Knowing Doing Gap Pfeffer

1. 1Start by naming the two or three elements of management practice, strategy, and culture that top managers genuinely believe drive long-term performance for your specific business model — not what conventional accounting happens to already track. Reichheld's point: measurement lies at the heart of vision and strategy, so senior leaders must own it rather than delegate it to 'the people with the green eye shades.'
2. 2Audit your inherited measures and delete anything that distorts strategy or fuels short-term, individual-outcome behavior. Apply the Otis Redding test — fewer, focused measures ('can't do what ten people tell me to do') beat comprehensive dashboards nobody acts on.
3. 3Shift each remaining measure from outcome-only to process-focused: measure whether the right practices are being performed, not just financial results after the fact. The Knowing-Doing Survey logic applies — attitude/outcome measures don't tell you what to change; practice measures do.
4. 4Aggregate where possible. Prefer measures at the team or area level tied to shared goals over individual scorekeeping that creates internal competition and destructive behavior.
5. 5Make measures visible and immediate. Post schedules and current performance where everyone can see them continuously, so the data drives real-time judgment on the floor rather than after-the-fact reporting.
6. 6Build in the three gap checks: compare what senior leaders know matters vs. what they think is happening vs. what people throughout the firm report is actually going on. Track those gaps over time as your real test of whether training and change efforts are working.
7. 7Sustain presence after launch. Leaders should be on the floor during the transition confirming the new measures are being used and resolving problems on the spot, not retreating to offices to read reports.

At the Bayport terminal, Kyle instituted a measurement system for each major area and displayed schedules plus current information on how well they were being met on a large board everyone could check at any time. The measures were aggregate and area-based, tied to the new self-managing operator model rather than to individual piecework or supervisor oversight. She reinforced the practice change physically — eliminating shift supervisors, demolishing the corner offices, auctioning her desk for $60 — signaling that measurement served team goals, not hierarchy. Crucially, she and her team stayed on hand during every shift for months to ensure the new methods were used and to resolve problems on the spot. The result was operational, not cosmetic: demurrage fees dropped from over $1,000,000 in the first half of 1995 to less than $10,000. The measures worked because they were simple, visible, process-linked, and owned by leadership present on the floor.

• If a measure is complex enough that only 'green eye shade' specialists understand it, then simplify or cut it — the most aggressive minds must be able to act on it.
• If a measure tracks only financial outcomes and not the underlying management practices, then add a process measure so people know what to change.
• If a measure encourages individuals to compete rather than teams to cooperate, then re-scope it to the aggregate area or team level.
• If a survey or metric reveals a gap between what leaders think is happening and what frontline people report, then treat that gap — not the raw score — as your action agenda.
• If training has run but the knowing-doing gap hasn't narrowed over time, then the training isn't accomplishing anything and the measurement of practice adoption is where to intervene.

• Delegating measurement as a back-office chore while leaders chase 'vision and strategy' — the tell is executives who can describe strategy fluently but work off inherited metrics that quietly distort it.
• Measuring what accounting already captures instead of what matters: the Silicon Valley study found most employers don't understand their total cost of employment or track training and recruitment costs, despite people being their largest cost.
• Piling on too many measures so attention scatters — the tell is people who 'can't do what ten people tell me to do' and default to old behavior.
• Measuring outcomes and attitudes rather than practices, so results move but nobody knows which action to change — the weakness of general surveys like the Gallup Workplace Audit.
• Installing the visible mechanics (the board, the dashboard) without driving the philosophy to the floor — like a plant that hangs an andon cord a 5-year-old can pull but never instills why. The tell: measures exist but managers complain bitterly about them while behavior stays unchanged.

• ▸Performance driver — the specific practice/strategy/culture element leaders believe drives long-term success
• ▸Current measure — what is actually tracked today for this driver (write 'none' if untracked)
• ▸Process vs. outcome — is this measure tracking the practice being done, or only the after-the-fact result?
• ▸Aggregate vs. individual — is it scored at team/area level or on individuals?
• ▸Visibility — where and how often can the people doing the work see current results?
• ▸Distortion check — does this measure push behavior opposite to what leaders want? (Y/N + note)
• ▸Knowing-doing gap — difference between what leaders think is happening here and what frontline people report
• ▸Action — simplify, delete, add process measure, or re-scope to team
• ▸Owner + review date — who sustains it and when the gap gets re-measured

1. 1Audit your reward architecture for zero-sum mechanics: list every practice that ranks, rates, or pays people relative to each other (forced rankings, named comparative sales data, individual commissions, contests). Each one signals that a colleague's gain is your loss.
2. 2Replace relative comparison with absolute standards. SAS Institute keeps sales targets 'mostly as a way of keeping score' rather than as the basis for commissions—set targets people hit together, not tournaments only some can win.
3. 3Strip individual incentive pay where you want cooperation. Follow the SAS model: pay good competitive base salaries, give a small year-end bonus (under 10% of salary), and remove sales commissions so account executives stop hoarding leads and start sharing them.
4. 4Stop posting comparative performance data by name. SAS deliberately does not post sales rankings by name—public comparative scoreboards convert coworkers into rivals and suppress the knowledge sharing you need.
5. 5Convert performance appraisal from documentation to conversation. Do what SAS, GM Powertrain, and Glenroy did—eliminate the annual ranked review; observe results, give tools, set short- and long-term goals, and manage through dialogue rather than a form that sorts winners from losers.
6. 6Reward the practices that require cooperation. Cross-training people to do each other's jobs, working in teams, active suggestion programs, and getting good ideas from other units in the chain all depend on people helping rivals—so measure and recognize those behaviors, not individual standings.
7. 7Tie the CEO philosophy to the practice. State plainly, as SAS co-founder James Goodnight does—'If you take care of your people, they will take care of the company'—and then check every incentive against whether it makes people compete with the person next to them.

