Setting up or renewing a corporate academy.
From a view on learning to the question that eventually lands on the table: which learning material suits us, and what will it cost to keep it standing. This guide walks that route and points, per subject, to the articles that go deeper.
Written for L&D managers, HSE leads and operations directors in the energy, process and manufacturing industries.
An academy is not a course catalogue.
Most organisations that say they want a corporate academy mean something a good deal more concrete. An onboarding in which new people can work independently after a few weeks rather than after six months. Expertise that does not walk out of the door with the colleague retiring in five years. Evidence that shows an auditor who is competent for what. And material that does not have to be rebuilt the moment a procedure changes.
Those are four operational problems. A course catalogue solves none of them. An academy in the sense that matters is not a collection of trainings but a production capability: the ability to make learning material, maintain it, and demonstrate that it has been followed, at a cost that survives a second and a third round.
That distinction sounds abstract until you see what happens when it is missing. Budget arrives, a package of trainings gets bought in, and eighteen months later part of that material is factually wrong. Nobody took a decision about that. It happened because every individual change was too small to make a project of.
So the question that decides whether such a programme succeeds arrives early: not which subjects go into it, but who updates the material in two years and what that costs per change. Anyone asking that only at delivery is asking too late.
Three triggers that really set it off
In practice an academy programme almost always starts through one of three events. Departure: experienced people leave and their knowledge is recorded nowhere. Growth: so many new people arrive that shadowing a colleague no longer scales. Or accountability: an audit, an incident or a client asks for evidence you cannot produce.
Which of the three it is decides where you begin. That is not a detail: with departure you start at capturing knowledge, with growth at onboarding and induction, with accountability at records and assessment. The same academy, three different first years.
Further reading on the setup
Why e-learning projects die in the concept phase
Projects rarely stall on the build. They stall the moment a steering group is handed a forty-page blueprint and feels nothing about it.
7 min read →Standardising without becoming generic
One programme for ten sites saves money and loses support. The distinction that solves it: what has to be the same everywhere and what has to differ per site.
6 min read →5 signs your safety training is not working
Green ticks, high completion rates and satisfied participants. And still the incidents keep repeating. These are the five patterns that point to training built to be ticked off rather than to be effective.
8 min read →Starting and renewing are two different tasks.
Starting from nothing
The risk is not that you build too little. It is that you build too much at once. The temptation to cover the whole curriculum in one programme is strong, and it is the most predictable way to run aground: by the time module fourteen is finished, module two no longer holds.
Start with one defined audience and one clear problem. New operators who need six months to run independently. Contractors who have to be on site on Monday. Build that out, measure whether it works, and use the source files and the working method as the basis for the next block.
What you choose in that first block stays fixed for years afterwards: the house style of the material, the file format, how you translate, how results reach your LMS. Those choices are cheap to make now and expensive to reverse later.
Renewing an existing academy
Here the first step is an inventory practically nobody has ever made: what material exists, when was it last changed, and which procedures have been amended since?
That list is uncomfortable and always clarifying. In practically every organisation, a small part of the material turns out to carry the greatest risk: precisely the modules about the operations where things can really go wrong, untouched for just as long as the rest.
Renewing rarely means remaking everything. It usually means making the part that matters maintainable, and accepting that some of the existing offering can simply go. An academy offering twenty percent less but holding up a hundred percent is worth more than the reverse.
In both cases the cost question arrives earlier than expected, and rarely in the form people expect. The build price of a module is the smallest variable. What decides the budget is how much source material is already usable, the number of languages, and whether the material can be updated internally after delivery.
Further reading on the first steps
Speeding up onboarding for technical staff
Nine to twelve months to working independently is the norm at many companies. A large part of that time does not go into learning, it goes into waiting.
7 min read →Capturing expertise before it retires
The operator who knows why the reactor behaves differently in winter will be in his garden in two years. What he knows is written down nowhere, and that is fixable while he is still here.
7 min read →Contractor site induction for 200 people in three weeks
Every turnaround starts the same way: hundreds of people who do not know the site, and one coordinator who has to induct all of them. Here is how to make that scale and stand up to an audit.
7 min read →What does it cost to have training material built?
Quotes for e-learning run from a few thousand to well into the tens of thousands for what looks like the same thing. Here is where that difference comes from, and the question you should be asking.
8 min read →Which learning material suits which question.
This is what most people come in with, and it is rarely what has to be answered first. The format follows from what somebody has to be able to do, not the other way round. A choice for VR that begins with the wish to do VR almost always ends in an expensive demonstration standing still six months later.
The usable distinction is not between technologies but between kinds of learning objective. Does somebody have to know something, be able to do something, or recognise something under pressure? Those three questions lead to different formats, and the costs differ between them by an order of magnitude.
| Learning objective | Suitable format | Watch out for |
|---|---|---|
| Transferring knowledge and rules | E-learning, short video | Cheap and scalable, but does not demonstrate that somebody can do it. |
| Holding knowledge after a training | Microlearning, spaced repetition | Works as an addition, not as a replacement for the initial training. |
| Performing an operation step by step | 3D work instruction, learning app on the floor | Stands or falls on whether a usable 3D model already exists. |
| Understanding a spatial or invisible process | 2D process animation, 3D visualisation | Explains what happens, does not rehearse operations. |
| Acting in a situation you cannot recreate | Scenario training, VR simulation | The most expensive format and the narrowest use. Defensible at real risk. |
| Building a skill with guidance | Blended: classroom plus digital | Asks a different role of your trainers, count on that. |
A programme practically never consists of one format. The question is which format carries the weight and which hang around it.
