Learning”, which you can download for free. This article is about the decision itself.
Bring one real piece of content along. Ideally the one on your desk right now.
Fork one: does someone need to be there when it goes wrong?
Some things you only learn by getting them wrong and hearing straight away why. A difficult feedback conversation, a sales call with an objection nobody saw coming, a manual step on a machine where a mistake costs money. What counts here is the reaction in the moment: the trainer sees the hesitation, hears the tone, stops the scene and runs it again. Then there is what happens between participants, the colleague’s question you would never have asked yourself, the discussion that brings a misunderstanding to the surface in the first place. A video can show all this. It cannot let you practise it.
If your answer is yes, the core of this content belongs in the room, or at least in a live video session with small groups. If it is no, because a mistake while learning costs nothing and can be repeated as often as needed, the content is a candidate for the digital part. That covers a surprising amount: product knowledge, process steps, how to use a system. Where short digital units work well and where they fall short is covered in more detail in our piece on microlearning in organisations.
Fork two: how long will the content stay correct?
Prices, product features, a changed approval rule. Content like this is out of date before the next training date has been agreed. Sales shows it most clearly: a field force spread across many regions cannot be brought together every time something changes in the product. Tie that kind of content to classroom sessions and you are systematically teaching last month’s version.
Fast-changing content belongs in short digital modules that can be swapped out one at a time. The room is kept for what changes slowly: method, attitude, handling the difficult cases.
This fork has a precondition that is often missing. The digital part needs a home where employees can find it. In the German-speaking market, Masterplan and Amadeus Fire (2025) found that 52% of companies have no digital learning platform, even though 70% consider digital topics relevant. If you take the digital branch, settle first where the content will live.
Fork three: does everyone know roughly the same?
This fork is about timing: before the session, after it, or alternating. Three routes:
- If the group shares a similar, sound foundation, prepare digitally and apply together. People who have studied beforehand bring questions rather than gaps, and the session time goes to the hard cases.
- If newcomers are involved, learn together first and go deeper digitally afterwards. Someone studying alone without a foundation gets things wrong, and the session opens with corrections. That costs time and motivation. After the session, on the other hand, participants know exactly what they want to look up.
- With a mixed group over several weeks, alternate: short classroom blocks and digital stages take turns, and theory and trying things out swap places each round.
Fork four: where are people, and when do they have time?
Shift work, field sales, sites in several countries. For audiences like these, every classroom session carries a cost that no seminar budget shows: travel, hotel nights, work left undone, and above all a date that suits every single person. The more dispersed the group, the further the design moves towards digital, with in-person time as a deliberate exception. For internationally spread teams, one split tends to work: a shared core for everyone, for instance on a product sold globally, plus digital modules per market with local prices, arguments and specifics.
Before anyone scraps the classroom day on principle, though: there is a good reason it is so hard to kill, and it rarely appears in the learning objectives. People from different sites meet, a question from day-to-day work that has been open for weeks gets settled over coffee, and afterwards you know who to call. That value is real. It is also hard to measure. If it matters to you, plan for it openly rather than disguising it as training.
Fork five: who notices when someone drops out?
Self-directed learning takes discipline. Anyone who cannot find it in a busy working week falls behind, and without support nobody notices. Learners feel left alone, questions stay open, and so do misunderstandings.
If nobody has time to accompany a digital learning path, a highly flexible model is the wrong choice, no matter how good it looks on paper. The programme then needs anchors: fixed milestones, a named contact person, a short check-in along the way. How to tell whether a programme is working rather than merely being completed is the subject of measuring learning success.
Where the forks lead
Six models have become standard in the literature. They were first described by the Innosight Institute (Horn and Staker, 2011) for schools and are now common in corporate training as well. Your answers above map onto them like this:
| If your answers look like this | the closest fit is | watch out for |
|---|---|---|
| Live feedback needed, little prior knowledge | Face-to-Face Driver: classroom at the centre, online material in support | Little individualisation, but easy to steer |
| Stable theory, practice can move out, group within reach | Rotation: fixed switch between classroom and online | Classroom dates remain the logistical hurdle |
| Dispersed, fast-changing content, independent learners | Flex: almost everything online, meetings only when needed | Without support, people drop out unnoticed |
| Equipment or software only available on site | Online Lab: digital content used in a training room | Little exchange with trainers |
| Long, part-time professional development | Self-Blend: rare classroom blocks, lots of independent work | Only works with strong self-motivation |
| Dispersed, but a familiar teaching format is wanted | Online Driver: digital teaching, in-person as the exception | Still schedule live sessions on purpose |
One admission belongs here. The models are not cleanly separable, and in practice most organisations end up with a mix of two. The learning theories often used to justify blended learning, from constructivism to behaviourism, do not help much with the actual assignment either. The five forks above are more pragmatic. Whether they lead to the best result in every case, we do not know. They do reliably lead to a decision you can explain.
What happens to last year’s slides
The digital part rarely starts from zero. The presentations, manuals and process descriptions used in classroom training so far are usually already there. What converts well into a short learning video:
- product launches and range changes
- process steps and click paths in a system
- basics for the first weeks of onboarding
- the theory ahead of a hands-on instruction
- answers to the questions trainers hear in every single run
Explained once, the content is available at any time, on mobile too, and as often as needed. For new hires, a series like this is laid out week by week in our piece on onboarding videos for the first 90 days. With an authoring tool like Omnora, an existing presentation becomes a learning video with a narrator and a quiz. From Omnora’s own experience: typically up to 80% less production time per training hour. That leaves the classroom time for whatever got stuck at the first fork.
A suggestion for your next programme: cross out the classroom day that is already in it, as an experiment. Then ask which piece of content demands it back.