SAS Institute deliberately dismantles internal competition. It offers no stock options or phantom stock, no short-term individual incentive schemes, and no sales commissions for account executives. It pays good competitive salaries with a year-end bonus under 10% of salary and merit raises once a year. Critically, the firm deemphasizes individual competition by not posting comparative sales data by name, so no one's standing is broadcast against a colleague's. As VP Barrett Joyner puts it, sales targets exist 'mostly as a way of keeping score,' because 'people are constantly finding holes in incentive plans'—so instead of incenting on a formula, they 'just tell people what we want them to do and what we expect.' The firm even eliminated performance appraisal forms, with HR VP David Russo arguing performance management should be 'based on conversation instead of documentation.' The result is a workplace where people stay because they enjoy the work and the people, not because they are winning a contest against them.

• If two people must share knowledge to succeed but your reward system ranks them against each other, then the ranking wins and the sharing stops—remove the ranking.
• If you cannot incent on a formula without people 'finding holes in it,' then don't use the formula—tell people directly what you expect and score progress without commissions.
• If a practice requires helping a rival (cross-training, suggestion programs, borrowing ideas from other units), then it will only diffuse when helping is not penalized by a zero-sum measure.
• If comparative data (sales, rankings) is posted by name, then assume it is manufacturing rivalry—make it absolute or anonymous.
• If your appraisal process is primarily a document that sorts and rates individuals, then replace it with observed results plus conversation, as SAS, GM Powertrain, and Glenroy did.

• Running contests and comparative scoreboards to 'motivate,' then wondering why units won't share good ideas—the tell is that 'getting good ideas from other units' scores lower in doing than knowing (4.9 vs 4.0 in the 120-unit study).
• Believing you can precision-engineer behavior through incentive formulas; the tell is people gaming the plan—Joyner's line that 'people are constantly finding holes in incentive plans.'
• Treating an annual ranked appraisal as performance management; the tell is that the form gets filled out but performance doesn't change because it drives comparison, not conversation.
• Espousing cooperative values on plaques and mission statements while the pay system still pits people against each other; the tell is a values statement that everyone can recite and nobody's incentives reflect.
• Keeping named comparative sales data 'for transparency'; the tell is that private hoarding of leads and reluctance to help colleagues rises even as talk of teamwork continues.

• ▸Practice name: name each ranking, bonus, contest, or comparative posting you use.
• ▸Zero-sum test: does one person's gain require another's loss? (Y/N)
• ▸Named comparison: is this data posted or ranked by individual name? (Y/N)
• ▸Cooperation required: does the outcome depend on people helping rivals? (Y/N)
• ▸Gaming risk: can people find holes in the formula rather than serve the goal? (Y/N)
• ▸Replacement: absolute target, anonymized score, or conversation-based review?
• ▸Values conflict: does this contradict a stated operating value about teams or people?
• ▸Owner + review date: who removes or redesigns it, and by when?

1. 1Diagnose the interdependence-learning profile of each task before choosing incentives. Ask two questions: Does this work require constant acquisition of new skills/knowledge, or is it well-learned and routine? Are the units performing it independent, or must they share information to succeed? Reserve internal competition for low-learning, low-interdependence tasks only (like the Intel building-maintenance example).
2. 2Build cooperation directly into the reward system rather than assuming it. Follow Lincoln Electric: tie roughly 50 percent of pay to overall enterprise success through profit-sharing, so relative standing is not the only thing that pays.
3. 3Make cooperation a scored dimension of formal evaluation. Lincoln Electric uses cooperation as one of four dimensions supervisors assess. Add peer evaluations of managers so status protection cannot survive review.
4. 4Kill forced ranking of people and units. Ranking operating units best-to-worst (as IBM did) converts colleagues into 'the competition' and shuts down information sharing; replace with measures tied to whole-enterprise performance.
5. 5Engineer physical and temporal room to talk. Prusak's point: learning is the socialization of knowledge, so give people space and unhurried time to tell each other what they are learning; without allocated time, sharing does not happen even when people want to do it.
6. 6Close the specific knowing-doing gaps on sharing. The restaurant data show managers rate 'getting good ideas from other units' at 4.9 knowing but 4.0 doing, and 'sharing financial performance with everyone' at 4.3 vs 3.8. Measure both, then act to shrink the gap.
7. 7Make mistakes shareable by celebrating them. Copy Zander ('How fascinating!') and IDEO ('fail early and fail often') so people surface errors and lessons instead of hiding them to protect status.
8. 8Cross-train employees to do each other's jobs so knowledge lives in more than one head and helping becomes normal rather than a threat.

A senior IBM executive told the senior management team that internal competition was damaging the firm's competitive capabilities. The mechanism was structural: operating units were ranked from best to worst, which introduced competitive dynamics between them. The executive's own words show the result: 'We don't talk to the people in the other operations. They have become the competition. There is no sharing of information and limited cooperation.' The ranking did not reveal weak units; it manufactured enemies out of colleagues. Contrast this with Lincoln Electric, famous for individual piece-rates yet built from the start on cooperation incentives: about 50 percent of pay comes from profit-sharing tied to overall success, and cooperation is one of four evaluation dimensions. James F. Lincoln glorified competition in his writing but understood workers were interdependent and designed teamwork into the system. The lesson: the same firm can preach competition while structurally rewarding cooperation, and the structure, not the rhetoric, determines whether knowledge is shared.

