Impact you can defend.

Most platforms can tell you who finished. Almost none can tell you who learned, who improved, who applied it at work - and whether any of it moved a business number. ThinkPair reports learning as a stack of separate evidence. Each layer earned, each layer reported on its own. We never convert one into another.

Request a pilot plan

Evidence, layer by layer

Figures illustrative
5Business impactOnly with clean data
4Transfer70% applying live
3Retention78% at 4 weeks
2Skill application3.1 → 4.2 / 5
1Learning gain61 → 84 · +23
Everything above this line has to be earned.
-Engagement - the floor, not the proof94% · 38 min avg

The evidence stack

An engagement baseline, plus five earned layers. Each answers a different question, is measured differently, and is reported separately. Click a layer to see how it works.

Only with clean data

Where a reliable KPI exists, we review it: ramp-up time, conversion, win rate, audit findings, early attrition - whatever the programme was built to move.

The discipline. This is the layer everyone oversells. We report it only when the underlying data is clean enough to support the claim. When it isn’t, we say so - and we don’t manufacture an ROI number to fill the gap.

70% applying live

We look for the behaviour in the workplace - manager check-ins, workplace tasks and observed application. Self-report alone is weak, so we triangulate it with observation rather than trusting a survey on its own.

The discipline. Transfer is the generalisation of learning to the job and its maintenance over time (Baldwin & Ford). We report it as evidence, not assumption.

78% at 4 weeks

Delayed checks 2-6 weeks out - spaced retrieval, not a re-watch. This is the layer most platforms quietly skip.

The discipline. Retrieval practice and spacing are what fight the forgetting curve. Passive review doesn’t hold; testing and distributed practice do.

3.1 → 4.2 / 5

Knowledge and skill are different outcomes, so we measure them differently. Learners work through scenarios, simulations and AI-facilitated role-plays - and we score the performance.

The discipline. Cognitive, skill-based and affective outcomes are distinct (Kraiger, Ford & Salas). A quiz score doesn’t tell you whether someone can do the thing.

61 → 84 · +23

We compare a pre-assessment against a post-assessment and report the delta, not the score.

The discipline. A pre/post design shows change around an intervention. It shows learning - it is not, on its own, proof of business return, and we never present it as such.

Engagement - the floor, not the proof

Completion, time on task, module progress. Where everyone else stops reporting.

Everything above this line has to be earned.

Figures illustrative.

Honest reporting by design

Engagement as participation - never as proof.

Learning gain as pre/post improvement - never as ROI.

Skill application as scored performance in scenarios.

Retention as delayed, spaced checks.

Transfer as workplace evidence, triangulated with observation.

Business impact only when reliable KPI data supports it.

We don’t turn completion into competence. We don’t dress learning gain up as return. We don’t claim a business result the data can’t defend. That restraint is what makes the evidence hold up in front of a CFO, an auditor, or a sceptical board.

Most platforms stop at completion. We don’t.

The same programme, measured two ways - a single tick, or six layers of evidence you can defend.

What’s reported

Most platforms: A single completion or satisfaction score.

ThinkPair: Six evidence layers, each reported on its own.

Knowledge vs skill

Most platforms: A quiz score stands in for competence.

ThinkPair: Skill scored separately in scenarios and role-play.

Weeks later

Most platforms: No check once the session ends.

ThinkPair: Spaced retention checks, 2-6 weeks out.

On the job

Most platforms: Transfer assumed from a completion tick.

ThinkPair: Transfer evidenced by manager check-ins and observation.

Business impact

Most platforms: An ROI number invented to impress.

ThinkPair: Reported only when the KPI data can defend it.

See it in action

The stack is easiest to understand when you follow one person through it. Names and figures are illustrative; the reporting discipline is the point.

Onboarding

Maya’s first 90 days

From a first-day baseline to a defensible impact number.

  1. 1

    Day 1: pre-assessment on processes, tools and role expectations. Baseline 41/100 - a starting line, not a verdict.

  2. 2

    First 30 days: cohort hits 96% completion, 38 min average engaged time. The floor, not a result.

  3. 3

    Week 4: post-assessment 79/100 · +38 gain. Understanding improved - not yet proof she can do the job.

  4. 4

    Weeks 2-4: real tasks - system navigation, a process request, a first stakeholder update. Applied score 4.0/5.

  5. 5

    Day 45: spaced check. 82% retained on critical processes.

  6. 6

    Days 60-90: manager confirms it in daily work, triangulated with two observed tasks. Now it’s transfer.

  7. 7

    Day 90+: HRIS data is clean, so we report it - time-to-productivity 74 → 58 days across the cohort.

    Data verified

Sales enablement

Daniel’s new product line

From a soft win rate to evidence a sales leader can trust.

  1. 1

    Pre-assessment on the new product and value story. Baseline 61/100.

  2. 2

    Enablement journey completed with 94% of the sales floor. He turned up - that’s all this proves.

  3. 3

    Post-assessment 84/100 · +23 gain on product knowledge and value messaging.

  4. 4

    AI-facilitated role-play on the objections this product actually attracts: 3.1 → 4.2 / 5.

  5. 5

    Four weeks later: 78% retained on the core messaging - still there when a prospect pushes back.

  6. 6

    Manager confirms the new framing in live conversations, triangulated with call reviews and CRM notes. Transfer.

  7. 7

    CRM data holds up for this pilot, so win rate, conversion and ramp-up are reviewed against a clean baseline. Where it wouldn’t, we claim nothing.

    Data verified

The same stack flexes to compliance, risk & EHS (audit findings, repeat incidents, policy breaches) and leadership development (manager-effectiveness, engagement, team performance). Business impact appears only when the data can carry it.

Academic foundations - the evidence base behind each layer

  1. Alliger & Janak (1989), Personnel Psychology 42(2).

    Why it matters: completion, learning, behaviour and results shouldn’t be collapsed into one number.

  2. Kraiger, Ford & Salas (1993), Journal of Applied Psychology 78(2).

    Why it matters: knowledge, skill application and affective outcomes are distinct.

  3. Dimitrov & Rumrill (2003), Work 20(2).

    Why it matters: pre/post measurement is learning-gain evidence, not automatic ROI.

  4. Roediger & Karpicke (2006), Psychological Science 17(3).

    Why it matters: the basis for retention checks and retrieval practice.

  5. Cepeda et al. (2006), Psychological Bulletin 132(3).

    Why it matters: the basis for spaced reinforcement and delayed checks.

  6. Baldwin & Ford (1988), Personnel Psychology 41(1).

    Why it matters: transfer as generalisation and maintenance on the job.

  7. Blume et al. (2010), Journal of Management 36(4).

    Why it matters: transfer depends on the learner, the environment and the design.

  8. Aguinis & Kraiger (2009), Annual Review of Psychology 60.

    Why it matters: business impact, discussed cautiously and only with appropriate data.

Learning investment is easy to make. Impact is harder to prove. See how far your learning actually travels.

See how it’s reportedRequest a pilot plan

Start with a two-week diagnostic.

Request a pilot and we’ll send a tailored plan within 48 hours - success criteria agreed in writing, no commitment.

  • A tailored pilot map in 48 hours
  • Success criteria agreed in writing
  • No commitment to start

We’ll get back to you as soon as possible. · Privacy

Methodology - ThinkPair