Human Performance Technology & Performance Support

Human Performance Technology & Performance Support

Quick orientation: HPT is the discipline UPSTREAM of instructional design. It asks “is training the right intervention?” before any course is designed. This skill covers the full diagnostic and intervention-selection toolkit: the ISPI model, Gilbert’s BEM, Mager-Pipe flowchart, Rummler-Brache, performance support (EPSS/DAPs), Five Moments of Need, and job aid design.

The foundational HPT claim: “If you pit a good performer against a bad system, the system will win almost every time.” — Rummler & Brache, Improving Performance (1990). About 80% of performance improvement opportunities are environmental, not individual (a practitioner estimate — see §2 for the calibration caveat). Training is the solution to a knowledge/skill gap specifically — not a performance gap generically.

Volatile claims — the DAP/EPSS vendor landscape in §6 and the market/effectiveness figures throughout — were verified-as-of 2026-06-16. The durable models (§§1–5, 7–11) are stable HPT canon (Gilbert 1978; Mager & Pipe 1970; Rummler & Brache 1990; Gery 1991).

Quick Reference: Intervention Selection Decision Table

1. The ISPI/HPT Model

Human Performance Technology (HPT) is ISPI’s (International Society for Performance Improvement) systematic, systemic, results-oriented framework. The ISPI model (Deterline & Rosenberg, 1992; updated 2000) is “the most representative and frequently utilized process model in the HPT field” (ERIC/TechTrends, 2017).¹

The fundamental distinction from instructional design: “HPT models allow for a much broader range of interventions to a given performance problem, which often include non-instructional interventions as well as those that involve instruction or training.” (Morrison, Ross, Kalman & Kemp, 2013, cited in jwkline.com).² The HPT model determines whether training or other interventions are effective — training is one output, never the assumed starting point.

Five Phases

Phase 1 — Performance Analysis (Need or Opportunity) Three sub-analyses run together:

• Organizational analysis: vision, mission, goals, strategies — the business context
• Environmental analysis: external and internal realities shaping actual performance
• Gap analysis: actual performance vs. desired performance; if actual = desired, no intervention needed

Output: a quantified performance gap statement with business rationale.

Phase 2 — Cause Analysis Why does the gap exist? Uses Gilbert’s BEM (see §2) to categorize causes:

• Environmental support factors: data/information/feedback; resources/tools; consequences/incentives
• Individual behavior factors: knowledge/skills; individual capacity; motivation/expectations

Technique: fishbone diagram applied to each BEM cell. Output: a statement of why performance is not happening — before any intervention is selected.

Phase 3 — Intervention Selection, Design, and Development Interventions are drawn from a broad taxonomy (see §4). Training is one category of seven. The rule: “Training should only be applied in those instances where no other cheaper and less timely intervention will work.” (Robinson & Robinson, 1995, as cited by multiple HPT references).³

Phase 4 — Intervention Implementation and Change Management Four components: the intervention itself, organizational change, leadership engagement, individuals affected.

Phase 5 — Evaluation Three tiers run in parallel with all prior phases:

• Formative: evaluates each phase as it runs
• Summative: reaction, competence, job transfer, organizational impact, ROI
• Confirmative/meta: validates the process, captures lessons learned

Empirical note: A content analysis of 30 actual HPT business cases (ERIC/TechTrends, 2017) found practitioners compress or adapt the model in practice — actual performance analysis processes differ from the sequential model. A “refined performance analysis process” based on actual practice has been proposed.¹

2. Gilbert’s Behavior Engineering Model (BEM)

Thomas Gilbert published Human Competence: Engineering Worthy Performance (1978). The BEM is his Third Leisurely Theorem, grounded in two prior theorems:

First Theorem (Worthy Performance): W = A/B — human competence is the ratio of valuable accomplishments (A) to costly behavior (B). Optimize the ratio, not the activity.

Second Theorem (Exemplary Performer Analysis): Potential to Improve Performance (PIP) = W(exemplary) ÷ W(typical). The gap between typical and exemplary performers represents the addressable opportunity — and the exemplary performer’s behaviors reveal the environmental conditions that enable good performance.

The Six Cells (2×3 Matrix)

Environmental factors (management’s responsibility — fix these first):

Individual factors (performer’s repertory — fix only after environment is addressed):

Gilbert’s Environmental-First Principle (“Gilbert’s Law”)

Diagnostic sequence: E1 → E2 → E3 → I1 → I2 → I3 (Data first, Motives last).

