Learning Analytics Platform Development

Learning analytics for decisions completion rates cannot answer.

We build a focused analytics layer across LMS or LRS activity, assessments, learner identity, and one approved business measure. Delivery covers event and score definitions, HRIS joins, cohorts, privacy, evidence limits, one decision view, review, monitoring, and analyst handover.

Bring the problem, the current workflow, or the existing code. We reply with a practical next step within one business day.

Evidence and scope

3,500+

Published learning-platform use

Daily active users reported for EMS Connect during its active use period.

30 min

Published engagement

Average daily session reported for that training platform, not proof of learning impact.

Starts at $20K

Focused analytics release

One learning source, one identity join, one cohort decision, and governed measures.

Evidence · planning contextSee the work

The brief

Start with what is not working.

Good software decisions begin with the constraint, not a list of features or a preferred technology.

01

Can L&D show who completed a module but not who retained the knowledge, needs follow-up, or improved on the agreed job measure?

02

Do LMS, assessment, HRIS, and operational records use different identities, periods, and definitions that make an impact claim hard to defend?

Plain answer

A learning analytics platform joins learning activity and assessment evidence with approved learner and business data so L&D can decide where content, coaching, or investment should change. RaftLabs scopes one learning source, identity join, cohort question, and decision first. A focused release starts at $20,000 and usually takes 8 to 12 weeks.

The programme reached 92 percent completion. The follow-up decision was still a guess.

One cohort took the assessment before a content update, another after it, and role changes reached the LMS a month late. Managers could see completions but not which knowledge gap required coaching or whether the reported performance difference used comparable groups. The analytics layer needed identity history, content versions, and an honest evidence label.

Published learning-platform evidence

3,500+
daily active users
EMS Connect during active use
30 min
average daily session
Engagement, not a retention measure
25%
reported training-time reduction
Published project outcome, method not public

The EMS Connect case covers a training platform with content, assessments, manager reporting, and 5,000-plus field employees. The public case reports use and training-time outcomes, but it does not publish a controlled causal study or a separate learning-analytics product. New claims need acceptance against your cohort, measures, and evidence design.

Build learning analytics when a specific L&D decision requires evidence beyond the LMS report.

Use native reporting when it already answers the question. Separate learning evidence from high-stakes employment decisions unless governance and validation warrant the connection.

A fit
01

The organisation has learning activity and assessment data plus a clear decision about content, coaching, pathways, capability, or programme investment.

02

L&D, HR, data, privacy, and business owners can define learner identity, cohort, measure, interpretation, access, and review.

03

The problem requires versioned learning events, repeated assessment, stable joins, evidence limits, or governed downstream measures.

Not a fit
01

The only requirement is a standard completion, enrolment, overdue, or score report the LMS already supports.

02

There is no stable learner identifier, representative history, approved measure, decision owner, or privacy path.

03

The project assumes activity proves competence, correlation proves cause, or a score should automatically determine employment.

Choose the layer by the learning decision

NeedBest fitBoundary
Enrolment, delivery, completion, certification, and standard reportsLMS or LXPLearning operations and native activity reporting
Assessment, retention, cohort, intervention, or programme evidenceLearning analyticsLearning events, identity, versions, measures, evidence limits, and decision view
Shared measures across HR, finance, sales, and operationsBusiness intelligenceEnterprise semantic model, role reporting, distribution, and adoption
Verified capabilities, gaps, role profiles, and workforce planningSkills managementSkills taxonomy, evidence, validation, proficiency, roles, and governance

Scope

What belongs in one learning analytics decision path

  • 01
    Learning event and assessment contract
    Define enrolment, assignment, start, completion, attempt, score, question, pathway, session, content version, and assessment timing. Preserve what each source can and cannot show.
  • 02
    Stable learner and cohort model
    Join LMS, LRS, HRIS, and approved operational records without erasing role history. Define cohort inclusion, transfers, contractors, leavers, small groups, and effective dates.
  • 03
    Governed learning measures
    Separate activity, assessment, retention, behaviour, and business results. Record formula, denominator, window, exclusion, version, source freshness, uncertainty, and the owner who approves interpretation.
  • 04
    Decision and review view
    Show where content, coaching, assessment, or investment may need attention. Let a reviewer inspect source evidence, compare appropriate cohorts, annotate context, and avoid unsupported causal language.
  • 05
    Privacy and operations
    Minimise employee data, restrict individual views, protect exports, trace use, monitor source and join failures, version changes, capture corrections, and hand the system to named L&D and data owners.

