In the second week of February, an L&D manager receives a spreadsheet titled Training Needs 2026. It has fourteen rows, and every row is already a course: Negotiation Skills, Presentation Impact, Time Management, Advanced Excel, and ten more. Each has a headcount and a preferred quarter. The business calls this its training needs analysis.
It is not one. It is a shopping list, and approving it commits the budget before anyone has established a single thing that needs to change.
Here is the claim worth arguing about: a needs analysis that ends in a list of courses has not found a need. It has collected solutions. On a spreadsheet the two look identical, and that is the problem.
Why the request arrives as a solution
The requester is not being careless. They are describing a real frustration in the only vocabulary the organisation gives them. A regional head watches three deals stall at the proposal stage and concludes the team needs negotiation training. The performance problem is real; the diagnosis has been skipped and the intervention chosen in the same sentence.
Two things cause this. Training is the one lever L&D is resourced to pull. If the answer is a broken lead handoff, an incentive that rewards volume over margin, or a manager who never reviews proposals, there is no budget line for it. So the problem is translated into the one thing that can be bought. Separating a business problem from a course request is consulting work, and it is the step the annual calendar quietly removes. Second, the word need does quiet work: add it to a course title and a preference starts sounding like a finding.
The framing is old. McGehee and Thayer set out three levels of analysis in 1961: organisational, task and person. Mager and Pipe, in Analyzing Performance Problems, made the sharper point that a performance discrepancy is not proof of a skill deficiency. It might be motivation, feedback, tools or consequences. Training is one remedy among several. The request list is not wrong about the symptom; it is silent about the cause.
Question one: what is the person doing, and what should they be doing?
Take one row. Ignore the course title. Ask the requester to finish two sentences about a specific person in a specific moment: “When [situation] happens, [role] currently does [action]. The standard we need is [different action].”
The test is whether a stranger could watch a recording and tick either column. “Lacks confidence in negotiation” fails. “Opens with a discount before the client states a constraint” passes. If you cannot write the behaviour as an action, you have an impression, not a performance problem.
This is where a competency framework earns its keep. An anchor written as an attitude cannot be rated, and the case for a short set is the one we made in a competency framework nobody uses. Two observable behaviours per role is enough. Twelve produces a document nobody can hold in their head.
Question two: is the shortfall a skill gap, or something else?
Now ask why the gap exists. This decides whether training is the answer at all. Work through the causes in order, cheapest check first.
| Check | Question | If yes |
|---|---|---|
| Standard | Does the person know what good looks like, in observable terms? | Clarify the standard. Not a course. |
| Feedback | Does anyone tell them, within a week, whether they did it well? | Fix the review. Often enough alone. |
| Consequence | Is the old behaviour rewarded, or the new one punished? | Change the incentive or the workload. |
| Tools and process | Is the information, authority or equipment available when needed? | Fix the handoff or the system. |
| Skill and knowledge | Having removed all of the above, can the person still not do it? | Now training is a candidate. |
Mager and Pipe made exactly this distinction: a discrepancy caused by a skill deficiency differs from one caused by motivation or poor feedback, and the remedies differ. Formal training is one option; practice, positive reinforcement and adequate feedback are others. A course request is a hypothesis about the last row. Most of the evidence, once you ask, sits in the first four.
Fourteen course requests therefore do not become fourteen courses. Several collapse into the same broken review loop. One or two have no performance gap at all, only a future capability that belongs in a dated plan, not a workshop.
Question three: what number moves, who records it, and where is it read?
A gap statement is not complete until it has an indicator and an owner. Ask which existing number would change, who writes it down today, and in which meeting a manager will look at it.
Prefer a number the business already collects: defect notes, proposal-to-meeting conversion, first-response time, on-time dispatch. A survey invented for one programme dies with the programme. This is the same logic as the smallest honest version of four levels of evaluation: measure something a finance review will recognise, not a satisfaction score.
If no such number exists, that is a finding. The organisation cannot see the performance it says it cares about. An assessments practice that baselines capability before a programme and re-measures after it is one way to make it visible.
The artefact: one gap statement per line
Write the answers as one sentence, no more than forty words.