Further reading per format
When do you choose e-learning, microlearning, video or VR?
The format often gets chosen before the problem has been defined. A practical order for turning that decision around, and the question that actually makes the choice.
8 min read →Microlearning on shift work: learning between tasks
Scheduling a training day around a shift rota is a rostering problem that rarely gets solved. Short modules sidestep it, provided they are designed properly.
6 min read →Learning apps for staff without a desk
Most of the industrial workforce has no laptop and no company email. For that group the phone is not an extra channel, it is the only channel.
7 min read →From classroom to blended: what changes for your trainers
The strongest resistance to digital training rarely comes from the participants. It comes from the trainers, and it is a legitimate concern you have to take seriously.
6 min read →Adaptive learning: skipping what somebody already knows
An engineer with twenty years of experience gets the same module as somebody on day one. Adaptive material solves that, provided you have enough content for it to work with.
6 min read →QR and NFC on equipment: instruction at the place of work
The least spectacular technology in this field and often the most useful. A sticker on the machine pointing at the right instruction, provided you answer one question properly.
6 min read →Still unsure about the format? The Training Format Scan walks the trade-offs through with you in a few minutes, or look at the technology overview.
Showing work you cannot describe.
There are three situations in which a written instruction falls short on principle, however well it is written. The process happens where you cannot look, inside a pipe or below ground. It turns on spatial relationships you have to see to understand. Or the sequence is complex and a mistake in it only comes to light much later.
In those cases, imagery is not a decoration on the instruction. It is the instruction.
The forms differ by an order of magnitude in price, and the cheapest one that solves the problem is almost always the right one. A 360 walkthrough shows a location for a fraction of what a model costs. Scanning captures an existing installation no drawing still matches. A 3D model makes operations possible that imagery alone cannot rehearse. VR adds consequences to that, at the highest price and for the narrowest use.
The argument almost nobody makes
With every form that needs imagery, the real value sits in reuse. A 3D model you have built once then feeds the work instruction on the tablet, the animation in the e-learning, the image in the manual and, if it comes to that, the VR scenario. The first application carries the full cost. Every one after it is a fraction.
That makes the first choice weigh more than it looks. Whoever has a model built as a one-off part of a single instruction pays the full price for one application. Whoever fixes up front that the model is their property and gets delivered reusable is buying a foundation rather than a product. The same applies to text, scenarios and translation memories.
Further reading on imagery
Visualising work: when 3D, animation or a digital twin pays off
Some things cannot be explained in text because they happen where nobody can look. Which form of visualisation suits which problem, and what it costs.
7 min read →Photogrammetry: scanning your installation instead of modelling it
The biggest cost in 3D work is the model that does not exist yet. On installations from the nineteen-eighties there is nothing. Scanning is an option, with sharp limits.
7 min read →360-degree capture: the cheapest way to show a site
Between a flat photograph and a full 3D model sits a form that often gets skipped. Fast to produce, immediately recognisable, and with a limit people underestimate.
6 min read →WebXR: VR without managing headsets
What VR programmes come unstuck on is rarely the simulation. It is the cupboard with twelve headsets nobody charges. WebXR removes that, and costs you something else.
6 min read →VR training: who is it really for?
VR sells itself at a trade fair and does not sell itself back in a business case. An honest boundary around where it pays off and where it is an expensive detour.
8 min read →Why VR training is more than a gadget
In industry VR is still too often seen as an expensive gimmick. But for roles where mistakes are fatal and rehearsal is impossible, it is the only training format that genuinely works.
7 min read →Designing for people who are not students.
The material in most corporate academies is designed as though the audience has time, a desk and a quiet environment. The real audience works shifts, has no fixed workplace, does not enjoy long walls of text, and carries twenty years of practical experience the material rarely acknowledges.
That asks three things with little to do with technology. Material has to start from what somebody can already do, otherwise it reads as condescension and the experienced engineer drops out at screen three. The division into blocks has to fit between two tasks, because a forty-minute module never gets done in one sitting in practice. And there has to be repetition in it, because without repetition most of it is gone within days.
That last one is the best evidenced and the most ignored. A training given once and never revisited delivers, after a month, a fraction of what the attendance list suggests. Spaced repetition is not an optional extra on top of the programme, it is what makes the programme effective.
Language is not a translation question
On practically every industrial site in the Netherlands there are people whose first language is not Dutch, and on large projects they are the majority. The reflex is to do everything in English. That solves less than it looks: English too is a learned language for a large part of the audience, and safety-critical instruction in a learned language is demonstrably remembered less well.
So the question is not whether you translate, but how you stop translating from becoming a recurring project. Material existing as a separate file per language is guaranteed to drift apart the moment anything changes. Material with one source and separated language layers can be updated in a single action. That is a design choice at the start, not a supplier choice afterwards.