• If a task requires even modest new learning or has any interdependence between units, then do not use internal competition or purely individual incentives—they will suppress needed knowledge sharing.
• If a task is well-learned, routine (often physical), and units are largely independent, then competition is acceptable because little new learning is required.
• If people's rewards depend on relative standing rather than whole-enterprise performance, then expect hoarding—so tie a large share of pay (Lincoln's ~50%) to collective results.
• If you find yourself ranking units or people best-to-worst, then stop: this is the exact move that turned IBM operations into 'the competition.'
• If knowing scores for sharing practices exceed doing scores (as in the restaurant chain), then the constraint is implementation, not awareness—act, don't retrain.

• Reasoning about knowledge work using sports analogies. The tell: leaders (often men) justify competition by pointing to races and teams beating opponents, ignoring that intellectual tasks needing learning and invention are best done without close scrutiny, constant assessment, or present competitors.
• Assuming rhetoric equals structure. The tell: a leader espouses cooperation but the pay and ranking systems still reward relative standing—look at what's measured, not what's said (Lincoln praised competition while paying for cooperation).
• Ranking units or people to 'motivate' them. The tell: colleagues start calling other units 'the competition' and stop talking, exactly the IBM symptom.
• Believing awareness is action. The tell: managers rate a sharing practice high on knowing (4.9) but low on doing (4.0)—for 17 of 25 practices the gap was statistically significant in the wrong direction.
• Wanting knowledge shared but allocating no time or space for it. The tell: people say they'd like to tell others what they're learning but 'do not have the time,' so socialization of knowledge never occurs.

• ▸Task name — one activity or process being incentivized
• ▸Learning intensity (1-6) — does it require constant new skills/knowledge?
• ▸Interdependence (1-6) — must units share info to succeed?
• ▸Verdict — competition-safe (low/low) or cooperation-required (any learning or interdependence)
• ▸% of pay tied to whole-enterprise success — is it near Lincoln's 50%?
• ▸Is cooperation a scored evaluation dimension? — yes/no, name the measure
• ▸Are units/people ranked best-to-worst? — yes/no (yes = red flag)
• ▸Knowing vs Doing gap for sharing — rate importance and actual use (1-6) each, note the difference
• ▸Time/space allocated for people to talk and share — specific mechanism or 'none'

1. 1Separate knowing from doing measurement: build two parallel item lists covering the same specific management practices, one asking whether each practice is important for performance, one asking whether it actually occurs in your unit.
2. 2Ask the knowing questions of senior management only; ask the doing questions of both senior management and people throughout the firm, so you capture perception gaps at every level.
3. 3Tailor items to your industry and company context rather than using a generic attitude survey—phrase them as concrete behaviors (e.g., 'sharing financial performance with all employees') not attitudes, so you know exactly what to change.
4. 4Rate each item on a 1–6 agree/disagree scale using an anchored procedure: first mark all 1s (strongly disagree), then all 6s (strongly agree), then fill the middle (2–5).
5. 5Compute three gaps: (1) what leaders know matters vs. what leaders think is happening, (2) what leaders think is happening vs. what frontline people report, (3) what leaders think affects performance vs. what actually occurs.
6. 6Treat the largest gaps as your action agenda; prioritize practices scored high on knowing and low on doing, because those are proven levers you already believe in but aren't executing.
7. 7Re-run the survey over time and judge training and change efforts solely by whether the gap shrinks—if the gap doesn't narrow, the training accomplished little regardless of how enlightened participants felt.
8. 8Bias toward learning by doing: convert priority gaps into small, immediate actions rather than more study, since implemented knowledge comes far more from acting than from reading, listening, or thinking.

Two consultants worked for a large electrical utility in Latin America facing deregulation. They discovered management already held a four-year-old, 500-page document from a prior firm containing benchmarking cost studies from best-practice utilities worldwide, summaries of successful training systems, and detailed implementation calendars. Working from the same information, their own recommendations came out 'basically a copy of the old document.' Their conclusion: 'The problem was not analysis. It was implementation. The client already had the basic information we were giving them.' The diagnostic move here is to stop buying the fifth analysis and instead measure why the existing, agreed-upon recommendations never became consistent action. A knowing-doing survey would have shown high scores on knowing (leaders agreed the practices mattered) against low scores on doing (the calendars sat unexecuted), pointing straight to implementation as the agenda.

• If four consulting studies keep reaching the same conclusion, then stop paying for analysis and buy implementation—the answer is known, the doing isn't.
• If a practice scores high on knowing but low on doing, then it belongs at the top of your action agenda; if it scores low on both, revisit whether it truly matters before acting.
• If a re-administered survey shows the knowing-doing gap has not narrowed after training, then the training isn't accomplishing much and should be changed or stopped.
• If you can only choose between more contemplation and thoughtful action, then choose action, because implemented knowledge comes mainly from learning by doing.
• If senior leaders' view of what's happening diverges sharply from frontline reports, then treat that perception gap itself as a performance risk, not a reporting error.

• Mistaking training or consulting spend for change—executives believed TQM training had changed how people worked, yet Zbaracki found quantitative methods went unused in four of five firms. Tell: high confidence in enlightenment paired with no observable shift in daily practice.
• Measuring attitudes instead of practices—generic surveys like the Gallup Workplace Audit capture opinions, leaving you unable to know what specifically to change. Tell: survey results move but no one can name which behavior to fix.
• Assuming good analysis equals implementation—the utility's 500-page plan was excellent and still unexecuted. Tell: repeated, high-quality reports that describe the same unchanged reality.
• Backsliding after adoption—some apparel plants adopted superior modular production then abandoned it and returned to the bundle method (80% still used bundle in 1992). Tell: a proven practice quietly reverting to the old way.
• Saying what you don't do—three-quarters of CEOs said companies should have fast-track programs but fewer than half had one. Tell: stated priorities that don't appear in your own doing scores.

