Home / Blog

Evaluating Training Without a Formal Kirkpatrick Programme: A Lightweight Two-Level Model for Corporate L&D Buyers

· By Ayda Editorial

Evaluating Training Without a Formal Kirkpatrick Programme: A Lightweight Two-Level Model for Corporate L&D Buyers

Why most Malaysian training programmes get evaluated at Level 1 and nothing else

A post-course "happy sheet" takes two minutes to fill and about five seconds to read. That is why so many HRD Corp claimable programmes end with a stack of satisfaction scores, a folder of attendance records, and no answer when a manager asks whether anyone actually learned anything.

You do not need a six-month measurement programme to fix that. The Kirkpatrick model, which Malaysian providers most often meet through Kirkpatrick certification programmes delivered locally, has four levels: reaction, learning, behaviour and results. A lightweight evaluation deliberately covers the first two properly and treats the last two as a separate, later conversation.

That split matters commercially. Reaction and learning evidence is generated inside the training day, using artefacts your trainer is already producing. Behaviour and results evidence is generated on the job, weeks later, and usually needs a manager, a coaching structure or a post-training retainer to make it exist at all. Mixing the two into one promise is where most evaluation plans quietly collapse.

What can you realistically measure at Level 1?

HR buyer and training provider reviewing an evaluation form together at a briefing meeting Ask to see the actual form fields before signing — a real form beats a promise of 'comprehensive evaluation'.

Level 1 measures reaction — how participants experienced the session — and it earns its bad reputation because most forms measure the wrong things. "The trainer was engaging" tells you nothing you can act on.

A reaction form worth keeping does three jobs: it captures relevance to the participant's actual job, confidence that they can apply one named technique, and a free-text field for what they intend to do differently on Monday. That third field is the highest-value item on the page and it costs nothing to add.

Fifteen minutes at the end of the day, a QR code, and a fifteen-second read per response is the whole operation. Replace the general satisfaction rating with a relevance rating and the form starts producing decisions instead of applause.

What can you realistically measure at Level 2?

Level 2 measures learning — whether knowledge or skill actually moved. The good news is that a well-built course already contains the instrument: the quiz, the case exercise, the role-play checklist.

Run the same short assessment before and after the module. The before-score tells the trainer where to spend time; the after-score is your learning evidence. Five to eight items is enough. A 20-question exam that nobody marks is worse than a six-item check the trainer reviews live.

For skills-based programmes, replace the quiz with a scored observation checklist — three observable behaviours, marked yes/no by the trainer during practice. That single page is more defensible than any multiple-choice test, and it doubles as feedback the participant keeps.

Keep a raw record: participant identifier, pre-score, post-score, date, trainer name. Store it with the attendance sheet. If a grant query ever arrives, that record answers the question quickly, rather than forcing an improvised reconstruction.

What should a trainer ask a provider at the briefing stage?

Before signing, ask four things. First: what is your Level 1 form, and can I see the actual fields? Second: what Level 2 instrument exists, and who marks it? Third: will the pre- and post-scores be handed to me as raw data, or only summarised? Fourth: what happens to the results — does anyone review them, or do they go into a folder?

A provider who can answer these in a five-minute conversation has usually done the work before. A provider who promises "comprehensive evaluation" without showing a form is describing an intention, not a process.

These are the same due-diligence questions that belong in choosing and vetting freelance HRD Corp trainers, where the evaluation instrument is a more useful signal than the trainer's marketing deck.

Where do Level 3 and Level 4 actually live?

Level 3 (behaviour) and Level 4 (results) rarely happen inside the training contract, and pretending otherwise is how providers end up over-promising and under-evidenced.

Behaviour change needs a line manager asking about application at the 30-day mark. Results need a business metric with a baseline. Neither is a trainer's deliverable unless it is explicitly scoped.

If a buyer wants 30- and 60-day follow-up, the honest structure is a separate engagement — which is exactly the territory covered in designing a post-training coaching retainer, where cadence, scope and fees are priced as their own line rather than folded into a day rate.

How does this evidence sit alongside an HRD Corp claim?