Rationale: “Environmental factors such as information, resources, and incentives are usually more cost-effective to fix than individual factors… Even if we were to successfully change individual factors, performance will most likely not improve if environmental factors remain unresolved.” (Chevalier, “Updating the Behavior Engineering Model”, Performance Improvement).⁴

Rummler-Brache practitioner estimate: “about 80 percent of performance improvement opportunities reside in the environment,” with 15–20% in skills/knowledge and fewer than 1% in individual capacity (rummlerbrache.com).⁵ Calibration caveat: this 80/20 split is grounded in Rummler & Brache’s consulting experience, not controlled empirical research (Clark, citing Christensen & Wallace, 2012).⁶ Treat it as a strong, directionally robust practitioner heuristic — not a measured statistic; the ratio shifts toward skills/knowledge for highly autonomous or novel-problem work.

Training’s BEM position: Knowledge (I1) is the fourth cell to check — after three environmental cells. The BEM structurally makes training a later consideration, not a first response.

Critique: Gilbert’s Motivational Claims

Gilbert asserted that improving environmental motivation “can usually obliterate all evidence of defective motives.” He dismissed non-monetary incentives as ineffective, citing no supporting research (PerriKennedy.com, 2012). Modern organizational psychology — self-determination theory, engagement research — contradicts this; organizations can influence intrinsic motivation through environmental design. Structural support for the BEM’s factor relationships is stronger than support for Gilbert’s motivational sequencing.⁷

3. Mager & Pipe: Analyzing Performance Problems

Robert Mager and Peter Pipe’s Analyzing Performance Problems (1970; 3rd ed. 1997) is a decision flowchart for diagnosing any performance discrepancy. It “validates the importance of examining and attacking the ‘non-training’ solutions that arise first instead of creating costly and in many cases un-needed learning solutions.” (GSU HPT Manual).⁸

The 7-Step Flowchart

The pivotal question: “Could they do it if their life depended on it?” — Yes means it’s NOT a skill deficiency; it’s an environmental, motivational, or feedback problem. Only a “no” routes to training.

Non-training “fast fixes” (Steps 3–4): Remove negative consequences for performing correctly; add positive consequences; provide clear standards and feedback; remove obstacles; simplify the task or provide decision support. These are Gilbert’s E-cells restated as actionable fixes.

Criticism: The Mager-Pipe model is “a bit too simple for complex problems because a yes/no answer is often not enough.” Many problems have multiple causes requiring multiple interventions. The flowchart is a diagnostic starting point, not a complete solution tool. (Rothwell, Hohne & King, 2013).⁹

4. Rummler-Brache: Three Levels and Nine Performance Variables

Geary Rummler and Alan Brache’s Improving Performance: How to Manage the White Space on the Organization Chart (Jossey-Bass, 1990) extends Gilbert’s BEM upward from the job/performer level to process and organizational levels.

The Three Levels of Performance

The “white space” concept: Performance failures frequently occur at process handoffs between departments. These cannot be fixed by individual training or job redesign — they require process-level intervention. This is Rummler-Brache’s distinctive contribution beyond Gilbert.

The Nine Performance Variables (3 Levels × 3 Needs)

The principle: “The majority of managers simply do not understand the variables that influence performance. They are not aware of the ‘performance levers’ that they should be pulling.” (Rummler & Brache, 1990).⁵

Intersection with HPT: Rummler-Brache operates at Process and Organization levels; Gilbert’s BEM operates at the Job/Performer level. Together they form the complete HPT diagnostic stack. When BEM analysis finds no individual-level cause, the problem is likely at the process or organizational level — requiring a Rummler-Brache intervention, not training.

5. The HPT Intervention Taxonomy

Once cause analysis is complete, interventions are selected from this full spectrum:

Key principle: Training should be “the solution of last resort among solutions of equal effectiveness — the most expensive option, the hardest to change.” (ATD, 2017).¹⁰ When multiple intervention types address the same gap, prefer the cheapest and fastest to change first.

6. Performance Support and EPSS

Gloria Gery’s Original Definition (1991)

“An orchestrated set of technology-enabled services that provide on-demand access to integrated information, guidance, advice, assistance, training, and tools to enable high-level job performance with a minimum of support from other people.” — Electronic Performance Support Systems (Gery Performance Press, 1991).¹¹

Gery’s vision was day-one performance — workers perform effectively from the first day because an expert system replaces lengthy apprenticeship.