• ▸Practice item (specific behavior): state one concrete management practice tailored to your industry, not an attitude
• ▸Knowing rating (1–6, senior mgmt): how strongly you agree this practice is important for performance
• ▸Doing rating (1–6, senior mgmt): how strongly you agree this practice actually occurs here
• ▸Doing rating (1–6, frontline): same question asked of people throughout the firm
• ▸Gap 1 (know vs. leader-perceived doing): subtract leader doing from leader knowing
• ▸Gap 2 (leader-perceived vs. frontline doing): subtract frontline doing from leader doing
• ▸Gap 3 (believed performance driver vs. actual occurrence): note where high-importance items are not actually happening
• ▸Priority flag: mark items high on knowing and low on doing as top action agenda
• ▸Next action / owner / date: the single doable step, who does it, and by when

1. 1Define organizational performance in your context concretely: quality, service, retention, innovation, and financial results — not just short-term financials. Financials 'largely reflect what has happened in the past,' so name the leading indicators too.
2. 2Build a knowing-doing survey. Ask senior leaders 'knowing' questions (which practices are important for performance) and ask both leaders and frontline people 'doing' questions (which practices actually occur). Tailor items to your industry, as done with the restaurant items.
3. 3Rate each item on the 1–6 scale (1 Strongly Disagree to 6 Strongly Agree). Have respondents first mark all 1s, then all 6s, then fill the middle — for both the knowing and the doing versions of each item.
4. 4Compute three gaps: (1) what leaders know vs. what they think is happening, (2) what leaders think is happening vs. what frontline people report, (3) what leaders think affects performance vs. what actually occurs.
5. 5Turn the largest gaps into an action agenda — these are the specific practices to implement, since 'firms that don't implement what they know will perform less well than those that do.'
6. 6Guard your measurement system: keep it few enough measures to act on, and don't let precise metrics crowd out hard-to-quantify drivers that matter long term.
7. 7Re-run the survey over time. Judge training and change efforts by whether the knowing-doing gap shrinks; if it doesn't, 'the training isn't accomplishing much.'

A comparatively small nonprofit fund-raising foundation knew what drove its performance: 'developing people internally is the best way to produce high-quality fundraisers and build a successful foundation.' Yet it kept hiring outside 'stars' with grant-writing and estate-planning skills who had immediate impact but did not fit the start-up culture or embrace the mission. The knowing-doing gap was severe and measurable: 19 people lost in three years — expensive, disruptive, and destructive to building a strong group. Executives all agreed the star-hiring strategy was failing, yet persisted, because Cialdini's 'social proof' led the inexperienced executive to imitate similar organizations that hired for skills over fit. Here superior retention and long-term capability (organizational performance) were sacrificed to a practice everyone knew was wrong. Closing the gap meant hiring less-skilled, high-potential people who fit — exactly what leaders said they believed but did not do.

• If leaders rate a practice high on 'knowing' but low on 'doing,' then treat that item as the priority action, not as a training topic they already understand.
• If frontline 'doing' scores fall well below leaders' assumed 'doing' scores, then fix leaders' visibility into operations before launching new initiatives.
• If you're unsure how to act and start copying similar organizations, then stop — that 'social proof' reflex is how you adopt practices that contradict your own knowledge.
• If a measurement system has grown too complex or too subjective to drive decisions, then cut measures back to the few most consequential for performance.
• If a driver of performance is hard to quantify (culture fit, learning from mistakes), then measure it imperfectly rather than omit it — precise metrics 'often miss important elements.'

• Buying one more book, seminar, or consulting study to 'know' more. Tell: the same recommendations recur — one consultant found four prior studies in six years reached identical conclusions; 'the problem was not analysis, it was implementation.'
• Measuring attitudes instead of practices. Tell: survey results move but you 'can't know what, specifically, to change,' because outcomes were measured rather than behaviors.
• Using only general, cross-industry surveys. Tell: the questions are too generic to reflect what leaders in your specific context believe drives performance.
• Judging by short-term financials alone. Tell: managers optimize the past and neglect customer satisfaction, employee development, and new products that shape future performance.
• Assuming training equals change. Tell: TQM training that executives believed had changed behavior — yet quantitative methods went unused in four of five firms Zbaracki studied.

• ▸Practice item — a specific management behavior tailored to your industry (e.g., 'sharing financial performance with all employees').
• ▸Knowing score (senior leaders) — 1–6: how important this practice is for performance.
• ▸Doing score, leader estimate — 1–6: how much leaders think this actually occurs.
• ▸Doing score, frontline — 1–6: how much frontline staff report it actually occurs.
• ▸Gap 1 (knowing minus leader-doing) — the say-do gap leaders can see.
• ▸Gap 2 (leader-doing minus frontline-doing) — leaders' blind spot into operations.
• ▸Performance dimension affected — quality / service / retention / innovation / financial.
• ▸Action owner and date — who closes this gap and by when.
• ▸Re-measure date — when you re-survey to confirm the gap shrank.

