How to Do a Training Needs Analysis – Before You Build Anything

The request arrives in a familiar form: "Sales are down in the northeast region. We need training." Or: "Customer complaints are up. Can you build a course on customer service?" It is tempting to say yes and start building, but a course built on a guess about the problem often misses, and everyone moves on assuming training didn't work. A training needs analysis is how you avoid that. How do you do a training needs analysis that tells you what to build, or whether to build anything at all?

To do a training needs analysis, define the business goal behind the request, describe the performance you need, measure current performance, find the causes of the gap and then decide which causes training can fix and which need a different solution. Gather evidence from several sources, such as interviews, observation, job data and short surveys, rather than relying on one stakeholder's view. Look at three levels: the organization, the job tasks and the people doing them. The output is a short report that recommends specific training, non-training fixes or both, with measures for success. Done well, it takes days, not months, and it often makes the eventual training smaller and far more effective.

What Is a Training Needs Analysis?

A training needs analysis (TNA) is a structured investigation into the gap between how people are performing and how they need to perform, and into whether training can help close that gap. It is the analysis phase of most instructional design models, done carefully enough that the rest of the project rests on facts.

The classic framework comes from William McGehee and Paul Thayer, who in 1961 described three levels of analysis:

• Organizational analysis: What are the organization's goals, and where is performance falling short of them?
• Task analysis: What does the job actually require people to do, and to what standard?
• Person analysis: Who needs to improve, by how much, and what do they already know?

All three levels matter, because if you skip the organizational level you may train on the wrong problem. Skip the task level and the course won't match the real work. Skip the person level and you'll waste the time of people who already perform well.

How Do You Do a Training Needs Analysis, Step by Step?

These steps work for projects of almost any size. On a small project, each one may take an hour, while on a large one it may take a few days.

Step 1: Clarify the Request and the Business Goal

Start by asking the requester what problem they are trying to solve and how they would know it was solved. "Build a customer service course" becomes "reduce repeat complaints about billing errors, which we track in the complaint system." A measurable goal anchors everything that follows.

Step 2: Describe the Desired Performance

Define what people should be doing, in observable terms. Talk to managers and, especially, to top performers. What do the best people do that others don't? Write a short list of specific behaviors and the standard for each.

Step 3: Measure Current Performance

Find out what is happening now. Look at existing data first: error reports, quality audits, complaint logs, sales figures, help desk tickets. Then observe the work or review samples of it. The gap between steps 2 and 3 is the performance gap.

Step 4: Find the Causes

This step separates a real needs analysis from a formality. For each important gap, ask why it exists. A useful test comes from Robert Mager and Peter Pipe's work on analyzing performance problems: if a person's life depended on it, could they do the task correctly? If yes, the problem isn't knowledge or skill, and training won't fix it.

Step 5: Match Solutions to Causes

Knowledge and skill gaps call for training or practice. Other causes call for other fixes: clearer processes, better tools, job aids, feedback, changed incentives or removed obstacles. Most real projects need a combination of these solutions.

Step 6: Profile the Audience and Constraints

For the gaps training can address, gather what you need to design well: who the learners are, what they already know, where and when they can learn, which devices they use and what time and budget are available.

Step 7: Recommend and Set Measures

Summarize findings, recommend solutions and define how success will be measured. Linking measures to the business goal from step 1 makes later evaluation straightforward. Many teams use the four levels of the Kirkpatrick Model (reaction, learning, behavior and results) to structure those measures.

Why Isn't Training Always the Answer?

Because many performance problems have little to do with what people know. Thomas Gilbert's Behavior Engineering Model, published in 1978, offers a helpful way to sort causes, which it splits into factors in the work environment and factors in the individual.

Gilbert argued that the environment side is usually the better place to look first, because it tends to have a larger effect and is easier to change. If people don't know what's expected, don't get feedback, lack the right tools or are rewarded for the wrong behavior, training will struggle to help. Only the "knowledge and skills" cell is fully a training problem.

Which Data Collection Methods Should You Use?

Use more than one method, because each has blind spots, and combining them gives a more reliable picture.

• Interviews with managers, top performers and struggling performers reveal reasons and context.
• Observation shows what people actually do, which often differs from what they say they do.
• Existing data, such as error logs and quality scores, shows the size and pattern of the problem.
• Short surveys reach larger groups, but work best for confirming patterns rather than discovering them.
• Work sample review, such as reading completed forms or listening to recorded calls, shows the quality of output.
• Task analysis breaks a job into steps and decisions, which becomes the backbone of any training you design.

The mistake I'd most like to retire is relying only on surveys that ask people what training they want. Preferences are useful, but they're not the same as needs.

What Does a Needs Analysis Look Like in Practice?

Consider a typical case. A company's field technicians are taking longer than expected to close service calls, and the operations director requests "advanced troubleshooting training."

The analysis starts with the data. Long calls are concentrated in one product line, not spread across everything. Interviews with top technicians show they carry a laminated wiring chart for that product that most others don't have. Ride-alongs with average technicians reveal two issues: they don't know a diagnostic shortcut the top performers use, and they often wait on the phone for parts availability because the mobile app shows inventory poorly.

The recommendation has three parts. Distribute the wiring chart as an official job aid and fix the inventory display in the app. Build a short practice-focused module on the diagnostic shortcut for that product line, with spaced follow-up questions. The measure is average time to close calls for that product line. The training piece is far smaller than "advanced troubleshooting training," and much more likely to work.

What Should the Final Report Include?

Keep it short, because decision-makers need findings and recommendations, not every interview note. A useful report covers:

1. The business goal and how it is measured
2. The performance gap, with evidence
3. The causes, sorted into training and non-training
4. Recommended solutions for each cause
5. For training, the audience profile, objectives, format and constraints
6. Success measures and when they will be checked

When recommending format, think about the full mix. Many needs are best met by combining short self-paced pieces, hands-on practice, live coaching and job aids. That kind of blended learning approach lets each part of the solution use the format that suits it.

How Long Should a Training Needs Analysis Take?

Scale the analysis to the stakes. A request for a short update to an existing module might need a single conversation and a look at one report. A new program for hundreds of people, or one tied to safety or regulation, deserves several days of interviews, observation and data review.

A useful rule of thumb is to keep analyzing until you can explain the main causes of the gap with evidence, and stop when new interviews keep repeating what you've already heard. Stakeholders who worry about delay usually relax once they see that a few days of analysis can save weeks of building the wrong thing.

What Mistakes Undermine a Training Needs Analysis?

A few patterns show up again and again:

• Treating the analysis as a formality after the solution has already been decided
• Talking only to managers, never to the people doing the work
• Asking people what training they want instead of observing what they need
• Ignoring environmental causes because they're outside the training team's control
• Skipping baseline measures, which makes later evaluation impossible

The fix for nearly all of these is the same: follow the evidence, and be willing to report that part of the problem isn't a training problem.

Analyze First, Build Less, Change More

A training needs analysis answers the most important question in any training project before money is spent: what is actually causing the performance gap, and what will fix it? Define the goal, describe the desired performance, measure the current state, find the causes and match solutions to them. Look at the organization, the tasks and the people.

The payoff is practical. Training that targets real skill gaps works better, and problems caused by tools, processes or incentives get fixed at the source.

The smallest useful version is one hour of analysis before you agree to any deliverable. Talk to a top performer and a struggling one, and look at one piece of real data. That single hour often changes the direction of the project.

← All articles