Level five is where training measurement goes to exaggerate.
Phillips extended Kirkpatrick with a fifth level: convert benefits to money, subtract costs, publish a percentage. Done honestly, it disciplines programs; done as usually practised, it launders assumptions into three-digit returns. Here’s the honest version, and the machinery that feeds it.
Where ROI calculations go wrong, step by step
The Phillips chain — impact data, isolation of training’s effect, monetary conversion, cost subtraction — fails predictably at its middle links. Isolation by participant estimate (‘what share of your improvement came from training?’) imports every optimism bias in the room; conversion multiplies that estimate by contested money values; and the resulting 340% ROI decorates a slide nobody quite believes, correctly.
The honest version strengthens what can be strengthened and confesses the rest. Capability at delay is measured, not estimated — the first link becomes evidence. Isolation uses design where possible (staggered rollouts, natural comparisons) and labelled conservative estimates where not. Conversion uses your finance team’s numbers with assumptions printed. The output is a range with its reasoning visible — less impressive than 340%, and infinitely more usable.
The first link, made of data
Retention and capability curves at delay replace the ‘learning occurred’ assumption — the chain starts from measurement, which disciplines everything downstream.
Isolation by design, not survey
Staggered rollouts and cohort comparisons — the platform’s deployment patterns create isolation opportunities most programs waste. We help you use them.
Ranges that survive the CFO
Sensitivity bands on every assumption: the ROI presents as a defensible range, and the CFO’s challenge improves the model instead of demolishing it.
Level five, with its receipts
Isolating training’s effect needs measurement at every step — the chain from reaction to ROI, each link on this screen.
Interface shown as an illustration with representative numbers, not a screenshot — the layout is the product’s.
Run one program’s honest ROI.
Measured inputs, design-based isolation, your finance values — the number will be smaller than the theatre version, and it will survive the meeting.
The evidence this page stands on
Questions buyers ask
Is calculating training ROI worth doing at all?
For expensive, strategic programs — yes, honestly done. For routine training, capability metrics serve better; forcing ROI everywhere produces exactly the theatre this page warns about.
What’s a credible training ROI figure?
Credible figures come with ranges and printed assumptions. The single confident percentage — especially a large one — is the genre’s tell for theatre.
How do we isolate training’s effect without control groups?
Staggered rollouts are the practical instrument: later cohorts serve as comparison before their start. Where even that fails, use conservative labelled estimates — honesty about weak isolation beats confident fiction.
Phillips or Kirkpatrick — which should we adopt?
Kirkpatrick’s levels for routine measurement; Phillips’s discipline for the programs that justify monetisation. They’re complements, not rivals — and both need measured inputs to mean anything.
What does the platform contribute to level five specifically?
The evidence layer: measured capability, cohort comparisons, cost data — the inputs that turn ROI arithmetic from assumption-stacking into calculation.
See it on your own content.
Bring one course. We’ll show you the retention curve your current training leaves behind — and what scheduled review does to it.
- 30 minutes, on your calendar — pick a slot here
- Run on your own content wherever possible, not a canned deck
- You see the dashboards, the learner surface and the evidence exports
- No commitment — and pilot data stays yours either way