13 October 2026

Completion isn’t the finish line: how to measure whether training worked

Most L&D teams can tell you who finished a course, but can we tell whether the course changed behaviour or moved the dial on a key KPI?

Will Thalheimer’s Learning-Transfer Evaluation Model, or LTEM (pronounced “L-tem”), was built to close that gap. If you’re a head of L&D being asked to prove value to the board, it’s a useful way to decide what to measure, and just as useful for deciding what to stop relying on.

Why LTEM exists

For decades, the Kirkpatrick four-level model has been the default answer to “how do we evaluate training?” Thalheimer’s argument is that it puts everything we call “learning” into one box. A tick-box knowledge check and a realistic on-the-job simulation both land at Level 2, though they tell you very different things.

LTEM, now in its thirteenth version (published October 2024), splits that box up and rates each type of measure by how well it can show learning worked.

The eight tiers at a glance

LTEM has eight tiers, split into what you measure in the learning and what you measure in the work.

Tier	What it measures	How far it can validate success
8. Effects of transfer	The impact of changed work behaviour on KPIs and stakeholders	Strong, but needs rigorous methods
7. Transfer to work performance	Learners using what they learned in their actual work	Strong, but needs rigorous methods
6. Task performance	Learners performing realistic tasks	Strong, especially when tested after a delay
5. Decision-making	Learners choosing well in realistic scenarios	Strong, especially when tested after a delay
4A. Foundational knowledge	Concepts that underpin work decisions	Adequate, but knowing isn’t doing
4B. Trivial knowledge	Facts and terminology not central to decisions	Inadequate
3A. Learner perceptions (effectiveness)	Learners’ views on things that drive learning, such as practice and motivation to apply	Useful hint, so back it up with objective data
3B. Learner perceptions (satisfaction)	Satisfaction, “would you recommend this?”	Inadequate
2. Learner activity	Attention, participation, interaction	Inadequate
1. Attendance/completion	Enrolment, starts, finishes	Inadequate

Four ideas worth taking away

1. Tiers 1, 2 and 3B can’t validate learning on their own. People can finish a course and still have learned nothing, learned the wrong thing, or decided not to act on it. Research summarised in Thalheimer’s report finds that traditional smile-sheet scores are virtually uncorrelated with learning results.
   
   
2. Those tiers still have a job. Completion and activity data are useful formative signals. They show you what’s happening in the learning and where to improve it. They just can’t be your proof of success.
   
   
3. Decisions and tasks matter more than knowledge. Foundational knowledge is worth checking, but the model nudges us toward measuring whether people can make realistic decisions (Tier 5) and perform realistic tasks (Tier 6). It also says to test remembering after several days or more, not just straight after the course.
   
   
4. You don’t need all eight tiers. LTEM isn’t a ladder to climb. It’s a map for choosing a sensible mix for each programme. Measuring work performance and business effects takes real investment, so it’s best reserved for your most strategic programmes.

Putting LTEM to work in Breeio

Breeio LMS gives you the tools, data and workflow to implement LTEM measures, and back up the training investment you’re making. Here’s how the two fit together.

Tiers 1 and 2: Foundations (use them formatively). Breeio’s dashboard, assignment and validity views, and learning logs show who has completed what, who’s overdue and where the risk sits, with exportable evidence for audits. This is your compliance backbone. For xAPI modules, Breeio also captures detailed learner activity, including responses to questions. Treat all of this as a diagnostic: it shows you where learning is stalling, not whether it has changed behaviour or impacted the KPI.

Tier 3A: Ask better survey questions. Breeio’s survey modules support multiple choice, Likert scales and typed-answer questions. Typed answers automatically get a sentiment score so you can review hundreds of responses. Results can be filtered by group, course and date range, scheduled as regular reports, downloaded to Excel (anonymised if you wish) and shared with line managers through My Team. You can use Breeio to move from “Did you enjoy it?” to questions about what drives transfer:

• How confident are you that you could apply this in your role?
• Did you get enough realistic practice?
• Will your manager support you in using this?