1. 1State the underlying philosophy and business model before you write any technique — answer why you operate this way, not just what you do. For Toyota, that means people, processes, quality, and continuous improvement, not kanban cards and andon cords.
2. 2When copying a successful practice from another firm, first extract the meta-theory that organizes it. Ask 'what view of people and the business makes this practice make sense?' and adopt that, since practices 'evolve and make sense only as part of some system.'
3. 3Screen partners and hires on attitude and philosophy fit, not only technical capability — Honda chooses supplier-partners on 'risk-taking attitude, motivation to improve, responsiveness to future needs' because teaching technical knowledge is easier than instilling philosophy.
4. 4Test whether your stated values are enacted, not just displayed: interview the people lowest in the hierarchy (analysts, plant-floor workers) and compare their lived experience to the values on the wall.
5. 5Drive philosophy down to every level through repeated effort, not proclamation — DaPrile's point is that a 5-year-old can pull the andon cord, but installing the philosophy behind it takes sustained work.
6. 6Assign someone to carry philosophy and its stories across the organization the way Corey Billington does at HP — 'part librarian, part consultant, part coach' who knows both the technical solutions and the surrounding stories.
7. 7When talk starts substituting for action, impose a real deadline with real measures — Xerox's Baldrige bid forced a gruelling internal analysis that focused the company on quality goals more than years of meetings had.
8. 8Use measurement and mistakes to support learning and drive out fear, letting measures evolve as your understanding of the business model and guiding philosophy sharpens.

Xerox spent years running Leadership Through Quality with meetings, task forces, and documents, yet assessment interviews found only 25 percent felt the company used the quality processes to a great extent, 15 percent felt rewards were based on quality, and 13 percent used cost of quality in decisions — the summary concluded 'quality is not the basic business principle today at Xerox.' The talk had become a substitute for action rather than a guide to it. The turnaround came not from more documents but from imposing a real deadline with real measures: in December 1988 Kearns and Allaire launched a year-long bid for the Malcolm Baldrige Award, forcing a gruelling internal analysis. According to Kearns, the real value lay in forcing Xerox to focus harder than ever on its quality goals. The following year, the Business Products and Systems division was one of only two recipients of the 1989 Baldrige Award. The lesson: a philosophy only guides action when a concrete external test forces people to enact it rather than recite it.

• If you can describe a competitor's practice but not the philosophy behind it, then you have copied the surface and not the soul — do not implement it yet.
• If a candidate or supplier is technically strong but lacks the right attitude and values, then favor philosophy fit, because technical knowledge is easier to teach than attitude.
• If a value is written, believed, and displayed but the lowest-level people report it isn't carried out, then treat it as unenacted talk, not a live value.
• If meetings, documents, and decisions keep multiplying without behavior change, then impose a real deadline with real external measures to force action.
• If you must choose what to store and transfer as 'knowledge,' then capture the guiding philosophy and its stories, not just the codified, tangible techniques.

• Confusing having a mission or values statement with implementing it — the tell is values printed on little cards, plaques, and posters treated as if they had 'magical powers,' while frontline experience contradicts them.
• Copying explicit techniques (kanban, andon cords, quality circles) while missing the philosophy — the tell is best practices that 'just doesn't spread' even between two plants across the street from each other.
• Overestimating the tangible, programmatic aspects of what competitors do and underestimating the underlying philosophy — the tell is a knowledge-management effort focused entirely on codified information and technology.
• Letting talk substitute for action — the tell is many meetings and documents producing little difference, as at Xerox, where the small effect came because of, not in spite of, the program.
• Espousing core values leaders genuinely believe but don't live — the tell at the securities firm: values of respect, teamwork, integrity on paper while analysts get secretarial work, recruiting is 'second tier,' and they're treated as short-term contractors.

• ▸Guiding philosophy (why): state the view of people and the business in one sentence
• ▸Business model link: how this philosophy explains why we make money
• ▸Core values (max 3): the values we claim to live
• ▸Enactment evidence: what frontline/lowest-level people actually experience for each value
• ▸Gap flag: mark any value that is displayed but not lived
• ▸Practice-to-philosophy trace: for each key technique, name the philosophy that makes it make sense
• ▸Attitude screen: the attitudes/values we require in hires and partners over technical skill
• ▸Philosophy carrier: person responsible for storing and coaching the philosophy and its stories
• ▸Forcing test: the real deadline and real measure that will force enactment, not talk

1. 1Move knowledge acquisition onto the floor: keep managers 'always on the floor helping employees when they're in the weeds' so learning happens inside the work, not in a classroom. Cross-train employees to do each other's jobs so knowledge transfers by doing and watching, not by reading manuals.
2. 2Replace contemplation with thoughtful action. When facing an organizational problem, take a small action to generate experience you can learn from, rather than scheduling more meetings and discussions about it.
3. 3Institutionalize learning from mistakes: have managers 'talk openly about learning from their own mistakes' so error becomes a shared teaching event instead of a hidden individual failure.
4. 4Build informal-learning channels deliberately—mentoring, supervising, on-the-job training of peers, interacting with customers—because up to 70 percent of workplace learning is informal and travels through stories and watching one another work.
5. 5Run the knowing-doing survey: ask senior leaders 'knowing' questions (is this practice important?) on a six-point scale, then ask 'doing' questions (does this actually occur?) of people throughout the firm using the identical items and scale.
6. 6Measure the three gaps: (1) what leaders know vs. what they think is happening, (2) what leaders think is happening vs. what staff report, (3) what leaders think affects performance vs. what actually occurs. Treat those gaps as your action agenda.
7. 7Judge training by gap reduction over time, not attendance or satisfaction. If the knowing-doing gap on a practice does not shrink after training, the training accomplished little.
8. 8De-emphasize codified-information technology (intranets, canned presentations, written reports) as the primary knowledge vehicle; use it to support, not replace, the social interaction through which working knowledge actually moves.

