The cost of non-performance

The invoice is the smallest number on the page.

Three prices, one invoice

An intervention is bought on one number and paid for with three. The number that reaches the approval paper is almost always the smallest of them.

The first price is direct spend: design, facilitation, licenses, a venue, materials. It arrives as an invoice, it lands in a budget line, and it is the number everyone argues about.

The second is travel and subsistence, where the work is done away from the desk. It is booked, but usually somewhere else: in a travel budget nobody connects to the decision.

The third is working hours. Every participant is paid to be somewhere, and while they are in a room or in a module they are not doing the work the organization is paying them to do. This price is real, it is often the largest of the three, and in our experience it is often not booked anywhere at all.

The arithmetic of a working day

The calculation is not difficult. It is simply never made, because no single person is accountable for all three lines at once.

The third card is where most organizations lose the most and notice the least. A four-hour module rolled out to twenty thousand people costs six hundred thousand dollars to buy and four point eight million dollars to sit through. The purchasing decision is taken on the first figure. The organization pays the second whether or not anyone learns anything.

Putting a price on a point

The full price of an intervention is only half the page. The other half is what the gap is costing while nobody closes it, and unlike the price, that figure is almost never on the table.

Four routes lead to it. Take the first one that applies and stop; what is wanted is a defensible order of magnitude, not a valuation.

Two disciplines keep this honest. State the rate you used, in the document, next to the answer: a number whose derivation is hidden invites an argument nobody can settle. And be conservative on purpose: if the plausible range is $8 to $14, use $10.60 and say so. An analysis caught being generous once loses every later argument on merit it should have won.

The unit that travels best

One last decision about the figure, and it decides whether the number survives its first change of sponsor.

This is where the first question hands over. Number 05 establishes that a measure exists, is owned and is moving; it deliberately stops short of pricing it, because an analysis that reaches for money before it has a measure ends up defending a figure instead of establishing a fact. The measure arrives here, and here it acquires a value.

What you are betting on

The full price is only half the exposure. The other half is the chance that the money is pointed at something that was never the problem.

Performance is made possible by six conditions. Three sit in the work environment and three sit in the person, and Gilbert’s original ordering has survived four decades of practice largely intact (Gilbert, 1978; Chevalier, 2003). Number 09 sets out what each one looks like on the floor while it is missing; the column that matters here is the second one.

This is not an argument that training rarely works. It is an argument about frequency. When the cause is a knowledge or skill shortfall, well-designed learning is exactly right and nothing substitutes for it. When the cause is one of the other five, the best course in the world closes nothing, and the result keeps costing what it was costing.

And the evidence says the same thing

You do not have to take the model on faith. Four independent bodies of evidence converge on where the improvement potential sits, and none of them is a marketing claim.

Pooling eighty-nine studies of training transfer: whether the manager backs it, and what the climate is like in the place people return to, both track how much of a program survives contact with the job, and the work environment weighs six times more heavily on open-ended work than on tightly scripted work (Blume et al., 2010). Support from peers, manager and organization together accounts for roughly a third of the difference in whether training transfers at all (Hughes et al., 2020). And the largest synthesis of behavioral evidence outside work ranks knowledge and general attitudes among the weakest intervention targets, and access to what people need among the strongest (Albarracín et al., 2024): evidence drawn from health and policy behavior, so its extension to work organizations is an inference rather than a finding.

Deming’s half-sentence, and the half that has aged

The most quoted number in this field is still right about where the trouble sits. It is no longer right about who can do something with it.

Deming reached the same conclusion from manufacturing quality half a century ago, and put a number on it that has been quoted ever since: most of what goes wrong belongs to the system rather than to the individual (Deming, 2000, p. 270). He wrote that in an era of strict hierarchy, and assigned the system to management accordingly. That half of the sentence has aged. Much of what surrounds the work is now within reach of the team doing it: the sequence of steps, the checklist, the local rule, the handoff with the team next door, what gets discussed in the stand-up. Work-redesign research finds participative redesign among the approaches with the more promising evidence (Knight & Parker, 2021), and a team that can change its own conditions can act on a finding in a week rather than waiting a quarter for permission.

What has not moved to the team is the rest: capacity at a queue, the incentive scheme, a system contract, the target itself. So the practical reading of Deming in 2026 is two-part. Establish the condition first, then ask who can actually change it, and expect the answer to be the team more often than the sentence in the book suggests, and management still, for the conditions that carry a budget or cross a boundary.