Tier 4A: Check the concepts that matter. Build assessments into e-learning modules and use question analysis (including xAPI response detail) to see which concepts people struggle with. Focus on the foundational ideas that drive decisions, not trivia. Because Breeio keeps the latest response by default, with an option to include every attempt, you can compare first attempts with retakes.

Tier 5: Test decision-making, not recall. Breeio’s AI-powered conversational and scenario modules let you build decision-making challenges tailored to the organisation and the learner’s role. The learner makes a call, the scenario probes their reasoning with follow-up questions, and everything is tracked in the LMS. This is a very different thing from a multiple-choice quiz.

Tier 6: See people do the task. Practical modules add manager (or designated expert) sign-off to a course. Learners upload evidence or reflect on what they did, and the manager signs it off or sends it back with notes. Use Breeio to confirm that people can do the thing, such as a mock client conversation or a supervised procedure. The AI Feedback Coach can also help learners reflect on their practice.

Tier 7: Watch for transfer on the job. The same practical-module process can be used for a later workplace observation. Pair it with a follow-up survey sent to learners or their managers a few weeks after training. Thalheimer is clear that Tier 7 needs rigorous methods to separate the effect of training from everything else going on, so treat sign-offs and surveys as a strong signal, not complete proof of causation.

Tier 8: Connect to business results. Breeio shows who was trained, when, in what, and how they did. Set that alongside your own operational data, such as error rates, incident numbers, customer scores or sales. Breeio's APIs will allow you to integrate with your Business Intelligence dashboards and bring business measures together with training data. Scheduled, exportable reports help you share the learning side of the story with senior leaders.

Skills tracking. Breeio lets you build a skills framework, award skills and levels on course completion, and report on gaps by group. To keep the data meaningful, tie a skill to a course that includes a Tier 5 or Tier 6 check, not just a completion.

A worked example: Know Your Customer training

Imagine you’re rolling out KYC training to 800 colleagues in a professional services or financial services firm.

1. Start with the end in mind. Tier 8: fewer KYC file errors, fewer audit findings and fewer late escalations. Tier 7: colleagues complete due diligence correctly on real client onboarding files.
   
   
2. Build the learning around the concepts that drive decisions. An e-learning course with Tier 4A checks on the foundational ideas, such as what a risk-based approach means, how to spot who ultimately owns or controls a client, and when something must be escalated. Skip the trivia, like memorising names of regulations.
   
   
3. Test decisions, not recall. Use Breeio’s AI scenario modules to put colleagues in realistic onboarding situations (Tier 5). Include a mix of relevant and irrelevant detail, such as an unusual ownership structure, a rushed partner or a client slow to supply documents. The learner decides whether to proceed, request more information or escalate, and the scenario probes their reasoning.
   
   
4. Add a practical sign-off for client-facing and onboarding roles. Use a practical module where the learner completes or reviews a real or simulated KYC file. For this module, choose the sign-off option “by learner first, then designated user(s)” so that your compliance or MLRO team signs it off, rather than the line manager (Tier 6).
   
   
5. Ask the right survey questions. A short Tier 3A survey: “How confident are you that you’d spot a red flag?” and “Do you know exactly when and how to escalate?” Filter results by team to see where confidence is lowest.
   
   
6. Follow up after a delay. A few weeks later, assign a short scenario refresher so you’re testing remembering, not just recall. For future years, the repeat pattern feature lets you give new joiners the full course and everyone else a shorter refresher.
   
   
7. Report. Use Breeio’s dashboard and scheduled reports for completion and overdue risk, plus exportable evidence for auditors. Then bring in your own compliance data, such as file review findings and escalation rates, to explore the Tier 7 and 8 story.

Completion still matters, since it’s your audit evidence. But now you also have evidence on whether people can make the right call.

Where to start

Pick one programme and ask where you’re measuring today. If the honest answer is “completion and a smile sheet”, the next move is a Tier 3A survey and a Tier 5 scenario, and Breeio can support you with both. Prove the approach on one programme, and then extend it.

Want to see how this works in practice? Book a Breeio demo.

Source: Will Thalheimer, “The Learning-Transfer Evaluation Model” (Version 13, 2024), worklearning.com/ltem. LTEM is © Will Thalheimer, shared under Creative Commons BY-ND.