A Five-Stage Diagnostic Framework
Imagine walking into a doctor’s office. Before you describe your symptoms, the doctor reaches for a prescription pad. “I know exactly what you need.” No examination. No medical history. No attempt to understand what brought you there. You would probably leave. In medicine, prescribing treatment before making a diagnosis would be considered poor professional practice. The physician begins with questions, observations, and evidence. The prescription comes later. Learning and Development (L&D) should operate on the same principle.
Yet many learning projects begin with the proposed treatment already decided.
“We need customer service training.”
“We need leadership training.”
“We need an eLearning module.”
“We need to retrain everyone before the next audit.”
These requests may sound like learning requirements. Usually, they are not. They are business problems expressed through the most familiar solution available to the person making the request.
The problem is not that managers ask for training. They are not expected to diagnose learning and performance issues. The problem begins when L&D accepts the requested solution without testing the assumptions behind it. The discussion quickly moves to learning objectives, course length, delivery method, budget, and launch date. The prescription has been written. The diagnosis never happened.
Why Diagnosis Matters
Training can improve performance when people lack relevant knowledge or skill and when the workplace allows them to use what they learn. The second condition is frequently underestimated.
A meta-analysis examining work-environment support found that peer, supervisor, and organizational support all had moderate, positive relationships with the transfer and continued use of trained knowledge and skills. In other words, the environment surrounding the learner materially affects whether training appears in workplace behaviour.
Manager support is also more complex than asking supervisors to “encourage participation.” A systematic review of 99 studies identified 24 distinct supervisor behaviours and attitudes that can support training transfer. These include setting expectations, providing opportunities to apply skills, discussing learning, offering feedback, and removing obstacles.
Motivation matters as well. A meta-analysis covering 148 studies, 197 effect sizes and 31,718 participants examined the relationship between motivation and the transfer of professional training. The scale of that evidence base reinforces a basic point: attending training does not automatically produce application. Employees must see the material as relevant, believe they can use it and have reasons to apply it.
The implication for L&D is straightforward. A well-designed course can still fail when managers do not reinforce it, employees lack opportunities to apply it, workflows discourage it, or incentives reward something else. That is why the training request cannot be the starting point.
A training request is an untested hypothesis When a manager says, “We need customer service training,” they are offering a theory: Customer service performance is poor because employees lack knowledge or skill. That theory may be correct.
It may also be wrong. Customer satisfaction could be falling because response times have increased. Response times may have increased because the organization reduced staffing or introduced a slower system. Employees may be following policies that frustrate customers. Managers may be measuring call duration while senior leaders say they want better conversations. None of those causes would be fixed by a course on empathy or communication.
The appropriate response is not to reject the request. It is to investigate the problem behind it. L&D needs a repeatable method for doing that.
The Performance Diagnosis Ladder
I use a five-stage diagnostic sequence before recommending a learning intervention:
Business outcome
Required behavior
Capability evidence
Performance conditions
Intervention decision
Each stage tests a different assumption. The order matters. Moving directly to the fifth stage usually means the solution is being selected before the problem is understood.
Stage 1: Business Outcome
Begin with: What business outcome are we trying to improve? This moves the conversation away from topics and deliverables. “Customer service” is not an outcome. “Reduce repeat customer complaints by 15%” is an outcome. “Leadership development” is not an outcome. “Increase the frequency and quality of performance conversations” is closer to one.
A useful outcome should identify what needs to change and why it matters. It may relate to quality, speed, safety, cost, retention, customer experience, compliance or another business measure. Without an outcome, L&D cannot determine whether the intervention made a difference. It can only report activity.
Stage 2: Required Behavior
Next ask: What must people do differently for that outcome to improve? This question forces the requestor to describe observable performance. Suppose customer complaints have increased. What are employees doing today? What should they do instead?
Perhaps strong performers verify the customer’s issue before proposing a solution. Perhaps they explain the next step clearly. Perhaps they escalate a specific category of problem sooner. These are behaviours that can be observed, practised, and measured. “Communicate better” is not sufficiently specific. “Confirm the customer’s issue in their own words before offering a resolution” is specific enough to design around. When the required behaviour cannot be described, the problem is not ready for a training solution.
Stage 3: Capability Evidence
Now ask: What evidence shows that people cannot perform the behavior? This is where many training requests begin to weaken. Has anyone observed employees doing the work? Have high and low performers been compared? Have employees been asked what prevents them from performing? Can they demonstrate the behaviour under realistic conditions? If employees can perform correctly when asked but do not perform consistently on the job, the problem may not be capability.
A knowledge or skill gap exists when people do not know what to do or cannot do it to the required standard. A performance gap can exist even when they can. Confusing these two conditions leads to unnecessary training.
