How to Run a Training Needs Analysis

Skipping it is rarely a deliberate decision. A department identifies a performance problem, somebody proposes training as the remedy, and a budget is approved before anyone has established whether the people concerned lack knowledge, lack practice, or are working around a system that makes the correct action difficult. Training addresses the first two. It does nothing at all for the third, and it is expensive.

A note on the name

Training needs analysis, learning needs analysis and skills gap analysis are used more or less interchangeably, and the distinctions people draw between them are mostly territorial. There is a broad convention: a skills gap analysis tends to look across a workforce for hiring and workforce planning, while a training needs analysis is narrower and tied to specific work. The CIPD uses learning needs analysis; AIHR and most North American practice use training needs analysis. Pick one and be consistent.

Start from the work, not the job description

A job description describes the intended role. It was probably written for a hiring process and has not been touched since. Assessing capability against it tells you whether people can do the job somebody imagined.

Assess against the process as it is actually performed. If you have a current-state map, use it. If you do not, that comes first - and it is genuinely first, not a nice-to-have, because the whole analysis inherits its accuracy from whatever you measure against.

Define the requirement as tasks, not topics

The most common error is writing requirements as subjects: “understands the credit process”, “familiar with the CRM”. Those cannot be assessed and cannot be trained against, because nobody can say when they have been achieved.

Write them as things a person can be observed doing, unaided. “Can run a credit check and correctly decide whether to release the order.” That is assessable, it is trainable, and it tells you immediately whether the training worked.

Establish the current position from more than one direction

1. The mapping interviews

   If you have already sat with these people, you know a great deal about what they can and cannot do, and that knowledge should be the starting point.

2. Supervisors

   They see the work daily and they know exactly who they route the difficult cases to, which is usually the most honest capability data in the building.

3. Watching the task performed

   Expensive but decisive, and worth it for the two or three tasks that matter most.

4. Self-assessment, carefully

   Unreliable in both directions - people overstate familiar tasks and understate ones they have quietly mastered. Useful as a cross-check, never on its own.

Classify the gap before you decide the remedy

This is the step that saves the most money, because four quite different problems all present as “they need training”.

A knowledge gap. They do not know something. Training or documentation genuinely helps.

A skill gap. They know what to do and cannot yet do it reliably. That needs practice with feedback, not a deck.

A tooling gap. The system makes the correct action difficult or impossible. No amount of training fixes this, and delivering training against it is actively corrosive - it tells people their competence is the problem when the tool is.

A communication gap. Nobody told them. Common, embarrassing, and fixed in an afternoon rather than a program.

In our experience a meaningful share of what arrives labelled as a training need turns out to be one of the last two.

Split the content three ways

For every requirement, decide: must be known from memory, must be findable in under a minute, or never needed. The third bucket is usually the largest, and moving it out is what turns an unusable program into a usable one. The second bucket is the argument for job aids over courses.

Rank by cost, not by volume

You will find more gaps than you can close. Rank them by what they cost - error rates, rework, delay, escalations, the risk if one person leaves - rather than by how many people have them or how loudly they were raised. A gap affecting one person on a critical path usually beats a gap affecting thirty people on something inconsequential.

What you should be holding at the end

A short document, not a report. For each role: what it must be able to do, where it currently falls short, what kind of gap each shortfall is, which ones are being addressed, and what is deliberately out of scope.

That last line is the most useful sentence in the document. It is the one that stops the training doubling in size between approval and delivery.