In 120 units of a restaurant chain, managers rated 25 practices twice—first on importance ('knowing'), then on actual use ('doing'), same six-point scale. For 17 of 25 practices the gap was statistically significant (p < .001), and every gap ran the same direction: they weren't doing what they knew mattered. 'Getting good ideas from other units in the chain' scored 4.9 known but 4.0 done; an active suggestion program 4.8 vs 3.9; detailed hiring assessment 5.0 vs 4.2; talking openly about mistakes 4.9 vs 4.3; frequent feedback 5.7 vs 5.2; sharing financial performance 4.3 vs 3.8. A second chain studied the same way produced nearly identical results. The lesson: the shortfall was not individual deficiency—the managers understood the issues—but a systematic failure to embed knowing into doing, which the survey made visible and actionable.

• If a practice scores high on knowing but low on doing, then the deficit is implementation, not knowledge—stop delivering more information and change the work.
• If someone blames the knowing-doing gap on individuals' skills or personality, then look at the organization instead; research suggests individual deficiency is not a large part of the story.
• If a piece of knowledge is technical or tacit, then transfer it by apprenticeship, watching, and doing—not by storing it in a report or intranet.
• If you're deciding between more analysis and a small action, then take the action, because implemented knowledge comes far more from doing than from reading, listening, or thinking.
• If a training program hasn't reduced the measured knowing-doing gap over time, then it isn't working—redesign or drop it.

• Treating knowledge management as a technology problem—'knowledge management starts with technology.' Tell: initiatives center on intranets and codified reports while day-to-day working knowledge goes uncaptured.
• Substituting talk for action—endless meetings contemplating problems. Tell: everyone can articulate the issue precisely, yet nothing on the floor changes.
• Cutting in-house training when the economy sours, hollowing out learning-by-doing exactly when you need adaptive capacity. Tell: training budget is the first line cut.
• Ignoring prior experience and repeating known failures—like New York City learning nothing from Kentucky's and Penn Manor's incentive-pay results. Tell: you 'reinvent' a practice whose problems were already well documented.
• Measuring only attitudes (general satisfaction surveys) instead of specific practices. Tell: you get scores that correlate with performance but can't tell you what specifically to change.

• ▸Practice statement: write one specific management/learning behavior in your context
• ▸Knowing score (1-6): senior leaders rate how important this is for performance
• ▸Doing score (senior view, 1-6): what leaders think is actually happening
• ▸Doing score (frontline, 1-6): what people throughout the firm report actually occurs
• ▸Gap 1 (leader knows − leader thinks happening): compute the difference
• ▸Gap 2 (leader thinks − frontline reports): compute the difference
• ▸Learning mode: mark whether this knowledge moves by doing/coaching/teaching or by report/intranet
• ▸Action to close gap: one on-the-floor experiment you will run this month
• ▸Re-measure date: when you re-survey to confirm the gap shrank

1. 1Ask senior leaders what management practices they believe drive performance (the 'knowing' questions), phrased as specific behaviors, not general attitudes — tailor them to your industry and company context.
2. 2Ask 'doing' questions of both senior management AND people throughout the firm about the extent those same practices actually occur in daily action; the doing sample must be broad because leaders overestimate implementation.
3. 3Measure the gap between what senior leaders say should be done and what people report is actually happening — that gap is your action agenda, not a scorecard for admiration.
4. 4For every decision, plan, mission statement, or presentation, name who does what by when; treat any output that stops at 'we now understand' as talk substitution, not action.
5. 5Back cultural or strategic talk with a measurement that tracks the behavior — the way Men's Wearhouse tracks transactions per consultant to enforce its team-selling philosophy rather than just preaching it.
6. 6Prefer learning by doing over learning by reading, listening, or thinking; schedule 'thoughtful action' — small implementations you can learn from — instead of another round of contemplation or a repeat training.
7. 7Re-run the knowing-doing survey over time to evaluate whether training and initiatives actually reduced the gap; if the gap doesn't shrink, the training isn't accomplishing anything and should stop being repeated.

Professor Mark Zbaracki studied Total Quality Management training in five organizations whose senior executives believed TQM would improve product and service quality and that the training had changed how people did their jobs. On the knowing side, leaders knew the quantitative TQM methods and had paid for training to install them. On the doing side, Zbaracki found the quantitative methods were used not at all in four of the five organizations and only on a limited basis in the fifth. The gap is total: talk and belief about TQM fully substituted for the practice of TQM, and no one measured actual method usage, so the illusion of implementation persisted. An action-oriented response would have surveyed workers on whether they actually ran the quantitative methods, exposed the near-zero doing rate, and killed the repeated training until real usage rose. Contrast this with Men's Wearhouse, where the team-selling talk is backed by transaction-count data that flags a consultant hogging walk-in traffic — talk plus a measure that forces the behavior.

• If a meeting, presentation, or mission statement produces no named owner, deliverable, and deadline, then treat it as talk substitution and do not count it as progress.
• If leaders assert a practice is 'in place' but no one measures its actual occurrence, then assume the doing rate is far below the knowing rate until proven otherwise — as in four of Zbaracki's five firms.
• If a repeated training or bought book covers ideas already known, then the problem is doing, not knowing; spend on implementation and measurement instead of more education.
• If you must choose between more analysis and a small real trial, then run the trial — implemented knowledge comes from learning by doing, not from reading or thinking.
• If a stated value (teamwork, quality, service) has no data tracking the underlying behavior, then add a specific metric before you trust the culture to hold, especially across many sites.

