How to Run a Training Needs Analysis That's Worth Doing

A training needs analysis that can only conclude 'yes, build the training someone already asked for' is not an analysis. It's a formality. A real one can conclude that training isn't the answer — and often should.

Start with the performance gap, not the training request

The request arrives as a solution: 'we need a course on the new expenses policy'. The analysis question is different: what is someone doing now that they should be doing differently, and what would change if they did?

If nobody can describe the observable difference, there's no gap to close and no way to know whether the training worked.

The five causes, and only one is training

Performance problems have a small number of root causes. Asking which applies takes ten minutes and saves entire projects:

1. They don't know how. A genuine knowledge or skill gap — training helps.
2. They don't know they should. Expectations were never made explicit — a communication problem.
3. They can't. Tools, access, time or authority are missing — a resourcing problem.
4. It's easier not to. The correct behaviour is slower or harder than the wrong one — a process problem.
5. Nothing happens if they don't. No consequence, no feedback — a management problem.

Training addresses cause one. Applied to causes two through five it produces a course that people complete and then continue exactly as before — the most common form of wasted training spend.

Who to talk to

Not just the person who requested it. At minimum:

• Someone doing the job well — what do they do that others don't?
• Someone struggling — what actually gets in the way? The answer is often not knowledge.
• Their direct manager — what do they see, and what do they do about it now?
• The requester — what triggered this? A specific incident usually lies underneath.

Four conversations of twenty minutes each will tell you more than a survey of two hundred people.

Look at data that already exists

Incident reports, error rates, complaint themes, help-desk tickets, quality assurance results, exit interviews. Most organisations are already recording evidence of the gap and nobody has looked at it in the context of the training request.

Be willing to write the unwelcome conclusion

Sometimes the honest finding is that the process is broken, the tool is unusable, or people are being asked to do something the system doesn't allow. Saying so is uncomfortable — particularly as an external consultant being paid to build training.

It is also the single most valuable thing you can do, and it tends to produce more work rather than less. A consultant who tells a client their real problem becomes the person they call next time.

Keep it proportionate

A needs analysis shouldn't cost more than the training. For a small internal course, a few conversations and a look at existing data is enough. Reserve formal analysis for programmes large enough to justify it — and be honest that the point is a better decision, not a thicker document.

What to produce

Short: the gap in observable terms, the root cause, the recommendation (including 'not training' if that's the answer), what success would look like, and what you'd need to measure it. One or two pages that someone will actually read and act on.