Be precise here. HRD Corp claim requirements are set by HRD Corp and administered through eTRiS, and the specific documents and conditions evolve. Confirm current requirements on the official HRD Corp portal before you build a file structure around them.

What evaluation evidence does for you is internal defensibility: it lets you answer a management question, justify next year's levy allocation, and show a participating manager that the day changed something. It is not a substitute for the claim documentation HRD Corp asks for.

Keeping evaluation records filed with attendance and programme records means one filing discipline serves both purposes. Our eTRiS query response playbook covers how to organise evidence when a query arrives, and the same folder logic applies here.

What does a lightweight model look like before and after?

Before. A provider runs a two-day supervisory skills programme. Evaluation is a 10-item satisfaction form. At the end, HR receives a PDF of averaged scores, no participant-level data, no learning evidence. When the regional manager asks in March what the RM30,000 bought, nobody can say.

After. The same programme carries a six-field reaction form and a six-item pre/post knowledge check. The trainer marks the post-check live, captures a scored behaviour checklist for the two coached role-plays, and hands over a one-page summary plus a spreadsheet of raw scores tied to participant identifiers. The cost difference is roughly ninety minutes of trainer time and one extra template. The March conversation now has a baseline and a delta.

That is the entire value of lightweight evaluation: it converts a satisfaction gesture into a baseline you can defend.

What should be written into the contract?

Contract clauses are where evaluation promises either survive or evaporate. Four clauses do most of the work:

  • Named instruments. List the reaction form fields and the Level 2 assessment method as annexes, not adjectives.
  • Data handover. Specify that raw, participant-level scores are delivered within a stated number of working days, in a stated format.
  • No-claim clause. State plainly that evaluation results are internal evidence and that no party guarantees any regulatory or funding outcome.
  • Re-run rights. Allow one re-run of the Level 2 assessment at no fee if the post-check is lost or mis-marked.

For providers, these clauses also protect margins — an unnamed deliverable is an unbounded one. If you are tightening your own commercial documents, course pricing for training providers in Malaysia shows how to price the measurement hours rather than absorbing them.

And if the programme is delivered in hybrid mode, where half the room is on screen, the evidence trail gets harder: see the hybrid training setup framework for attendance evidence for how to keep online participants' scores and attendance records as clean as the in-room sheet.

Which numbers should a lightweight model actually track?

Keep the dashboard to five figures per programme: response rate on the reaction form, mean relevance score, mean pre-score, mean post-score, and the delta. Add one qualitative line — the most common "what I will do differently" answer.

Resist the temptation to convert that delta into a ringgit ROI figure. Phillips-style ROI work is a real methodology with its own evidence demands, and it needs financial data a trainer rarely holds. A cautious buyer keeps Level 2 delta and Level 3 follow-up as separate reports; conflating them produces a number nobody senior will trust twice.

Run the whole thing as a repeatable template across programmes so the numbers are comparable year on year. That comparison, not any single score, is what makes training evaluation for HR teams genuinely useful at budget time.

Frequently Asked Questions

Do I need Kirkpatrick certification to run a lightweight evaluation?

No. Certification helps if you want to design and defend a full four-level programme or train others in the methodology, but a Level 1 reaction form and a pre/post knowledge check can be built without it. Certification matters most when you are selling evaluation design as a service.

How many questions should a Level 1 reaction form have?

Five to seven fields is the practical ceiling. Prioritise relevance to the job, confidence to apply one named technique, and a free-text field on what the participant will do differently. Drop generic satisfaction ratings that produce no decision.

Is a pre- and post-quiz enough evidence for a training claim?

It is useful internal evidence, not a claim document. HRD Corp sets its own claim requirements and these change over time, so confirm the current list on the official HRD Corp portal before relying on any file set.

Who should mark the Level 2 assessment?

The trainer, on the day, where possible. Live marking gives participants immediate feedback and produces a clean record. If marking happens later, fix a stated delivery timeline for the raw scores so it does not drift.

How do we evaluate behaviour change without a large budget?

Use a short manager check-in at 30 days with three application questions, or scope a small coaching retainer. Either way, make it a named, priced deliverable — unpriced follow-up almost never happens.

← All articles