• Mistaking enlightenment for change: leaders leave seminars 'wiser but not much happens' — the tell is high praise for the content and zero shift in daily practice.
• Measuring attitudes instead of practices: general surveys like a workplace audit tell you people feel good but not what specifically to change — the tell is survey results improve while operations don't.
• Assuming implementation because you paid for it: executives believe training 'changed how people performed their jobs' when the method is used nowhere — the tell is nobody can point to the practice in actual work.
• Repackaging known ideas as action: buying and re-teaching timeless principles yearly — the tell is the same prescriptions in new graphics with no reduction in the knowing-doing gap over time.
• Talk without a backing measure: preaching a value like team selling but never tracking the behavior — the tell is one person quietly outperforming by hogging traffic while the culture is assumed intact.

• ▸Practice / behavior (name one specific, observable management practice, not an attitude)
• ▸Knowing score (senior leaders: how important is this practice to performance, 1-5)
• ▸Doing score – leaders (senior leaders: how much does it actually occur, 1-5)
• ▸Doing score – workforce (people throughout the firm: how much does it actually occur, 1-5)
• ▸Gap (Knowing minus lowest Doing score; the larger the gap, the higher the action priority)
• ▸Backing measure (what data currently tracks this behavior — or 'none', which is a red flag)
• ▸Named owner + deadline (who implements the fix and by when; blank = talk substitution)
• ▸Next thoughtful action (the smallest real trial you will run to learn by doing)
• ▸Re-measure date (when you re-run the survey to confirm the gap actually shrank)

1. 1Name the sacred cows explicitly. Ask people to list policies and procedures 'that everybody knows about but that nobody talks about' — the ones that have outlived their usefulness. Quesnelle at Mitel used exactly this framing to surface barriers before Mitel went broke clinging to them.
2. 2Trace each questioned practice to its origin. Ask 'why do we do this?' and separate real reasons from 'we've always done it this way' and 'that's how our competitors do it.' Precedents from unknown origin and social proof are the two justifications to challenge first.
3. 3Re-derive the practice against current facts, not historical ones. Recompute using today's cost structure, demand, and knowledge — not a five-year history. Lockheed-Martin lost the value of $700M in achieved cost reductions because bids were built on historical costs even after the cost structure changed.
4. 4Reframe from 'whether' to 'how.' Once obstacles are named, stop debating if change is possible and ask 'how do you solve the problems you described?' This turns the same people who listed reasons not to act into people generating ways around obstacles.
5. 5Make the old way hard to use and make people mindful of its harm — sometimes through dramatic means — then design and implement a replacement while resistance is low.
6. 6Watch for threat-rigidity during downturns. When the firm is in trouble, expect people to cling harder to old practices; deliberately protect and resource any new-learning effort (like acquired intellectual capital) because financial pressure makes learning harder, not easier.
7. 7Guard new units against importing precedent. If you start a subunit to escape history, recruit people who won't drag old habits in, and watch for 'monkey see, monkey do' copying of what every peer firm does.

Fresh Choice, a buffet salad chain, saw its stock fall from about $32 in 1994 to $3 in 1998. As part of a turnaround it acquired three units of Zoopa specifically to acquire Zoopa's intellectual capital — the whole rationale for the deal. But financial threat triggered exactly the threat-rigidity effect Barry Staw documented: instead of learning from Zoopa, executives clung to what they already knew how to do, and their anxiety made absorbing new practices harder. Atari showed the same pattern earlier: facing declining demand and rising competition in the early 1980s, it doubled down on outdated video-game hardware and abandoned a working 1982 prototype much like the Apple Macintosh (which arrived in 1984) — even as executives told each other they'd probably regret it. In both cases the organization abandoned superior new ideas and retreated to ingrained practice precisely when breaking from the past mattered most.

• If the only reason given for a practice is 'we've always done it this way' or 'competitors do it too,' then treat it as unexamined precedent and re-derive it from current facts.
• If a practice's ineffectiveness is already acknowledged by leaders but it persists anyway, then you are looking at a sacred cow — attack the fear of touching it, not the logic (the logic is already conceded).
• If the organization is under financial or competitive threat, then expect and counteract threat-rigidity; deliberately protect new learning rather than letting people hunker down on the familiar.
• If the present genuinely resembles the past and the old way was done well, then keep relying on precedent — re-examining every decision daily is infeasible and precedent rightly economizes effort.
• If you spin up a new unit to escape history, then vet recruits for imported habits, or the new unit will simply re-embody old practices.

• Treating a single past action as legitimate forever. The tell: 'that way of operating' became automatically acceptable after being done once, regardless of whether it worked — no one can point to evidence it succeeded.
• Confusing efficiency with correctness. The tell: people defend a shortcut because it saves collecting new information and weighing options, without checking whether old ways are still the right ways.
• Silencing the person who questions the cow. The tell: a courageous objector with good reasons is ignored or rebuked, or mentored to 'not mention this idea to anyone' — the car-replacement manager was told he'd be 'asking for trouble' and 15 years later still can't kill it.
• Assuming a new organization guarantees fresh thinking. The tell: 'monkey see, monkey do' — the venture capitalist admits funding the same people and imposing the same structures as every peer, 'just making them do what we always make them do.'
• Reading threat as a reason to narrow. The tell: when demand drops the firm focuses harder on its oldest competencies and shelves superior new work, verbalizing regret in advance yet doing it anyway.

