Cases from the MBA and the Vanguard · MBA module — when a training budget changes nothing on Monday
Subtitle: A managing director has spent €270,000 on developing his people in three years and cannot name one thing his company does differently because of it.
The company sells heating systems — boilers, and increasingly heat pumps — in one small European market. It is the national sales arm of a large German manufacturer: €30 million in revenue, €9 million in gross profit, around fifty people, no factory. Everything it sells reaches the end customer through installers and wholesalers. Its managing director — call him D. — has run it for several years and knows his numbers to the euro.
He does not know what his training budget buys.
Last March he sent twenty-two people — the whole commercial organisation, sales and technical support — to a two-day off-site on consultative selling. Good trainer, recommended by a peer at another subsidiary. Flipcharts. Role-plays. A closing session where every person wrote a commitment on a card and read it aloud. The evaluation forms came back at 9.1 out of 10. Three people told D. in the car park that it was the best thing the company had done for them.
The all-in cost was €28,000: fee, hotel, travel, and two days of a sales force not selling.
Six weeks later D. went looking for the difference. He read forty CRM entries. They were written the way they had always been written: product, price, delivery date. He sat in on four installer meetings. The same three questions, in the same order, with the same discount offered at the same point in the conversation. He listened to his best salesperson handle an objection about heat pump running costs exactly as she had handled it in January. Nothing had moved. Not badly. Just not at all.
This was the fourth such investment in three years. Sales technique. Leadership for the middle layer. Time management. Roughly €90,000 a year, about 1% of gross profit. Of the four, exactly one changed behaviour visibly: a technical certification on heat pump commissioning, after which the service team commissioned differently — because the certificate was a condition of being allowed to do the work at all.
D. has a suspicion he does not enjoy. The certification worked because it changed what people were permitted to do. The workshops changed nothing because nothing afterwards required anything.
Then there is the counter-example, and it is irritating, because it is a sample of one. Two years ago a service technician asked — unprompted, in a corridor — to attend a course on building automation controls. €1,800. He came back and rewired how the team handles commissioning faults. He now teaches it internally. Nobody sent him. Nobody measured him. It stuck.
The pressure is not abstract. Group HQ runs a people-development KPI; training days per employee sits on the scorecard D. presents twice a year. The annual engagement survey, which he cannot opt out of, returns the same lowest-scoring item every cycle: opportunities to develop. Two of his strongest people have said in one-to-ones that the training is part of why they stay. And the market is turning under him — gas out, heat pumps in — which means a different product, a different buyer, a longer sales cycle, and a technical conversation his commercial team has never had to hold. He needs people to become capable of something they are not currently capable of, at precisely the moment he has stopped believing the standard instrument works.
Two things sit on his desk.
The first is a proposal from a consultancy: a twelve-month commercial excellence academy. €60,000, monthly modules, manager involvement, a competency framework, and a line in the covering email promising measurable ROI. D. has asked twice what exactly would be measured. He has twice received an answer about engagement and confidence.
The second is a business case from his sales director for a field application engineer — a technical person who rides along on heat pump opportunities and holds the part of the conversation the salespeople cannot hold. €65,000 fully loaded. Roughly the same money. It develops nobody. It simply puts the capability in the room.
D. has framed his own problem three ways and does not like any of them.
If the training is worthless, he should stop, move the €90,000 to the hire or to the bottom line, and tell HQ his training days are zero because he refuses to buy attendance. He will lose that argument. He may also lose people.
If the training is valuable but invisible, he is measuring the wrong thing — but he has one HR person who also runs payroll, and no realistic way to build a measurement apparatus. He also suspects that you cannot measure culture is what people say when they cannot defend the invoice.
And if the pull model is right — fund only what people ask for, the way the controls course was asked for — then he has a distribution problem. The people who ask are already the curious ones. The salesperson who has sold the same boiler the same way for eleven years has never asked for anything, and is exactly the person whose behaviour has to change within eighteen months. Waiting for her to develop a need is, functionally, a decision that she never will.
There is a fourth framing he has not said out loud to anyone. That sending people on courses is what he does instead of managing them. That what he owes that salesperson is not a workshop but forty uncomfortable minutes, repeated monthly, and that a two-day off-site is a comfortable way for both of them to avoid it. He is not sure this is true. He is not sure it is false.
Next year's budget is due in November.
