Job title
Training Catalogues Should Not Become Job Architectures
A corporate catalogue is organised around suppliers, formats and budgets. A job architecture is organised around work, scope and accountability; the two should meet, not merge.
What the wrapper says
A corporate catalogue organises learning products and administration, while a job architecture organises roles, scope and accountability; course-to-role mappings are the point of contact.
The all-clear
Structured catalogues and job families are useful operational controls for learning delivery, procurement, compensation, reporting and career administration.
What is durable
The capability needed to achieve a role's intended work outcomes.
The catalogue has a category called Leadership Essentials.
The job architecture has a family called Product Management.
By Friday afternoon, somebody has made the category mandatory for the family, mapped the completion badge to “product leadership,” and produced a dashboard showing a pleasantly rectangular readiness score.
The deck is named FY26_CAPABILITY_COVERAGE_FINAL_v6_USE_THIS.xlsx.
Nobody is doing anything unreasonable.
Procurement needs supplier groupings.
LMS administrators need enrolment rules.
Finance needs a chargeback.
A manager needs a development plan that fits on one screen.
The trouble is only that all of those reasonable things have been allowed to answer a question none of them can observe.
A training catalogue is an inventory of interventions. A job architecture is a model of work.
They should meet at an explicit mapping.
They should not merge into one highly confident noun.
The catalogue has a proper job
A catalogue lets you buy content, plan cohorts, attach prerequisites, schedule facilitators, manage seat time and retain a transcript.
That is real operational work, and it needs real governance.
A proposal above a hundred thousand in direct budget, or mandatory for more than five hundred people, goes to a formal business case.
A course drawing less than 15 percent annual enrolment can be retired in the annual curriculum audit.
A cancellation may be free more than 21 business days before the course, cost half between 11 and 20 days, and forfeit everything inside 10.
That is exactly how a catalogue avoids becoming an expensive attic full of duplicate content and unloved licences.
Good governance of learning supply.
It is not evidence that attendance produces the work outcomes of every role assigned to a category.
The role has a different job
A job profile carries a family, a management level, a default pay grade and a job description.
A family connects a set of profiles across seniority tiers.
That architecture makes a role legible for compensation, reporting and career administration.
Also useful work.
And the family marks meaningful differences in outcome.
Product management answers what will be built and why.
Technical programme management owns cross-functional execution, dependencies and timing.
Their staffing ratios differ by roughly a factor of five, which is a clue that they are different accountabilities rather than adjacent nouns.
An assigned catalogue can support either population.
It cannot infer which scope of accountability a particular role carries.
The people in the first group decide direction.
The people in the second group make a complicated initiative land without acquiring a small private collection of untracked dependencies.
Both might complete the same course before lunch.
Completion did not settle the question.
The completion record knows what it knows
Your LMS can know that a person was enrolled, because an HRIS job code triggered an audience rule.
It can know the course package, the completion time, the attendance, the assessment attempt and the badge.
It can preserve immutable, time-stamped audit logs, and retain transcripts, certification records and compliance audits for the full retention period.
Valuable provenance.
Keep it.
But none of those fields observes product judgement.
Or incident response.
Or an accounting close.
Or a customer renewal.
Or a manager’s ability to apply a framework when the meeting has gone from manageable to archaeological.
A reaction score records reaction.
A learning-assessment pass records an assessment result.
Both are useful signals about the course.
They are not role-ready signals because your dashboard placed them next to the job title.
The bridge is behavioural: an audit at 90 days, across a randomized sample of participants and their supervisors, asking whether the capability appears in weekly operational planning.
That is a serious question.
Did the intervention transfer into work?
It is still different from the next question.
Is this person ready for the role, at this level and scope?
Answering that may use transfer evidence, observed work, a role assessment or manager validation.
It deserves to be reviewed as its own decision.
A course title needs provenance
There is another small administrative trap here.
A catalogue presents a course title as though it were a stable thing. Leadership Essentials. Data Foundations. Managing Change Without Making Everyone Hate Tuesday.
It is not stable by itself.
External courseware moves through acceptance gates: a design document, an alpha prototype, a gold master, a post-pilot acceptance.
