How much does it cost to build a training programme? The industry's answer was written in 2010

Where the money actually goes now, how to use a pilot properly, and the five questions worth asking instead of cost per finished hour.

It is the first thing almost every buyer wants to know and the last thing most suppliers will tell them. You ask for a number. You get offered a discovery call.

Some of the evasion is honest. A serious programme genuinely cannot be priced before anyone knows what it is for. But the effect on buyers is not honest at all. People end up comparing quotes with no sense of what a reasonable number looks like, which is how an organisation ends up paying a great deal for something thin, or paying very little for something that has to be built again properly eighteen months later.

I have been on both sides of this. I have commissioned large programmes as a buyer and quoted for them as a supplier. So here is the arithmetic, in public, with no sales call attached.

The number the industry actually uses

Almost every cost calculator, agency estimate and "how long does eLearning take" article in circulation traces back to a single study. It was run by Bryan Chapman at Chapman Alliance, drawing on 249 organisations and nearly 4,000 learning professionals, covering content consumed by close to twenty million learners. It remains the most cited piece of research in this market.

It found that one finished hour of basic eLearning, meaning content pages and a quiz, takes around 79 hours to produce. Interactive material averages 184 hours. Advanced work, simulations and scenario based design, runs from 217 hours at the low end to 716 at the top.

The part almost nobody quotes is the money. Chapman asked the same organisations what a finished hour actually cost them to make.

So the industry's own benchmark says the cheapest thing you can build costs five figures an hour to produce. Worth holding on to the next time a quote surprises you.

The data was collected in September 2010.

That is worth sitting with too. The benchmark underpinning most of the pricing conversations happening in this market this week was gathered before Articulate Storyline existed. Before responsive delivery to phones was standard. Before video was cheap. More than a decade before generative AI.

The industry has not replaced it. The most serious recent attempt, from ATD in 2023, largely concluded that development time now varies so widely between organisations that no single ratio holds. Which is true, and unhelpful, and is why a study from 2010 is still doing the work.

One thing to understand about those figures before you use them

They are US organisations reporting their own internal cost of production. Salaried instructional designers, developers, media people and project managers, on a US payroll, in 2010.

That matters in two directions.

If you are comparing suppliers who design and build entirely in Ireland, the UK or Western Europe, this is close to the right benchmark for you, and it is a high one. A quote in that world that comes in far below it is telling you something about scope, not about efficiency.

If a supplier's number comes in materially lower and they produce outside a Western cost base, that is not automatically a worse deal. It is a different cost base, and there is nothing wrong with a different cost base. What decides whether it works is not where the building happens. It is who is directing it, how close they are to you, and whether they carry the accountability when something is wrong. Production travels well. Judgement does not travel on its own.

The question to ask is not where the work is made. It is who is answerable for it, and can you get them on the phone.

What has actually changed, and what has not

Production has collapsed. That part is real and anyone telling you otherwise is protecting something. Drafting, scripting, media, voice, build, the mechanical assembly of a course. All of it is faster and cheaper than it was even three years ago, and it will be cheaper again next year.

If your programme is mostly production, you should be paying less than you did. Possibly a great deal less.

But production was never where the risk sat. Five things have not moved at all.

Deciding what the programme is actually for. Not the topic. The decision the learner has to be able to make afterwards, and the evidence you will accept that they can make it. Most programmes are commissioned without this written down anywhere, which is why so many of them are content rather than capability.

Getting the knowledge out of the people who hold it. The expensive part of subject expertise is not the expert's time. It is the extraction. The people who know how the judgement actually works are usually the busiest people in the organisation, and what they know is mostly tacit. Nothing about this has got cheaper, and a plausible AI draft makes it worse, because a fluent wrong answer takes longer to correct than a blank page takes to fill.

Designing an assessment that survives being challenged. If your credential carries any consequence, somebody will eventually contest a result. An assessment built to be defensible looks very different from one built to produce a completion record, and the difference is entirely in the design.

Getting it into your systems and proving it works. This one is almost always missing from the brief. Putting a course on a website is straightforward. Packaging it for your learning platform, tracking who did what, testing it end to end on the actual platform rather than the supplier's, and fixing what breaks, is a separate piece of technical work. If your quote does not mention it, ask whether it is included or whether your own team is expected to do it.

Keeping it right afterwards. A regulated programme is accurate on launch day and quietly wrong eighteen months later. Almost nobody prices this and almost everybody pays for it.

And there is one more that eats budgets quietly. Chapman found that uncontrolled design revisions add around 40 per cent to development time. Cheap drafting makes this worse rather than better. When producing another version costs almost nothing, you get more versions, more opinions and a longer review. The constraint that used to impose discipline has gone, and nothing has replaced it except an agreed sign off process, which most projects do not have.

Ask for a pilot, then ask the second question

Most suppliers will build you one module at a reduced rate before you commit to a programme. If nobody has offered, ask. It is a normal part of how this market works and there is no reason to be shy about it.

