"Did the training work?" is the question every L&D head is asked and few can answer, because the evidence sits in four systems that do not speak to each other. Learning analytics joins them, while the programme is still running.
The problem
Attendance lives in the LMS, feedback in a form, assessment in a spreadsheet, and the only evidence of on-the-job change is a manager's recollection. By the time anyone assembles a report, the programme has finished, the cohort has dispersed, and the finding — that a third of the region never practised — is useless.
How it works
Every event in an Indus programme writes to one place: attendance, each practice attempt and its score, assessment results, manager observations, completed on-the-job tasks. The dashboard presents them in four layers — participation, capability, application, and business effect — and cuts them by project, city, cohort, designation and supervisor, because those are the dimensions along which anyone can actually act.
The AI layer does two jobs: it scores the open-ended material so capability data exists at all, and it flags patterns worth a human look — a supervisor whose entire team stopped practising in week three, a scenario where every cohort fails the same beat, a city where scores are excellent and business metrics have not moved.
What changes at work
- Interventions happen during the programme: the region that has not practised is called in week two, not named in a post-mortem.
- L&D walks into the business review with evidence at cohort level instead of satisfaction scores.
- Bad programme design is caught early — if every cohort fails the same beat, the content is wrong, not the learners.
- Budget conversations change character, because the discussion is about a business metric with training data beside it.
How we prove it
By being falsifiable. We agree the metric before the programme starts, take a baseline, and report movement with the caveats intact — seasonality, sample size, everything else that changed in the same quarter. Training is one of several influences on a business number and we present it that way; overclaiming is how measurement programmes lose their audience.
What it cannot do
It cannot prove causation on its own. A dashboard showing capability up and sales up is consistent with training having worked, and also with a good quarter. Where a client wants more certainty we run a staggered rollout — one region trained first, the other as a comparison — which is a modest amount of extra coordination and a great deal more confidence in the answer.
Frequently asked
What do the dashboards actually show?
Four layers: participation (who attended, who practised, how often), capability (assessment and practice scores, by person and cohort), application (manager observation and on-the-job tasks completed), and business effect where the client shares the metric — ramp time, discount given, first-call resolution, safety incidents.
Do you need access to our business data?
Not to start. The first three layers come entirely from the programme. The fourth needs a single metric from you, at cohort level — not individual records — and it is what turns a training report into a business conversation.
How is this different from LMS reporting?
An LMS reports completion, which tells you almost nothing. These dashboards report capability and application, and are cut by project, city, cohort and supervisor — the dimensions along which action is actually taken.
Can our team access it directly?
Yes. L&D and HR get the full view, project managers see their own teams, and clients on outsourced programmes get a scoped view of their own population. Access is role-based and audited.
What if the numbers are bad?
Then we show them. We have ended cohorts early and rewritten programmes mid-flight because the data said the design was wrong. A dashboard that only ever produces good news is a marketing asset, not a measurement system.