The gold master is a specific release, following a specific pilot and a formal sign-off.
The course title is what humans see.
The versioned package is what was actually delivered.
That distinction gets awkward the moment somebody asks whether an earlier completion satisfies a newly assigned requirement.
The title may be identical.
The supplier may have changed the scenarios, the assessment, the prerequisite, the modality, or the content that made the course useful for that role in the first place.
So the learning record keeps its supplier, course title, version or release, format, assignment basis, assessment result and mapped role outcome.
A completion badge without that provenance is a receipt with the purchase line removed.
The same restraint applies to your supplier controls.
An LMS availability commitment is an important contractual promise.
It does not establish that a manager who completed the course can exercise judgement under pressure.
Both systems are working exactly as designed.
They simply make different claims.
The catalogue is not the villain here
None of this argues against structured catalogues.
A well-run one is how a large organisation gets consistent learning in front of thousands of people without every business unit commissioning its own private course from a supplier nobody has checked.
Retirement rules stop it bloating.
Cancellation terms make cohort planning honest.
Business-case thresholds stop a well-meaning category becoming a mandatory expense for five thousand people because somebody liked the pilot.
Keep all of it.
The failure is not the catalogue.
The failure is the moment a browse filter becomes a definition of readiness.
AI will make the category look like a capability
Job families emerged to organise an employer’s work.
Catalogues emerged to organise learning products, suppliers and delivery.
HRIS job codes make tidy enrolment rules.
LMS categories make tidy reporting.
Then a recommendation engine sees both fields and asks the obvious question.
Which employees are ready for which jobs?
It maps Leadership Essentials to Product Management, sees the badge, and reports a clean readiness score.
The score is fictional, but exceptionally well formatted.
This is not a model failure.
The system was handed a job label, a catalogue label and a completion event, then invited to infer capability, scope and accountability.
It acted on the objects you gave it.
The repair is boring, which is usually encouraging.
Store a canonical role, with its expected outcomes, family, level and scope.
Store the course supplier, version, format, prerequisites, assessment method and completion separately.
Then put an explicit intervention mapping between the course and the outcome it is meant to support.
Preserve transfer evidence separately from completion.
And require a role-readiness review to inspect the evidence appropriate to the work, rather than accepting a supplier category as proof.
The model can recommend a course.
It can identify that a learning record exists.
It can surface a missing prerequisite.
It should not promote a completion badge into a claim that somebody can do the job.
Let the systems meet, then leave them distinct
Training can be the right response to a defined capability gap.
A catalogue can make that response available at scale.
A job architecture can state which outcomes, scope and accountability the role requires.
A completion record can show what intervention occurred.
A transfer measure can show whether it entered the work.
Related objects.
Different claims.
The durable object is capability: the ability to achieve the intended work outcome under the role’s conditions.
The catalogue category and the job-family label are wrappers.
One helps you buy and administer learning.
The other helps you organise employment.
The useful system maps them.
The dangerous one lets a catalogue browse filter become your definition of readiness, then calls the result workforce intelligence because the bar chart is green.
What the record establishes
- Write capability-based role outcomes before selecting learning content.
- Map courses to roles as learning interventions rather than proof of role readiness.
- Retain supplier, course-version, assessment and completion metadata with every learning record.
- Measure behavioural transfer after learning rather than relying on completion or satisfaction alone.
- Review role-readiness decisions separately from course-completion decisions.
- Keep job-family, level and role-scope definitions distinct from catalogue categories and supplier labels.
Asked in the review
- What is the difference between a training catalogue and a job architecture?
- A training catalogue organises learning products for sourcing, licensing, scheduling, budgets, enrolment and completion records. A job architecture organises roles for work design, scope, accountability, career levels, pay grades and reporting. The systems meet when a course is mapped to a role outcome or capability. A catalogue category remains an administrative label; it does not become the definition of the role or proof that its work can be performed.
- What should be defined before a course is assigned to a role?
- The employer defines the role's intended work outcomes, scope, accountability and required capability before selecting content. A Job Profile can hold a job-profile identifier, family identifier, family-group identifier, management level, default pay-grade identifier and job-description object. That structure identifies the work being administered. The course is then selected as an intervention for a stated gap rather than as a substitute definition of the role.
