Learning in the Flow of Work: The Shift From Push to Pull
68% of employees prefer to learn at work, embedded in the workflow, rather than in a separate training event. Only 10% of skills are learned through formal training — the other 90% come through on-the-job practice, social learning, and performance support at the moment of need. Yet most enterprise L&D investment still goes into push…
1. Push vs Pull: The Fundamental Design Shift
Push learning is scheduled by the organisation and delivered to learners through module release dates, mandatory training windows, and calendar-based programmes. Pull learning is initiated by the learner at the moment they need support — a job aid accessed before executing a rarely-performed task, a microlearning module retrieved before a high-stakes conversation, a knowledge base consulted in the moment of decision. Most enterprise L&D is push. Most of how people actually learn at work is pull.
68%
of employees prefer to learn at work — embedded in operational context — rather than in separate training events that take them out of workflow (LinkedIn Learning via LevelUp LMS Statistics 2026)
Only 10%
of skills are learned through formal training sessions — the other 90% are acquired on the job, through peers, and through practice in the real work context (eLearning Industry Training Statistics 2025)
72%
of LXP use cases are for on-demand content consumption — versus 41% for LMS — reflecting the learner preference for pull access over scheduled push delivery (LLCBuddy via Continu eLearning Statistics 2025)
25–67%
improvement in retention when employees learn at their own pace in their workflow context — the range across multiple studies connecting learner-controlled access to measurable retention improvement (eLearning Industry Training Statistics 2025)
Key Distinction
Push training delivers learning in advance of the moment it will be needed. Pull support delivers learning at the moment it is needed. Push is appropriate when the learner needs foundational capability development, understanding, a conceptual framework, initial skill that cannot be acquired at the moment of performance. Pull is appropriate when the learner has the foundational capability and needs performance scaffolding, such as a reference, a checklist, or a reminder, at the specific moment of application.
2. What Belongs in Flow-of-Work Design
| Content Type | Flow-of-Work Format | Moment of Need |
|---|---|---|
| Infrequent procedure: correct steps required but rarely executed | Step-by-step job aid or decision tree, accessible in the relevant system | Immediately before executing the procedure |
| Regulatory Reference: Applicable Rule in a Specific Situation | Searchable knowledge base entry, integrated in the CRM or workflow tool | During the client interaction or decision point where the regulation applies |
| Spaced reinforcement: counteracting forgetting from prior training | 3–5 minute microlearning scenario, pushed at scheduled interval post-training | Days 7, 30, 90 after initial training event |
| Troubleshooting: solving an operational problem | Decision-tree troubleshooting guide in the relevant operational system | At the moment the problem presents |
| Pre-task confidence building: before a high-stakes interaction | Short scenario or reminder accessible via LXP or mobile platform | Immediately before the high-stakes task or conversation |
3. How Flow-of-Work Learning Complements Formal Training
“The formal training event develops the capability. The flow-of-work support deploys it. The learner who completed the onboarding module on the compliance procedure three months ago and now faces the boundary case in a real customer interaction does not need to re-attend the module. They need the 90-second decision tree that walks them through the specific situation they are facing. The module built the capability. The job aid applies it.“
- Design formal training to build capability; design performance support to deploy it. Formal training should be retained for objectives that require the learner to develop foundational understanding, decision-making frameworks, and initial skills. Performance support should handle the procedural reference, regulatory lookup, and operational checklist that the learner needs at the point of application without needing to rebuild understanding from scratch.
- Identify which training content is currently formal that should be performance support. A significant proportion of every enterprise training catalogue consists of content the learner cannot apply without reference to procedural steps for infrequent tasks, regulatory requirements for specific situations, and product specifications for rare customer questions. This content is often delivered as modules that the learner completes and immediately forgets, when a searchable job aid accessed at the moment of need would produce far better operational performance at a fraction of the development cost.
4. Designing Performance Support for the Moment of Need
- Map the workflow moments where performance support is needed — not the topics that could benefit from training. The design process starts from the specific operational moment: the task the employee must complete, the decision they must make, the situation they must respond to. The performance support asset is designed for that specific moment — not for the general topic. A job aid for “data protection” is not a moment-of-need asset. A job aid for “deciding whether a customer request qualifies as a subject access request and what to do next” is.
- Integrate performance support into the tools employees use — not into a separate learning platform. The performance support asset that requires the employee to navigate to the learning platform, log in, search for the resource, and retrieve it under time pressure will not be used. The same content embedded in the CRM, the ERP, the Teams workspace, or the mobile app the employee already has open at the moment of need will be used. Integration into existing workflow tools, not the addition of another tool, is the design standard for effective performance support.
- Measure access rate and task performance, not completion rate. The effectiveness indicator for performance support is whether employees use it at the moments it was designed for, measured by access rate in context, and whether operational performance improves at those moments. The completion rate that measures formal training is not the right metric for performance support. Usage rate and task performance data are.
In Summary
Learning in the flow of work is not a replacement for formal training; it is the performance infrastructure that extends formal training’s value into the moments where it needs to produce operational performance. The 10% of skills developed through formal training requires the 90% of on-the-job application infrastructure to produce the reliable performance that the training investment was designed to enable.
Most enterprise L&D functions are still investing predominantly in the 10% and assuming the 90% will happen organically. Designing performance support assets, job aids, decision trees, searchable knowledge bases, and integrated microlearning for the specific workflow moments where employees need knowledge and guidance is the investment that extends formal training ROI at relatively low additional cost and significantly higher impact than additional formal content production.
Qquench · 25+ Years · Performance Support Design · Job Aid and Decision Tree Development · Workflow-Integrated Knowledge Assets · Point-of-Need Microlearning · Fortune 100 · Global
Qquench designs performance support for the specific moments that formal training cannot reach when the learner is in the task, facing the decision, or executing the procedure that their foundational training prepared them for, but cannot accompany them through.
We map workflow moments of need, design the right support format for each, and integrate assets into the tools employees use — rather than adding another platform they will not navigate to under operational pressure.
Frequently Asked Questions
Q1
What is learning in the flow of work?
Knowledge and guidance are provided at the moment of need, embedded in the work context rather than delivered in a separate training event. Job aids, performance support tools, searchable knowledge bases, microlearning at the point of application, and AI-powered in-workflow assistants. The learner accesses support when they need it, in the context where they need it.
Q2
What content is best suited to flow-of-work delivery?
Procedural guidance for infrequent tasks. Reference content regulations, product specifications, and policy summaries. Troubleshooting guides. Spaced reinforcement microlearning at defined intervals after formal training. Pre-task confidence building before high-stakes interactions.
Q3
What is the relationship between flow-of-work learning and formal training?
Flow-of-work is not a replacement for formal training. Formal training develops foundational capability, conceptual understanding, a decision-making framework, and initial skill. Flow-of-work support deploys that capability in operational practice, providing the scaffolding that allows application without requiring the learner to rebuild understanding from memory under pressure.
Q4
How should L&D design for learning in the flow of work?
Map specific workflow moments where employees need knowledge to perform correctly. Design support assets for those specific moments not general topics. Integrate into tools employees already use at those moments. Measure access rate and task performance, not completion rate.
QS
Qquench Specialists
Performance Support and Learning Architecture Practice · Qquench
25+ years designing performance support that embeds learning in the workflow moments that formal training cannot reach. We write from practice, not position papers.