The request is not the problem

Built on human performance improvement Gilbert, Mager and Pipe, Rummler and Brache, and the Dutch line of Arets, Overduin and Heijnen. Half a century of method, put within reach of every learning advisor.

A request is a solution in disguise

A manager comes to L&D and asks for training. The sentence sounds like a problem statement. It is a solution, chosen before anyone established what is actually stopping the result. Reaching for training before finding the cause is the most expensive habit in the field, and the Dutch literature gave it a name: the costly misunderstanding (Arets & Overduin, 2008; Arets & Heijnen, 2008).

The request is easy to act on, which is precisely the trouble. It arrives with the solution attached, a number of participants, sometimes a date. A department can process it without ever asking which result is missing, in whose work, and how anyone would see it. The programme runs, the dashboard looks healthy, and the problem sits where it sat.

What makes the habit expensive is not the invoice. It is that the request names the one cause a course can address, and that cause is rarely the one holding the result back. Meanwhile the working hours disappear, in numbers nobody put on the request.

This paper describes the alternative, which is old, well documented and until recently almost impossible to run on every request: find the cause first, then decide. And it describes what changes when every learning advisor has a performance consultant beside them at the intake, instead of one specialist for the few requests there is time for.

What a request really costs

50 people · 2 days · classroom

1,000 people · 1 day de-escalation training,

customer-experience training, complaint scores are down

sales per visit are falling

€70k

€720k

working time

50%

working time direct (trainer, venue)

36%

direct (trainer, venue) travel

14%

travel

never on the request

never on the request

Illustrative figures. Direct cost per participant €300 to €500 for classroom teaching or €25 for e-learning; travel €120 to €200; working time at a loaded €40 an hour. The proportions carry the argument, not the totals.

Two things stand out the moment the numbers sit on the table. The invoice is the small number, and the large one is working time, which almost never appears on the request. And time scales without mercy. The e-learning for twenty thousand people looks cheap because it is digital; it consumes roughly 160,000 working hours.

None of this is controversial. It is arithmetic. What has changed is that the arithmetic now takes less time than reading the request it belongs to.

Most of the leverage sits around the person

Thomas Gilbert built the discipline on one observation: a person’s environment engineers performance far more than what sits inside their head (Gilbert, 1978). Knowledge, skill and motive matter, and they operate inside a work system that either makes good performance easy or quietly makes it impossible.

Four decades of quantitative work point the same way. A synthesis of meta-analyses across sixteen determinants of behaviour finds that knowledge, general skills, attitudes and beliefs have negligible effects as targets of an intervention; habits and social support have medium effects; and access, the material and logistic means that make a behaviour feasible, has the largest (Albarracín et al., 2024). A training request aims at the weakest determinant on that list.

theworkaroundtheperson

expectations and feedback

tools, information, time

theperson

knowledge and skill

capacity

the only box training can fill

The consultant looks in this order: first at the conditions around the work, which are the likeliest cause and the cheapest to repair, and only then at the person. Knowledge and skill is the single box a course can fill.

Work the case backwards

Performance is produced from left to right: people carry out tasks, tasks combine into processes, processes deliver results. A case gets analysed the other way. Start with the business result the organisation wants, then work back through the processes and the tasks to the place where performance actually leaks. Starting at the tasks, which is where the training request points, means starting at the wrong end (Rummler & Brache, 1995).

performance is produced this way

the person

the task

01 · The result

02 · The gap Stated as a measurable outcome, not as an

The quantified distance between the result now activity. “More customer focus” is a wish. “Repeat

and the result needed, with a value attached. Only orders at 62 percent against a needed 75” is a

a number tells you whether the problem is worth case.

solving and whether it got solved.

None of this takes months. A good performance analysis is lean and fast, aimed at the few causes that matter rather than at an exhaustive audit. Its output is a short, evidence-based diagnosis that decides whether training is the answer at all.

Why so few functions work this way

If the case is this strong and this old, the obvious question is why almost no L&D function runs on it. Not because the method is in doubt. Because the daily pressures of the job run against it at every turn.

No access to the real client

The confidence to challenge

The conspiracy of convenience

Between L&D and the manager who

Investigating a cause means

Everyone prefers the quick answer. owns the problem sit intermediaries:

questioning a manager’s own

The manager sees action, L&D gets a HR business partners, competency

diagnosis. Many advisors have neither

clear deliverable, and nobody has to owners, whoever handles training. The

the mandate nor the practice for that

sit with the possibility that the person who could describe the

conversation.

problem is the system they run. business priority is rarely in the room.

None of these barriers is an argument against the method. Each one is an argument for making the method cheap enough to run on every request, which is exactly what changed.

What the Performance Consultant AI does

The tool guides a conversation, not a form. It sits beside the learning advisor at the intake and runs the same five moves every time.

In plain language

01 · The request It starts with one field and one sentence: “a two-day sales training for all ten thousand reps, is training the fix?” No intake form, no template to fill in first.