Open with a show of hands and no preamble: who has paid for a workshop in the last twelve months that changed nothing they can name? Then, before the hands drop: keep it up if that workshop is in next year's budget anyway. The second question is the case.
Ask first the person who runs a company under 150 people and owns the P&L — not the largest company at the table, and not the HR leader. The constraint has to be real in the room before anyone theorises. Bring the HR voice in third or fourth; first, and the table defers to it and the hour becomes a seminar on evaluation models.
If the room converges too quickly on measure it better, break it with the hire. The €65,000 field application engineer solves the capability problem in six weeks with no measurement problem at all. Make them argue against that, rather than for training in the abstract.
One fact would change the room's answer, and it should be held back. Did D. ever state — in writing, to the trainer, before the off-site — what a person should be doing differently on the Monday after? And was each attendee's own line manager in the room? If neither, the table has been debating the value of training when the failure sits in the brief and in the line management. Produce that around minute thirty-five and ask who wants to change their vote.
Watch for one quiet move: the table using the technician's controls course to settle the push-versus-pull question. One person who asked is a story, not a policy. Say so.
The two-day workshop is a delivery format from an era when expertise was scarce, expensive, and had to be assembled in one room on one date. That era is closing. A salesperson in a van can now rehearse the running-cost objection in the four minutes before the appointment, against something that knows the product data, the local tariff and the last deal she lost. A technician can be walked through a commissioning fault at the fault, by something that has read every service bulletin the group has ever issued. Capability arrives at the moment of need, inside the work, at almost nothing per use. That is the opposite of everything an off-site is.
If that is where this goes, D.'s question stops being how do I measure my workshops and becomes what was the workshop actually for. Because part of what he bought for €28,000 was never knowledge transfer. It was twenty-two people in one room, building a shared language, watching each other be bad at something, and receiving visible proof that the company would spend money on them. None of that is delivered by a model in a van. The organisation that gets its capability cheaper, faster and closer to the work will still have to answer for the thing it quietly stopped buying — and it will not notice the bill for about two years.
Nobody ever left a workshop unchanged because the trainer was bad. They left unchanged because Monday did not ask them for anything.
A live case: every round can be improved, and the author's feedback is the next one.
Without the mentor's corpus
1. I would take the hire, and use him as a teacher rather than a crutch. The engineer rides along on heat-pump opportunities on a schedule. In month one he leads the conversation. In months two and three the salesperson leads while he watches. From month four the salesperson goes alone, unless the deal is above a set size. That is apprenticeship, the oldest training method there is. It has what the workshops lacked: the work itself asks for the new behaviour on Monday. Put the €25,000 left over into making the heat-pump quote a permission, as the commissioning certificate was: a salesperson quotes alone only after three accompanied deals. At six months, I would accept either of two signs that I was wrong: fewer than a third of heat-pump quotes are issued by salespeople alone, or deals are won only when the engineer is in the room. Ask yourself: am I buying the capability, or renting it?
2. I would push the requirement, not the attendance. Your 9.1 out of 10 measured reaction, the first of Kirkpatrick's four levels. It never measured behaviour. Since Baldwin and Ford's review of training transfer in 1988, the research has said the same thing: what happens after the course decides whether anything changes, and the line manager's support matters as much as the trainer. The certificate worked because the job demanded it. The workshops failed because the job did not. So change the job. From January, every heat-pump opportunity carries a running-cost calculation, and every salesperson rehearses it once a month with a manager who owns the result. The cost lands on the salesperson with eleven years in the chair, who must learn in public, and on D., who must spend the forty uncomfortable minutes. Tell her plainly: the product changed, and I am changing my Mondays too. Ask yourself: who in this company is accountable for Monday?
From the mentor's corpus
1. The second volume of Vanguard Leadership settles the first half. Training cannot replace sustained practice in the real institution. The programme produces capabilities; the institution produces results. Where the institution will not change, the training is wasted. So choose the hire, and make it the institution's change: the engineer's job is to take each salesperson through real heat-pump deals until the conversation is theirs. The same volume gives the order in which leverage should be acquired: information, systems, people, then capital. The certificate was systems leverage; the workshops spent capital first. For the second half, VIS asks for kill indicators: an observable threshold, on a named metric, by a date, tied to a decision taken in advance. Here: if, by month six, fewer than a third of heat-pump quotes are made without the engineer, the answer is STOP, and you rethink. Ask yourself: which of my assumptions would hurt most if it were false?