Stage 4: Performance Conditions
The fourth stage examines the environment: What helps or prevents people from performing correctly? This is often the most important stage. Performance depends on more than individual capability. Employees also need clear expectations, usable tools, supportive workflows, relevant information, time, authority, feedback and consequences that align with the desired behaviour. A practical review should examine six conditions:
Expectations
Do people know what good performance looks like?
Tools and information
Can they access what they need while working?
Workflow
Does the process make the desired behaviour realistic?
Manager reinforcement
Do managers observe, coach and reinforce the behaviour?
Incentives and consequences
What is actually rewarded or discouraged?
Opportunity
Do employees have a chance to practise and apply the skill?
Research into workplace eLearning transfer continues to treat these environmental conditions as central because learning outcomes and workplace application are not the same measure. A recent systematic review of workplace eLearning literature examined studies published from 2000 through 2024 and highlighted the continuing need to understand and measure transfer beyond course participation and immediate learning results. This stage frequently reveals that training is only part of the answer.
Stage 5: Intervention Decision
Only now should L&D ask: What role, if any, should learning play? There are four common answers.
Training is the primary intervention.
People lack essential knowledge or skill, and the environment supports application.
Training is one part of the intervention.
Capability must improve, but managers, systems or processes must also change.
Performance support is more appropriate.
People do not need to memorize the information. They need access to it during the work.
Training is not the solution.
The problem is caused primarily by processes, tools, workload, incentives, leadership or another environmental condition.
The quality of the recommendation depends on the diagnosis, not on how much content L&D produces.
A Composite Example: The Training That Was Not Needed
Consider a composite case based on a pattern common in service organizations. A business unit requested refresher training because employees were entering incorrect information into a customer-management system. The error rate had increased after a software update, and the initial assumption was that employees had forgotten the procedure. The obvious response was to rebuild the system-training module. The diagnosis produced a different answer.
Employees could explain the correct procedure and complete it accurately in a controlled demonstration. The capability existed. The errors occurred during high-volume periods. A field that had previously been optional had become mandatory, but the system gave no clear explanation of what information belonged there. Employees created workarounds to complete transactions quickly. Managers knew about the workarounds but tolerated them because they reduced wait times.
The proposed training would have reminded people of a procedure they already understood. The eventual intervention was smaller:
The field label was rewritten.
A short in-system prompt was added.
Managers were given a consistent escalation process.
Error reports were reviewed weekly for one month.
A five-minute update explained the system change.
The organization did not need a new course. It needed a better performance environment.
This example also illustrates an important financial point. The cheapest course is still wasteful when it addresses the wrong cause. A short diagnostic can prevent weeks of unnecessary development and reduce the operational cost of leaving the real problem unresolved.
What Diagnostic Practice Changes For L&D
When L&D diagnoses before designing, its conversations with stakeholders change. Instead of asking, “How many modules do you need?” it asks, “What must improve?” Instead of asking, “When should the course launch?” it asks, “What evidence supports the proposed solution?” Instead of promising that training will solve the problem, it explains what training can influence and what requires action elsewhere. This does not make L&D less responsive. It makes the function more useful.
The strongest minority objection is that stakeholders do not always want a diagnostic process. They may have a fixed deadline, a compliance requirement, or a senior leader who has already decided that training is necessary. That constraint is real.
Diagnosis does not always require a large research project. It can begin with a 30-minute conversation, 2 employee interviews, a review of error data, and 1 workplace observation. The depth of analysis should reflect the cost and risk of being wrong. The important point is that someone tests the central assumption.
An Operational Standard For Accepting Training Work
Before L&D commits to development, it should be able to complete a one-page diagnosis containing:
The business outcome
The performance gap
The required behaviour
Evidence of the capability gap
Environmental barriers
Manager responsibilities
The proposed intervention
Measures of workplace change
This document creates an explicit performance hypothesis: If we strengthen this capability, while addressing these workplace conditions, we expect this behaviour and business measure to change.That statement can be tested. “We delivered the course” cannot.
The Next Credibility Test For L&D
Content production is becoming easier. Generative AI can already draft objectives, scripts, scenarios, assessments and supporting materials in minutes. This will make diagnostic capability more valuable, not less. When almost anyone can produce a plausible course, L&D’s credibility cannot depend on its ability to generate content. It will depend on its ability to determine whether that content should exist.
My prediction is that the gap between order-taking L&D teams and performance-focused teams will widen. Order takers will use technology to produce more deliverables. Performance consultants will use it to reduce development effort and invest more time in evidence gathering, workflow analysis, practice, reinforcement and measurement.
Both groups may become faster. Only one is likely to become more valuable. The professional standard should be simple: Do not begin by asking what training to build. Begin by asking what is happening, what should be happening and what prevents people from getting there. The prescription comes last. The diagnosis comes first.
Share on