How it works

From learning question to defensible cohort evidence

  1. Phase 1
    01

    Define learner decision and evidence

    Choose one programme or pathway, audience, decision, learning events, assessments, identity source, cohort, approved business measure, costly interpretation errors, privacy limits, owners, and acceptance measures.

  2. Phase 2
    02

    Reconcile learning and people data

    Profile completions, attempts, scores, event statements, content versions, assignments, employee identifiers, role history, missing records, timing, consent, retention, exports, and reviewer agreement.

  3. Phase 3
    03

    Build the focused analytics layer

    Implement ingestion, stable joins, governed measures, cohort and time-window logic, assessment views, evidence labels, role access, correction, one decision view, monitoring, and recovery.

  4. Phase 4
    04

    Validate interpretation and hand over

    Reproduce historical periods, reconcile source totals, test identity and content changes, review confounders and small cohorts, document definitions and limits, train users, and release.

Risk

What the evidence contract must settle

Completion versus learning
Treat opens and completions as activity. Use an appropriate assessment and follow-up window before claiming knowledge or behaviour changed.
Identity and role history
Preserve effective dates, transfers, managers, contractors, and content assignments. A current HRIS snapshot can mislabel the historical cohort.
Correlation and causation
Name confounders, selection, sample size, time lag, and comparison method. Use experiments when a causal investment claim matters.
Employee consequence
Keep weak or exploratory analytics away from automated performance, promotion, discipline, or termination decisions. Require accountable review and approved policy.

Scope and price

A focused learning analytics release starts at $20,000.

Start with one learning source, one identity join, one programme, one cohort question, approved measures, role access, and a named L&D decision owner.

Use native LMS reporting for standard activity questions. A broader learning-data platform commonly reaches $45,000 to $100,000 after the first decision path proves useful.

Starting investment

Starts at $20,000

A focused release usually takes 8 to 12 weeks. More sources, countries, skills models, experiments, historical reconciliation, or individual-level use increase scope.

Activity is not labelled impact

Completion, assessment, retention, behaviour, and business results remain separate measures with visible evidence limits.

Cohorts keep their context

Identity, role, content version, timing, inclusion, and known confounders travel with each reported comparison.

Related learning and analytics services

Learning analytics questions

It combines learning activity, assessment, learner identity, content version, and selected business or performance data to support a defined L&D decision. Useful outputs can include assessment gaps, retention checks, cohort differences, content abandonment, follow-up queues, or programme evidence. It should show definitions and limits, not only attractive completion charts.

LMS reporting usually covers enrolment, completion, score, and course activity inside one platform. Learning analytics can join event-level activity, repeated assessments, HRIS history, role or cohort context, and an approved downstream measure. If the LMS report already answers the decision with trusted definitions, a custom layer is unnecessary.

Usually not from observational correlation alone. Role, manager, tenure, season, selection, incentives, and other changes can affect the same result. We label associations honestly, use time windows and matched cohorts where appropriate, and recommend a controlled rollout or experiment when the organisation needs stronger causal evidence.

Collect only fields needed for the approved decision, separate aggregate and individual views, restrict access by role, protect exports, log use, define retention, and avoid turning a weak model into an employment decision. The employer owns its lawful basis, consultation, policy, and jurisdiction-specific obligations with its advisers.

A first release starts at $20,000 and usually takes 8 to 12 weeks. It covers one learning source, one learner-identity join, one programme or pathway, one cohort decision, governed measures, role access, monitoring, and handover. More systems, countries, skills models, experiments, or individual-level decisions increase scope.

Work with us

Bring the learning decision, not a list of dashboard tiles.

Share the programme, LMS or LRS, assessments, HRIS join, content versions, current reports, cohort question, approved business measure, privacy owner, and the decision L&D needs to make.

  • Scope and cost agreed before work starts. No surprises. No obligation.
  • Working prototype within 3 weeks of kickoff.
  • Pay by milestone. You see progress before each invoice.
  • 60-day post-launch warranty. Bug fixes, UI tweaks, and deployment support. No retainer.
  • All conversations are NDA-protected.