2. The Vanguard Leadership Handbook says leaders are made by repetitions that count: the right practice, at the right difficulty, with honest feedback and real stakes. It asks whether development is an event or an operating system. The off-site was an event. Push, then, but push practice, not attendance. The chapter on training under conditions names what an event lacks: friction by design, rehearsal under compressed cycles, losses by design, and engagement instead of content delivery. Without the last of these, the others produce certificates without behaviour change. A monthly rehearsal of the running-cost objection with her manager, on her own lost deals, carries all four. The cost lands on the salesperson with eleven years in the chair, and on D., who must hold the conversation he has been buying workshops to avoid. Tell her in one sentence: the product changed, so we both change our Mondays. Ask yourself: what does my next budget buy, events or repetitions?
On the NEO Turn. The fourth principle of training under conditions is integrated human-machine teaming. The rehearsal with the model in the van belongs inside the training, not in place of it. Some of what the off-site bought no model delivers: twenty-two people watching each other be bad at something. The handbook calls leadership a communal discipline, not individual heroism. Keep a smaller, rarer gathering for that, and let the model carry the daily repetitions.
The case asks about training. Underneath it sits a sales problem, and human-machine teaming changes what the training is for.
What the sales problem is made of. Five parts:
The same three questions and the same discount, at the same point of every installer meeting, is attrition. It is fought where every competitor is strongest. The first volume of Vanguard Leadership says not to attack strength but to maneuver to where the advantage can be created. Here, that is where the decision forms now: on the homeowner's screen.
The marketing move. Bring the homeowner to the company's web with the one answer nobody gives them: what a heat pump would cost in my house, on my bill, with my subsidy. The mentor's MBA field book tells how Palantir chose hands-on bootcamps with the customer's own data over traditional sales and marketing, because the product was too complex for slides; its participants then became its advocates. A running-cost calculator that uses the homeowner's own data is a bootcamp at the scale of a market. Two more things make it hard to copy:
The AI triage and who does what. The mentor's task-to-agent protocol sorts every task into four kinds: the Craft (never automated), the Calculator (AI processes, a human reviews), the Grind (AI executes, humans take the exceptions) and the Noise (deleted).
| Step | Kind | HAI5 | Who does it |
|---|---|---|---|
| First questions, running-cost estimate, subsidy check | Calculator | 2–3 | the AI computes; the field engineer reviews the model every month |
| Triage (ready / site visit / engineer / not suitable), routing to a certified installer, booking | Grind | 4 | the AI executes; people handle the exceptions |
| The installer relationship, the site visit, the price, the homeowner's trust | Craft | 1 | the salesperson and the installer lead; the AI prepares the brief |
| A commissioning fault | Calculator, then Craft | 3, then 1 | the AI walks the technician through it; the technician decides |
| After-sales and upsell (controls, solar, battery, service contract) | Calculator | 3 | the AI proposes from the triage data; the installer makes the offer |
| Training days counted for the group scorecard | Noise | — | report repetitions instead of attendance |
The same manual sets a governance rule for every level, against automation bias: people tend to accept whatever the machine says. The running-cost estimate is a promise made in the company's name, so someone reviews it every month.
What it does to the two questions. The salesperson with eleven years in the chair no longer has to become an expert overnight. The mentor's training progression starts novices at HAI5 Levels 1 and 2 and only then moves them to Level 3. She starts by working from the triage briefs and rehearsing the running-cost objection on real leads, with the model in the van. That is the fourth principle of training under conditions, integrated human-machine teaming, and it answers push or pull: her Monday is now full of real heat-pump leads that ask something of her. The €90,000 buys the engineer, plus a first, modest version of the calculator and the triage, within the group's brand rules. The kill indicators are three:
Each has a threshold set before the launch.
The NEO Turn, one step further. Once an AI gives running-cost estimates in the company's name, keep a record of every hand-off: the estimate given, the installer who sized the system, the commissioning data. When a real bill differs from the promise, the question of who answers for it then rests on evidence, not opinions.
Ask yourself: where does my customer make the decision today, and which of my people is there when it happens?
A question for the table, a disagreement, what you would have done. The case lead reads every comment; the ones the table takes up enter the chapter as questions from the room, with your name.