Gery’s Three-Type EPSS Taxonomy (1995)

Empirical finding: Higher integration produces better performance outcomes, attitudes, and system use. Any level of EPSS support outperforms no support (Altalib, Performance Improvement Quarterly, 2005).¹²

From EPSS to Digital Adoption Platforms (DAPs)

The term “EPSS” faded in the 2000s into knowledge management and e-learning silos. Gartner named the category in 2019 (initially “Digital Adoption Solutions,” now Digital Adoption Platforms (DAPs)). The modern DAP market:

• WalkMe (SAP): Enterprise-leading DAP; extrinsic model; DeepUI AI adapts to application changes automatically. A WalkMe-commissioned Forrester Total Economic Impact study (2020) reported a 368% three-year ROI for a composite organization (sub-3-month payback, ~$20.0M present-value benefits) driven by reduced training/onboarding time and fewer support tickets¹³ — a vendor-sponsored figure, so treat it as directional, not independent evidence; resolves adoption friction at point of use.
• Whatfix: spans both Gery types — intrinsic embedded step-by-step Flows and extrinsic Smart Tips overlays (so it appears as the extrinsic example above); plus AI Agents, ScreenSense contextual guidance; SCORM-compliant; includes Mirror sandbox simulation for pre-go-live readiness.
• Pendo: Analytics-first; designed for SaaS product activation (customer-facing) rather than enterprise ERP workflows.
• Conversational AI agents (Continu Eddy, CGS Cicero Coach): Chatbot-embedded in Slack/Teams; resolve 85%+ of routine policy/procedure queries via LLM + knowledge base, implementing Gery’s extrinsic EPSS in conversational form.

AI evolution: “Generative AI produces quality text, images, audio, and video via natural language prompts, further fulfilling performance support promises. However, the potential of enterprise performance-centeredness remains unrealized in the AI-integrated workplace.” (Dickelman & Christensen, EPSS Central).¹⁴ See genai-education-instructional-design for AI-EPSS platform engineering.

7. The Five Moments of Need (Gottfredson & Mosher)

Conrad Gottfredson and Bob Mosher formalized this framework (2011 book; Learning Guild article 2012). It maps when learning and performance support are needed in relation to the work workflow.

The Five Moments

The Critical Insight

Traditional L&D addresses only Moments 1 and 2. Moments 3, 4, and 5 — Apply, Solve, Change — occur entirely in the workflow and require performance support (EPSS/DAP), not formal training. When organizations shift to performance-first design using 5MoN: training time is reduced approximately by half; time to competency is also reduced by approximately half. (5momentsofneed.com, 2020–2023).¹⁵

Moment 5 (Change) is the most neglected and most costly: Once skills are automatic behaviors, replacing them requires sustained workflow performance support over time — formal training alone cannot accomplish this unlearning/relearning.

The Two Deliverables Model

The 5MoN methodology produces exactly two output types:

1. Targeted Training — for Moments 1 and 2 (takes learners out of the workflow)
2. Digital Coach (EPSS) — for Moments 3, 4, and 5 (embedded in the workflow)

Design sequence is reversed: start with Apply (Moment 3), then work backward to determine what training is actually necessary.

Misapplication warning: Organizations frequently claim to “do 5MoN” by chunking eLearning into micropieces. Mosher explicitly calls this insufficient — true workflow learning eliminates the transfer phase entirely; microchunking still requires the learner to stop working. (Apply Synergies, 2024).¹⁶

8. Job Aid Design

Definition and When to Use

Job aids are any artifacts that direct or guide work performance, reducing recall burden and minimizing errors during task execution (ATD / Klein).¹⁷

Use a job aid (not training) when:

• Task is infrequent (less than monthly) — memorization is waste
• Task has strict sequential steps that must be followed exactly
• Information changes frequently — a course would be obsolete at launch
• High consequence if done wrong — support reduces error rate
• Cognitive load reduction matters more than building durable memory
• Many procedural steps — “if a process has thirty steps but only happens once a quarter, memorization is a waste of resources” (Sherpa Intelligence, 2026).¹⁸

Do NOT use a job aid when:

• Task requires rapid, fluid performance (frequent use demands automation)
• Job aid use would damage credibility (e.g., medical professional in emergency)
• Novel situations predominate — judgment replaces procedure
• Performer lacks prerequisite skills needed to use the aid

The “Job Aid as Default” Principle

“Guidance with job aids should always be the choice, unless barriers exist. Job aids generally cost less to develop than instruction, are easier to revise when performance requirements change, reduce the time to achieve on-the-job performance, and are not subject to forgetting.” (Barry Boothby, HPT Treasures).⁴

Four Format Types

Design principles: one task per job aid; brevity (keywords over sentences); visual design (icons, white space, contrast); point-of-use design (matches the physical/digital environment where the task occurs); test with actual performers before deployment.