- Does completing the assigned catalogue make someone ready for the job?
- No. Completion records that a person finished an assigned learning product; it does not observe role performance. A course can assess learning through scenarios or simulations, but role readiness concerns whether the person achieves the role's intended outcomes at the required scope and accountability. The employer keeps course completion, learning assessment, behavioural transfer and role-readiness decisions as separate records so that one administrative event does not become a fictional capability claim.
- Why can a generic catalogue course not define an internal role?
- Commercial off-the-shelf content can provide broad subject coverage and rapid deployment, but it cannot teach proprietary software architectures, internal compliance workflows or company-specific management routines. The corpus describes annual catalogue licensing at $120 to $200 per user, while bespoke interactive courseware is quoted at $25,000 to $60,000 per finished instructional seat hour. The cost and format of content therefore describe an intervention choice, not the full work definition of an internal role.
- How should a course be connected to a job role?
- A course is connected to a role through an explicit intervention mapping. The mapping identifies the role outcome or capability addressed, the expected learner population, the course supplier and version, the learning format, prerequisite capability, assessment method and review status. The connection can state that the course supports development or remediation. It does not state that the course title, supplier category or completion badge is identical to the role capability.
- What learning metadata needs to stay with a completion record?
- A learning record retains the supplier, course title, course version or Gold Master release, delivery format, assignment basis, completion date, assessment attempts and result, along with the mapped role outcome or capability. LMS governance also requires immutable, time-stamped audit logs for completions, test attempts and administrative alterations. That provenance allows later reviewers to distinguish a changed course, a changed role definition and an unchanged historical completion.
- Why is an 80 percent course score not evidence of role readiness?
- An 80 percent passing score can earn a formal competency badge in the LMS and trigger remedial loops after a failed attempt. It records performance on the course's end-of-module scenarios or simulated cases. It does not establish that the learner applies the capability in the role's live conditions, with its scope and accountability. A role-readiness decision therefore needs evidence beyond the LMS badge, including observed work or a separate assessment of the intended outcome.
- What does behavioural transfer measure after training?
- Behavioural transfer measures whether learning is applied in work after the course. The corpus uses Kirkpatrick Level 3 for behavioural application at 90 days: three months after training, the L&D team audits a randomized 15 percent sample of participants and immediate supervisors for use of the structured frameworks in weekly operational planning. This evidence tests the bridge between learning and work; a reaction score or completion status does not test it.
- Why should job families remain separate from catalogue categories?
- Job families group roles around functional work, career administration, compensation and organisational accountability. Catalogue categories group learning products around supplier offerings, formats, procurement and scheduling. Software Engineering, Systems Infrastructure and Data Science have different primary deliverables even though each can require programming capability. Treating a course category as a family definition can erase those distinctions and make a supplier's product structure decide what the employer believes a role is.
- How should an employer decide whether training is the right response?
- The employer uses a Training Needs Assessment before prescribing learning. The corpus describes a 30-to-60-day discovery sprint across 8 to 12 affected business units, reviewing process logs, customer measures and production metrics. The diagnostic distinguishes knowledge deficits from broken tooling or flawed incentives; it reports skills gaps as 15 percent to 25 percent of operational error variance, with the remainder attributed to workflow bottlenecks. Training is then one possible response, not the default answer.
- What makes a role-to-course recommendation safe for an AI system?
- An AI system retains the canonical role, job family, level, expected outcome, course supplier, course version, delivery and assessment metadata, and the intervention mapping between them. It reports completion as completion and preserves transfer evidence separately. A recommendation engine can use those records to suggest learning, but it cannot safely report a readiness score merely because a catalogue label and a job label appear to match. Human review remains necessary for the readiness decision.
- Why do catalogues and job families still need governance?
- Catalogues need governance because they control licensing, attendance, supplier commitments, curriculum quality and learning records. Job families need governance because they control profiles, levels, pay grades, survey matches and accountable ownership. The corpus describes annual curriculum audits that retire low-enrolment courseware and a job-family parent record that commonly connects to 6 to 12 job profiles. Separate controls keep each system useful without allowing either to impersonate the other.