Further reading on design and language
Instructional design for technical audiences
Engineers and operators are not a difficult audience. They are a specific audience, and most teaching material was made for a different one.
7 min read →Breaking the forgetting curve: repetition that does work
Within a month most of a training day has gone. That is not a motivation problem but a design problem, and it is fixable.
6 min read →Training in your people's own language: why English is not enough
In your mother tongue you understand it. In a foreign language you hope you do. With safety instruction that difference is not academic.
7 min read →Maintaining multilingual material without translation chaos
Making five languages is doable. Keeping five languages current is where it goes wrong, and where organisations quietly fall back on one language not everybody understands.
7 min read →The task starts at delivery.
This is the part that carries least weight in the tender and costs most money after two years. With traditional production of learning material, the price of a change sits close to the price of building new. The source file is with an outside party, in a format you cannot edit yourself. There has to be an order, a schedule, a review, a new export, a new upload. To change three screens.
So everybody does the same sum: this change is too small to be worth that. That sum is correct per change. It is not correct after twelve changes. What then emerges is a second reality: the official procedure sits in the system, the real procedure sits in the heads of the people who work with it daily. Usually that second version is better. Until the moment it matters, at an audit or an incident.
The way out sits in the design, not in discipline. Modular instead of linear, so you can replace component by component. One source for several formats, so a change is one action. Ownership of the source files, arranged contractually, even when you outsource the updating itself. And a fixed review moment, because "update when needed" does not happen.
Where AI changes the sum
This is where the change of recent years sits, and it is a sober one. AI does not write good training. What it does do is cut the cost of the laborious steps sharply: turning an SOP into a first outline of a module, generating variants of assessment questions, producing voice-over in eight languages, synchronising subtitles, running an existing module against what has changed.
That does not lift the quality of the design, which stays human work. It does shift which changes are worth making. When updating drops from an hour to ten minutes, the threshold below which a change is left alone shifts. That is precisely the difference between material that ages and material that moves with you.
Further reading on maintenance and production
Procedures change. Your training material does not.
Practically every organisation has training material that no longer holds. Not through neglect, but because the cost of updating is higher than the pain of outdated material, until something goes wrong.
7 min read →From SOP to learning module: converting technical documentation
Your procedures contain everything somebody needs to know. They were simply never written to learn from. Here is what sits between those two forms, and what you should not automate.
8 min read →AI in training material production: what it does and does not do
AI does not write good training. But it does change which training projects are feasible at all, and that is a bigger difference than it looks.
8 min read →Showing that it worked.
An academy is ultimately judged on something other than satisfaction scores. The question that lands on the table at an audit, an incident investigation or a client declaration is narrow and hard: can you demonstrate that this person was competent for this operation on that date, and on what do you base that?
A completion rate does not answer that question. It demonstrates that somebody clicked through the material. The difference between "followed the module" and "can perform the operation" is exactly the difference an investigation magnifies.
That does not mean everything has to become a practical exam. It means deciding per subject, in advance, what evidence is sufficient: a knowledge assessment for rules and definitions, a scenario assessment for judgement, a signed-off workplace observation for operations carrying risk. And recording how long that evidence stays valid, because competence has a shelf life.
Recording it is a systems question better asked early than late. Which data you want to be able to retrieve, how long you keep it, and whether your learning platform delivers it in a form an auditor accepts. Building a report afterwards out of a system that was not set up for it is considerably more expensive than setting it up in advance.
Further reading on assessment and records
Assessing what somebody can actually do: from knowledge question to competence
An assessment that measures whether somebody clicked through the module measures nothing. Here is the difference between knowing and being able, and how to assess the second.
7 min read →Demonstrably competent: records that survive an audit
At an audit the question is not whether somebody can do it. The question is whether you can show it, and that is where most organisations stumble, not on the training itself.
7 min read →Learning analytics: what the numbers from your LMS do and do not say
Practically every learning platform reports completion. That is the figure that says least. Four measurements that do mean something, and the reversal nobody makes.
7 min read →Every article in one place.
The same articles as above, grouped by subject, so you can quickly find where you left off.
Vision
Building or renewing
Learning formats
- › When do you choose e-learning, microlearning, video or VR?
- › Microlearning on shift work: learning between tasks
- › Learning apps for staff without a desk
- › From classroom to blended: what changes for your trainers
- › Adaptive learning: skipping what somebody already knows
- › QR and NFC on equipment: instruction at the place of work
Imagery and simulation
- › Visualising work: when 3D, animation or a digital twin pays off
- › Photogrammetry: scanning your installation instead of modelling it
- › 360-degree capture: the cheapest way to show a site
- › WebXR: VR without managing headsets
- › VR training: who is it really for?
- › Why VR training is more than a gadget
Design and language
Staying current
New articles appear on Insights. Terminology is in the glossary, and concrete questions in the FAQ.
Are you building an academy, or has the existing one stalled?
We are happy to think along about where to start and what it realistically costs to keep it standing. No sales conversation.
Milan
Business development, STARK Learning