• ▸Practice/policy: name the specific thing being done
• ▸Stated justification: write the exact reason people give (watch for 'always done it' / 'competitors do it')
• ▸Actual origin: when and why did this start — is the origin even known?
• ▸Current facts vs. historical basis: what data was it built on vs. what is true now (e.g., new cost structure)
• ▸Acknowledged as flawed? yes/no — and who has already admitted it
• ▸Fear/who guards it: name the sacred cow's protector and what happens to objectors
• ▸Cost of continuing: quantify waste of keeping it (dollars, turnover, lost intellectual capital)
• ▸Reframe to how: state the change as an obstacle to solve, not a yes/no decision
• ▸Replacement + how to make old way hard: the new practice and the step that blocks reverting

1. 1Audit where bad news dies: ask managers whether subordinates bring them problems they don't want to hear, and watch how they react. Deming's prescription—drive out fear—starts by finding the specific moments where fear silences information. The tell that you have a problem: no one brings you news you don't want to hear.
2. 2Separate acting-on-knowledge from being-punished-for-it. People act on what they know only when they believe taking a risk based on new insight will be rewarded, not punished. Publicly protect the first people who question precedent or take an unfamiliar assignment, so others see that courage is safe.
3. 3Measure fear directly with anonymous surveys rather than assuming your workplace is enlightened. Use the Princeton Survey questions: how many employees have withheld a suggestion because it might cost someone a job (national baseline: 16%, one in six); how many trust the company to keep its promises (baseline: 38%). Even 'nice' firms score badly.
4. 4Close the loyalty-trust gap by keeping specific commitments. Distrust builds when people believe promises were broken—Saturn people trusted 'if you succeed, we'll give you the rest of the capital' and then it went to Delaware. Track every promise made to employees and whether it was honored, because unkept commitments convert into strikes, disidentification, and hoarded knowledge.
5. 5Stop rewarding fear-based bosses. The market rewarded 'Chainsaw Al' Dunlap—Sunbeam stock rose 49% the day he was named CEO—for bellowing at subordinates who brought bad news. Change your internal incentives so managers admired for being 'tough' or 'mercurial' are not the ones who get promoted.
6. 6During threats and turnarounds, actively counter threat-rigidity. Fear plus deadlines, fatigue, and financial pressure make people freeze on past knowledge and abandon new ideas (Atari shelved its Macintosh-like prototype). When performance is bad, deliberately protect experimentation instead of letting anxiety drive everyone back to what they've always done.

Fresh Choice, a buffet salad chain, saw its stock fall from about $32 in 1994 to $3 in 1998. To turn around, it acquired three Zoopa units in Seattle specifically to import Zoopa's 'guest first' service culture and design—that intellectual capital was the main rationale for the deal. But Fresh Choice's top-down style, described by an insider as 'always hammered down policies and procedures,' meant it kept imposing its own precedents on the acquired stores. The financial threat made executives cling even more tightly to what they knew—Staw's threat-rigidity effect—so the very anxiety driving the acquisition also blocked learning from it. The knowledge sat there; fear and rigidity prevented acting on it. Fresh Choice got the locations but lost the intellectual capital it paid for.

• If people stop bringing you news you don't want to hear, then assume fear is now filtering your information—do not read the silence as everything being fine.
• If a manager is admired internally as 'tough,' 'hard-nosed,' or unpredictable, then treat that reputation as a performance risk, not a credential—no one wants to work for that person, and fear suppresses the knowledge you need.
• If you are in a financial threat or under deadline pressure, then expect threat-rigidity and cognitive closure to push everyone toward old habits—install explicit permission to try new things before anxiety freezes the organization.
• If you have made a success-contingent promise (like Saturn's capital commitment), then keep it exactly, because a broken promise converts loyalty into distrust, disidentification, and hoarded effort.
• If employees withhold suggestions because they might cost someone a job, then the problem is safety, not idea quality—fix the punishment structure before asking for more ideas.

• Assuming your enlightened, well-trained workplace is immune. The passages show two firms with strong reputations and civil-service-like protections still surfaced pervasive fear in their own surveys—the tell is leaders who say 'that doesn't happen here' without ever measuring it.
• 'Making nice' only during a labor shortage. Efforts to put people first evaporate once the shortage eases and firms return to old mean-spirited ways—the tell is that humane treatment is framed as a temporary recruiting tactic rather than a standing policy.
• Confusing consistency with wisdom. Commitment to past decisions signals persistence and shields those decisions from being questioned—the tell is that no one can explain why a practice exists except that it's always been done that way (mindless, precedent-driven behavior).
• Letting threat trigger rigidity instead of learning. Under financial or competitive pressure, people cling to what they do best and abandon untried ideas—the tell is executives who admit they'll 'probably be sorry' for killing a promising project (as at Atari) but kill it anyway.
• Treating care for people as soft-headed and unbusinesslike. The tell is that putting employees first is dismissed as not serious, while bellowing bosses draw laughs and stock bumps—so fear stays rewarded and knowledge stays buried.

• ▸Bad-news test: name the last time someone brought leadership news it didn't want to hear—and what happened to them.
• ▸Suggestion-withholding rate: % of employees who withheld an improvement idea for fear it would cost a job (national baseline 16%).
• ▸Promise-trust score: % of employees who trust the company to keep its promises (national baseline 38%).
• ▸Broken-commitment log: list every success-contingent promise made to staff and whether it was actually honored.
• ▸Rewarded-fear check: which admired managers are known as 'tough'/'mercurial'—and are they being promoted?
• ▸Threat-rigidity flag: current deadline/financial/competitive pressures pushing the group back to old habits.
• ▸Abandoned-innovation list: promising new ideas dropped 'to be safe'—and who said they'd regret it.
• ▸Care-as-soft test: where is putting people first currently dismissed as unbusinesslike?
• ▸Safety action: one specific protection you will give the next person who questions precedent.