“For [role], [current behaviour] occurs in [X] of [N] [situations], against a standard of [expected behaviour]. Cause: [skill or environment]. Success: [existing indicator] moving from [a] to [b], reviewed [cadence] by [named owner].”
Compare the two forms. A row that says “Negotiation Skills, 18 people, Q2” becomes: “For account managers, the first response to a pricing objection offers a discount before the client states a constraint, in 9 of 12 sampled calls. Standard: two diagnostic questions first. Cause: skill. Success: discount-free first proposals from 25% to 60%, reviewed in the Monday pipeline meeting by the regional head.” A row that says “Time Management, 30 people, Q3” becomes “No agreed observable behaviour; requester to supply one before this line is costed.” That honest gap is better than a confident fiction.
Only now does format matter. The training solutions conversation is about which behaviour to build and how to rehearse it, not which title to buy. A two-hour drill, a manager review script and a changed pipeline template may cost less than the two-day course and do more.
A worked example
Illustrative example. A consumer-goods company sent fourteen requested courses from four regional heads, covering 96 nominations. One morning of applying the three questions with each requester collapsed the list into four gap statements.
Two lines, Presentation Impact and Business Writing, were the same problem: proposals went out without a named decision-maker or a date. The cause was a template and a review habit, not a skill. No course ran. The template changed and the regional head added one line to the weekly review. One line, Advanced Excel for nine people, had no gap at all; the team would need it in six months, so it became a dated plan item.
One line was a genuine skill gap. Eleven supervisors ran the morning shift handover without naming open exceptions. The night log showed 31 exceptions in a fortnight; the handover note mentioned 6. The standard became “every open exception named, with an owner and a time”. The intervention was a forty-minute drill, a one-page handover card and a week-three review by the plant head. At six weeks, 27 of 31 exceptions were being carried into the note.
The nominal budget was about Rs 14 lakh. The work that shipped cost a fraction of that and four hours of management time. These figures are illustrative, not a client result or a benchmark. The point that generalises is the collapse: fourteen requests, four gaps, one training intervention.
The honest limit
This method needs three things, and fails honestly without them.
It needs performance data. With no calls, defect notes, handover logs or conversion numbers, the three questions produce hypotheses, not gaps. You can still write the standard and the cause test, but the indicator is a promise. Say so. Do not invent a baseline to make the template look complete.
It needs a sponsor who will hear the wrong answer. Question two often puts the cause in an incentive, a reporting line or a manager who does not review. That finding can embarrass the person who signed the request. Where L&D is measured on heads trained and budget spent, fewer courses is a hard sell even when correct. The method makes the trade-off visible; it does not fix the incentive.
And a gap statement is no guarantee of transfer. A behaviour that is designed, rehearsed and still unrewarded will fade, for the reasons we set out in why behavioural training fades. Three gap statements across a portfolio is manageable; thirty means the diagnosis has stretched back into a catalogue.
What to do on Monday
Open the request list you are working from now. Pick the single most expensive line. Book thirty minutes with the person who requested it, ask the three questions in order, and write the answers as one gap statement on a single line. Do not discuss vendors, dates or formats until that line exists. If the requester cannot name an observable behaviour, the line is not ready to be costed, and that is a legitimate outcome.
Then send one sentence to whoever approves the budget: this course is on hold until we can state which behaviour changes, what causes it and which existing number will show the change. On a fourteen-line calendar that usually removes three or four lines before a request for proposal goes out, and gives the rest something a designer can build against. Start with the most expensive line; the rest is easier on Tuesday.
Sources
No external statistics are used in this article. The worked example above is illustrative. The following sources were consulted for the framework and the sequence of the argument.
- McGehee and Thayer’s three levels, and TNA as the gap between current and desired performance: https://trainercentric.in/the-3-levels-of-training-needs-analysis
- Define the performance gap before a training request, and the rule that a non-skill cause needs a different remedy: https://ceriusexecutives.com/performance-gap-needs-analysis
- Anchor a needs analysis on the business problem rather than the course wish list: https://www.talaera.com/corporate-training/training-needs-analysis
- Mager and Pipe on distinguishing a skill deficiency from motivation and feedback causes: https://eric.ed.gov/?id=ED050560