A pilot is worth far more than a demo, and the two get confused. A demo is something the supplier has already built for somebody else, polished, and shows to everyone. A pilot is your content, your subject experts, your review process, your accessibility requirements, your platform. It tells you what working with these people is actually like, which is the thing you cannot learn from a proposal.

So insist the pilot is a real module, not the easiest module. Pick something with genuine assessment weight and at least one awkward subject expert. If it goes well on the hard one, the rest is scheduling.

Then ask the second question, which most buyers forget. Is the reduced rate a genuine reduction, or is it recovered across the rest of the programme? Both answers are legitimate. A supplier absorbing real cost to win the work is behaving normally, and so is one who discounts the first module and prices the remainder accordingly. What you cannot afford is not knowing which one you are dealing with when the full quote arrives and it is larger than the pilot led you to expect.

And be honest with yourself about what a pilot commits you to. If you have run a pilot with one supplier, absorbed their thinking and adopted their structure, you are not really going back out to tender. Price that in.

The most valuable person in the room is the one who says no

Here is where a lot of budget goes and almost nobody writes it down.

You will want things. A branching simulation because a colleague saw one. Custom animation on the introduction. A gamified leaderboard. Video where a paragraph would do. Every one of those is a real cost, and some of them are large costs, and a proportion of them will add nothing whatsoever to what a learner can do at the end.

The supplier is being paid to build. If they say yes to all of it, they get a bigger project and you get a bigger invoice, and everyone is polite about it.

What you actually want is a partner with somebody senior enough to tell you that the animation is not earning its place. That the leaderboard will produce a spike of activity and no change in capability. That the branching simulation is exactly right for the two decisions carrying real consequence and a waste of money on the other nine sections. That is not a supplier being difficult. That is a supplier with a learning specialist in the room rather than a sales lead, and it is the single largest difference between a programme that costs what it should and one that costs half as much again for the same result.

You can test for this early. Put one clearly indulgent idea into the brief on purpose and see what comes back. If it is priced without comment, you have learned something useful before you have spent anything.

Why two honest quotes for the same brief differ by a factor of ten

They usually are both honest. That is the part buyers find hardest to believe.

A quote can cover writing and building what you send over. Or it can cover working out what the programme should teach, extracting that from your people, designing an assessment that will hold up, building it, packaging it to a standard, testing it against accessibility requirements, loading it into your platform, and being on the hook when the regulator or the accreditation panel asks a question.

Those are not the same job. They are not close to the same job. The gap between them is not margin. It is scope, and scope is very easy to leave unstated in a document that looks, on the surface, like a like for like comparison.

The five questions worth asking a supplier

Cost per finished hour is the least useful number to compare suppliers on, because it prices the component that has collapsed and ignores every component that has not. Two suppliers can quote an identical rate per hour and be selling completely different things.

Hours per finished hour is still a perfectly good way to understand how much harder one kind of course is to make than another. It is a bad way to compare two prices. Those are different uses of the same measure, and the second one is where buyers get caught.

These five do more work.

What proportion of this quote is production, and what proportion is judgement? A quote that is almost entirely production is a quote for something you can very nearly do in house now. That may be exactly what you want. Just be clear that is what you are buying.

Who is making the design decisions, and will I ever speak to them? In this market the person who sells the work is frequently not the person who scopes it, and neither is the person who builds it.

What would you cut from this brief? If the answer is nothing, they are selling you build hours. Anyone who has done this properly can name two things in your brief that will not repay their cost.

How many review rounds are included, and what happens on round four? Ask before you sign, not after. If the answer is vague, the overrun has either been priced in somewhere or left out and will arrive later as a variation.

What happens in month eighteen? If there is no answer, the maintenance cost is not zero. It has been moved onto you and left out of the quote.

The part you control, and a tool for working it out

Everything above is about reading somebody else's number. There is one part of this you control entirely, and it is the largest single variable in what you will pay.

How much is decided before you brief anyone, and how you intend to run the project once it starts.

A programme where the outcomes are agreed, the assessment approach is settled, the experts are named and released, and one person can say yes, costs a fraction of the same programme briefed as a folder of slides and a hope. Not because the building is different. Because the deciding gets done once, by you, instead of repeatedly, in review rounds, at supplier rates.

If you would rather start further back, the Programme Design Audit scores an existing programme across six design dimensions and tells you how ready it actually is before you go near a supplier. Also free, also no form.

The point

The cheap quote and the expensive quote are usually both real prices. They are prices for different things.

The useful question is not which number is lower. It is which one has priced the part that will still matter in two years, and whether you can tell from the document in front of you.

If you cannot tell, that is not a failure of your procurement process. It is a feature of how this market writes its proposals.

For the prior question of whether to build at all, see custom eLearning versus off the shelf. For what a build actually involves once you commission one, see what a custom eLearning build actually looks like. For choosing who builds it, see how to choose a custom eLearning company. See more insights from LearnFrame.