The refresher that was never needed

A field service operation, forty-two thousand jobs a year, six hundred and twenty technicians. The numbers below are illustrative and reconcile from the rates stated with them.

First-time fix rate fell from 88 to 74 percent over eleven months. Every job that has to be revisited costs about $310 in technician time, vehicle and parts handling. At 88 percent the operation absorbed 5,040 repeat visits a year. At 74 percent it absorbed 10,920. The difference (5,880 extra visits) is $1.82 million (€1.60M) a year, and it had been running for the better part of one.

The service director’s proposal was a two-day technical refresher for all 620 technicians, run regionally so that nobody flew. Priced in full at the rates in section 02, with travel at the regional $250 a day rather than the off-site $450, that is $558k of direct spend, $310k of travel and $595k of technician hours: $1.46 million (€1.29M), and 9,920 technician-hours (some 1,240 working days) taken out of an operation already behind on its jobs.

What the fix cost instead

Republish the mapping to the terminals; print the crosswalk until the release lands. About $40,000 (€35k) and six weeks, most of it one integration engineer.

The refresher was not run. Nobody was told they had been wrong, and nobody was contradicted: the request was reasonable, the technicians were reasonable, and the fault codes were invisible from every desk above the van. What changed is that the money went to the cause instead of to the symptom.

It is worth being precise about what the analysis did and did not do. It did not prove that a refresher would have failed; nobody ran the counterfactual. It established that the condition the refresher addresses, knowledge and skill, was not the condition that was short, and it produced a named, testable alternative that could be implemented in six weeks and measured in eight. That is a different kind of claim from a promise, and it is the only kind an analysis is entitled to make.

The dividend, counted twice

A diagnosis pays in two directions at once, and organizations habitually count only one of them, usually the smaller.

The hours are worth dwelling on, because they are the line that closes the argument with an operations director. Twelve hundred and forty working days is not an abstraction in a service business running behind on its jobs. It is roughly five technicians for a year. Nobody had asked whether the organization could afford to take them out, because nobody had put that question in the same document as the invoice.

What to require before you sign

You do not have to become an analyst to raise the standard. You have to stop accepting proposals that are missing five elements, and start noticing which one is missing most often.

None of this slows a decision down by more than days. The analysis behind the case in section 05 took under a week, most of it spent riding along with four technicians and reading a release note.

What this does not claim

An argument this convenient for one side deserves its boundaries stated first. Three of them, plainly.

• It is not an argument against training. Where the cause genuinely is a shortfall in knowledge or skill, well-designed learning is exactly right and nothing substitutes for it. The claim here is about frequency, not about value.
• The rates are illustrative. The figures in section 02 are stated so that every total in this paper can be recomputed, not because they are your rates. Substitute your own, and your own exchange rate with them. The proportions move very little, and the argument does not depend on either.
• An analysis is not a guarantee. It establishes what is causing a gap and what would close it. It does not promise the fix will be implemented, or implemented well, and it cannot protect a decision from being overruled for reasons that were never about the evidence.

Price the request in front of you

Ten minutes, on the proposal on your desk now. Answer these five and you have the input a diagnosis needs. Leave one blank and you have found the first thing to establish.

Two problems, one method, sixteen papers

Two models carry the method. Gilbert’s six conditions ask what kind of cause; Rummler and Brache’s three levels ask where it sits. Both are sound and this series uses both, and it leads with Gilbert, because his unit is the accomplishment, which is what an organization already counts, and because his six conditions are the categories an organization can actually change.

Where the method comes from

Two further pieces close the set and carry no number, because each points at all the numbered papers. Before the guide sits Arets L&D Business Models, the destination piece for the people who receive the requests: four models a learning function can run, and the route from HR-driven to business-driven. Last sits Solutions that cannot work, the field guide, ordered by the sentence that lands on your desk: seventeen forms a performance problem arrives in, each with an address in one of the numbered papers. Three orders in all: the destination you choose, the question you are on, and the form that landed on your desk.

Sixteen papers, six parts

Number 01 is the map and number 02 is the foundation. From number 05 on, every number is depth on exactly one step of the method. The filled block is the paper you are holding; the two unnumbered pieces close the set.

Where every claim in this paper comes from

The three books named earlier in this paper are listed in full in number 02, so that they do not recur in every reference list.