Blog

Training needs analysis — a 4-step guide for small teams

Four-step framework to run a training needs analysis in a team under 100 people. Interview prompts, a gap matrix, and two concrete SMB examples with timelines.

·7 min read·Mentor Team

An HR lead at a 70-person company gets an €18,000 L&D budget. The director's first question: "Where is the money going?" Without a needs analysis, the answer falls into generic training for everyone. Six months later the director looks at the numbers and sees 180 training hours with no measurable effect.

A training needs analysis (TNA) is a four-step framework that turns a business problem into concrete modules. This guide breaks each step into questions, tools and two SMB examples.

Why TNA usually fails in SMBs

Three mistakes keep repeating:

  • It starts with content, not the problem. Someone suggests an Excel course. Six months later the team still does not use Excel in the actual workflow.
  • Interviews only with managers. Managers describe the ideal picture; employees know the reality. Both are required.
  • Priorities without a yardstick. Fifteen suggested modules, all "important". Without a rule for ordering them, most never reach production.

The four-step framework addresses all three mistakes. Each step produces one concrete output that feeds the next step.

Step 1: Define the business outcome

TNA does not start with "what should employees learn". It starts with "which business outcome do we want to move". Three outcome types in SMBs:

  • Compliance. A deadline, a fine or an audit risk. Example: GDPR, workplace safety, NIS2.
  • Operations. A process defect, slow onboarding, a customer complaint. Example: support ticket time, invoice error rate.
  • Growth. New product, new market, new role. Example: sales reps selling a new module.

Write the outcome in one sentence with a metric. "Reduce sales rep ramp from 90 to 60 days" is a good outcome. "Better sales team" is not.

Without a metric, the project has no end. The metric is also a test for the director. If they refuse the metric, the module has no executive sponsor.

Step 2: Interviews across three layers

In an SMB, three layers of interviews are enough — 6 to 8 conversations of 30 minutes each:

  • Two department heads. What is the real problem, what was tried, what did the data show.
  • Three or four employees who run the process. Where it stalls, what they do differently from the written procedure, how much time it takes.
  • One recipient of the outcome. A customer, a colleague from another team, a sales lead. Where the problem shows up externally.

Three practical tips for interviews:

  • Record, do not take notes. A short transcript gives more than live notes, which HR filters while listening.
  • Ask for the last incident. "Describe the last time the process broke" uncovers more than general questions.
  • Check what employees actually do. A common finding: the written procedure exists, the team runs its own version in practice.

After the interviews you have 15 to 25 raw findings. Each one goes into the matrix in the next step.

Step 3: Gap matrix

The gap matrix is a simple table that separates three causes of a problem:

  • Knowledge. The employee does not know how.
  • Skill. They know, but cannot execute.
  • System. They know and can, but the tool or process blocks them.

Only the first two causes are solved by training. The third one needs a process or tool change. In SMBs a common split is roughly 30 % knowledge, 20 % skill, 50 % system.

That is an important finding. Half the suggested modules do not solve the real problem. Example: "We need a customer-communication course", when the real issue is a CRM that never sends automatic reminders.

Build the matrix by placing each interview finding in one row and tagging the cause. After a week of work you have 20 rows. Eight to ten need a module; the rest go into process changes.

Step 4: Prioritise by three criteria

Eight modules are still too many for the first quarter. The priority yardstick has three axes:

  • Risk × size. A fine or the cost of an error. Example: a GDPR fine, or an invoice error of €500 × 12 months.
  • Number of people affected. A module for 50 people has a bigger total effect than one for 5.
  • Time to proof. How many weeks until a measurable result. A module with a 4-week cycle shows impact in the same quarter.

Score each module 1 to 3 on each axis. The sum is its priority. The top three go into the first quarter; the rest wait for the next cycle.

The yardstick takes the decision out of debate. HR does not say "this one is more important" — they show a table with three columns. Managers agree faster because the rule applies to all.

Example 1: Services company, 55 people

The managing director had a €15,000 L&D budget and an owner request to drop customer churn from 18 % to 12 % per year.

TNA in three weeks:

  • Week 1. Business outcome: churn 18 → 12 %. Metric: quarterly churn by cohort.
  • Week 2. Interviews: two team leads, four account managers, two users from CRM support. 19 raw findings.
  • Weeks 2 to 3. Gap matrix. 5 knowledge rows (product knowledge on the new line), 4 skill rows (handling an unhappy customer), 10 system rows (CRM without alerts on signals, no customer segmentation).
  • Week 3. Prioritisation. Three modules for Q1: product refresher (55 people, 20 min, medium risk), de-escalation on the phone (12 account managers, 45 min, high risk), internal reference sheet (55 people, 10 min).

Four months in, churn sits at 14 %. The remaining gap is addressed by a CRM alert rebuild in the next quarter. Spent: €9,400. The rest goes into the second cycle.

Example 2: Manufacturing company, 85 people

The operations director reported 7 safety incidents in the past year. None serious, but an inspection was announced.

TNA in four weeks:

  • Week 1. Business outcome: 0 recordable incidents in the next 12 months, passing inspection.
  • Week 2. Interviews: production lead, maintenance lead, safety engineer, six operators across three shifts, worker-council rep.
  • Week 3. Gap matrix. 3 knowledge rows (new machine line), 2 skill rows (first-aid procedure), 9 system rows (poorly marked routes, broken visual alarm on two machines).
  • Week 4. Prioritisation by risk. Two modules in month one: a 5-minute safety refresher (all staff) and a 15-minute briefing on the new machine (12 line-B operators). System changes go into a parallel project.

Four months after launch: 0 incidents, inspection passed. The operations director at the review: "Month eight would have been too late. Four weeks of TNA gave us the right focus."

For structuring the modules themselves, see the skills matrix template for SMBs.

Common mistakes

  • TNA once, then silence. Results decay without a cycle. Repeat TNA every six months with a lighter scope.
  • Too many interviews. Twenty conversations in an SMB stop being signal. Eight is enough for a team under 100 people.
  • Matrix without a system column. If every row reads "knowledge", someone is not listening. Always split knowledge, skill and system.
  • Priority list without a sponsor. If the sponsor does not sign off on the top three, the first urgent fire will push them aside.

The last mistake is the most expensive. A TNA without sponsor sign-off becomes an HR document in a folder, not a quarter-level agenda.

What to do this week

Write one business outcome with a metric and a 30-day deadline for the first interview. That is step 1. Without it, every later interview pulls in a different direction.

An LMS like Mentor covers the second half of TNA — from the gap matrix to module production. The four-step framework decides what ever reaches the system in the first place. Once TNA becomes a quarterly habit, the L&D budget is easy to defend to the director with a metric, not with anecdotes.

Ready to try Mentor?

14-day free trial, no card. Set-up in under 5 minutes.

Start trial