9. Learning in the Flow of Work

Josh Bersin articulated this concept in 2018, synthesizing Gery’s EPSS and Gottfredson’s 5MoN with modern platform realities.

Key empirical anchor: The average employee has only 24 minutes per week for formal learning (Bersin by Deloitte, 2015, 700+ organizations). Approximately 50% of all learning interactions from O’Reilly’s technical community are “in the moment of need” technical support (Bersin, 2018).¹⁹

Just-in-time vs. just-in-case:

Platform pattern: Learning experience platforms (LXPs) plugged into productivity tools (Office 365, Salesforce, Slack) recommend microlearning based on tasks being performed — implementing Gery’s intrinsic/extrinsic EPSS in modern SaaS infrastructure.

Academic validation: Chacko & Cox (2025, Industrial and Organizational Psychology, Cambridge University Press) provides peer-reviewed support for the just-in-time microlearning framework.¹⁹

10. The Performance Consultant Role

The Robinson model (Performance Consulting: Moving Beyond Training, 1995/2015) defines the shift from training order-taker to performance partner. The central argument: training programs “focus on developing excellent learning experiences, while failing to ensure that the newly acquired skills are transferred to the job.” (ERIC abstract, ED382791).³

The diagnostic sequence a performance consultant follows:

1. Name the business outcome — not the learning need. “What’s the business problem we’re solving?”
2. Conduct performance analysis — desired vs. actual state and the measurable gap.
3. Conduct cause analysis (BEM cells) — before prescribing anything.
4. Apply the tacit/codified knowledge split (see §11 Limitations).
5. Apply the frequency/consequence test.
6. Select and blend: train for foundational skill, support for procedural reference.

The “order-taker” pattern: A client says “I want a training program because of increased customer dissatisfaction.” The performance consultant recognizes this as a business problem with an assumed-training solution, pulls back to cause analysis, and may discover the cause is unclear expectations, not skill gaps. (Dana Robinson, ATD interview, 2015).³

US organizations spend $175 billion annually on training initiatives “yet often have little to show for it. One reason is that people jump to solutions before they identify the causes.” — Robinson, Robinson & Handshaw (3rd ed., 2015).

11. Limitations and Anti-Patterns

The Tacit Knowledge Limitation

HPT performs best on codified knowledge (procedures, checklists, reference data) and underperforms on tacit knowledge (clinical judgment, adaptive expertise, complex troubleshooting).

Edmondson et al. (Harvard Business School, 2002) studied minimally invasive cardiac surgery adoption across hospitals. Performance improvement via codified knowledge accelerated predictably — later adopters benefited from documented learnings. Performance improvement via tacit knowledge varied significantly and unpredictably across organizations. Conclusion: tacit-knowledge-dependent performance improvement requires “intense communication among past and future users, perhaps involving moving people around so that new users can work closely with more experienced users.” A job aid cannot substitute for this.²⁰

Implication: When the operative performance driver is tacit/adaptive expertise, the HPT default toward performance support is wrong — investment in training, mentorship, and deliberate practice is required.

Other HPT Limitations

• Time-consuming: Full performance analysis, cause analysis, and intervention design require significant data collection and stakeholder involvement, slowing implementation.
• Expertise required: Not all organizations have trained HPT practitioners; the model’s quality depends on the depth of the cause analysis.
• Systems overemphasis: HPT’s environmental focus can miss personal/attitudinal barriers — individual accountability matters when environmental factors are adequate. (247Teach, 2024).²¹
• Knowledge-worker limitation: HPT models designed for procedural work need adaptation for knowledge workers whose performance is driven by complex problem-solving, not procedures. Systems thinking (Senge, Checkland) is a corrective. (Korean academic analysis).²¹

Training Theater Anti-Pattern

“Training without reinforcement becomes entertainment… Many programs are built backward: start with content, build the experience, then ‘add later’ objectives. The harder work — diagnosing the real drivers of performance — gets skipped. Without diagnosis, training becomes content delivery.” (Training Industry, Feb 2026).¹⁸

The mature HPT position: blend training and performance support rather than replace one with the other — train for foundational understanding, support for complex procedural reference and workflow reinforcement.

References