About numbers, not wishes

02 · The questions back The tool asks where the problem shows up in the figures, who is missing which target, and where the revenue leaks. The advisor supplies the detail the organisation already has.

What the organisation knows

03 · The data Documents, dashboards and loose figures go in. The diagnosis rests on the organisation’s own evidence rather than on an assumption about it.

Cause by cause

04 · The gap, quantified Current and needed results side by side, and under each one the cause, tagged by factor, by level, and by how far the evidence carries.

And the documents to use it

05 · The advice Four versions of the same diagnosis, edited by the advisor, ready for the meeting where the decision gets made.

The method does not change. What changes is the scarcity: an analysis that took a specialist weeks now fits inside the intake conversation, on every request rather than the few there was time for.

The anatomy of a diagnosis

Diagnosis

retail · 1,000 people · request: customer-experience training

sales per visit €18.40, target €21.00, worth about €2.6M a year

out-of-stocks on the top 40 lines

pricing errors at the register

no feedback on visit value per store

fix replenishment and the price file first; coach three stores; no company-wide training

STILL TO CONFIRM · does the pattern hold in the four stores that already meet the target?

Illustrative diagnosis. The same anatomy whether the request covers fifty people or twenty thousand.

Every cause carries its confidence

It names what would settle the matter Hypothesis, substantiated or triangulated. The

The questions it would ask next, so that a strong tool marks its own uncertainty instead of hiding it

diagnosis becomes a watertight one before behind fluent language.

anyone spends money.

One conversation, four documents

The person who owns the money, the manager who owns the team and the analyst who owns the detail do not read the same document. The same diagnosis therefore comes out four ways, from one conversation, without a second analysis.

Full

Exec

one intake conversation

Manager

Slides

This is what changes the report L&D brings to the table. For years the function reported what it did: completions, participants, satisfaction. It can now report what it protected and what it made possible, in the language the board already speaks.

The dividend as a standing metric

€1.42M

94%

78%

4.2/5

€1.42M

Illustrative dashboard. The dividend reported in the same cadence as completions, skills coverage and engagement.

Performance consulting is the crossing

The Arets L&D Business Models place four ways of working on one rising line. The two on the left begin after someone else has decided that learning is the answer. The Performance Partner begins before that decision, establishes the cause and prescribes what the business needs, training or not. The Performance Consultant AI is what makes that crossing survivable on a normal working day.

Learning Specialist Order Taker

Order Taker

Learning Specialist

Performance Partner Takes requests and delivers courses.

Designs excellent evidence-informed

Establishes the cause first and Attendance counts as success. The

learning, and starts only once somebody

prescribes accordingly: process, tooling, cause of the problem never gets

else has decided that learning is the

guidance, or training where training is established, so the result rarely moves.

answer.

genuinely the answer. Includes 02.

The programme, the tool, the author

The programme

The Performance Consultant AI Tulser delivers performance consulting as one

The tool runs beside the intake conversation and work, learn and implement programme for an L&D

is free to try. It replaces neither the L&D team nor team: ten real requests from your own

the consultant: it removes the reason that used to department, asked again; an advice per request

decide the matter by default, which was never that names what is going on, what helps and what

having the time for an analysis. tulser.ai it delivers; and one way of working the whole team uses afterwards. tulser.com

Selected sources

Albarracín, D., Fayaz-Farkhad, B., & Granados Samayoa, J.

Gilbert, T. F. (1978).

Human competence: Engineering A. (2024). Determinants of behaviour and their efficacy

worthy performance

. McGraw-Hill. [no DOI available] as targets of behavioural change interventions.

Nature Kahneman, D. (2011).

Thinking, fast and slow

. Farrar, Reviews Psychology, 3

(6), 377–392. https://doi.org/ Straus and Giroux. [no DOI available] 10.1038/s44159-024-00305-0 Mager, R. F., & Pipe, P. (1997).

Analyzing performance Arets, J. (2026).

Learning that crosses the levels: The problems: Or, you really oughta wanna

(3rd ed.). Center Learning Conversion Framework

[White paper]. Tulser. for Effective Performance. [no DOI available] [DOI to be added] Rummler, G. A., & Brache, A. P. (1995).

Improving Arets, J., & Heijnen, V. (2008).

Kostbaar misverstand: van performance: How to manage the white space on the training naar business improvement

. Academic Service. organization chart

(2nd ed.). Jossey-Bass. [no DOI [no DOI available] available] Arets, J., & Overduin, B. (2008).

Liever (g)een training: op Stolovitch, H. D., & Keeps, E. J. (Eds.). (1999).

Handbook weg naar performanceverbetering

. Academic Service. of human performance technology

(2nd ed.). Jossey- [no DOI available] Bass/Pfeiffer. [no DOI available] Binder, C. (1998). The six boxes: A descendant of Gilbert’s behavior engineering model.

Performance Improvement, 37

(6), 48–52. https://doi.org/10.1002/pfi.