design-tactical-doctrine

Build Tactical Doctrine

Identify the expert judgment that drives superior outcomes in your domain, observe and document the decision patterns of top performers as explicit rules with situational triggers, test those rules with average practitioners until they can execute without deep expertise, and maintain the resulting doctrine as a living system — so that organisational performance is no longer bounded by the supply of exceptional individuals.

Why This Is Best Practice

Origin: Zhuge Liang (181–234 AD), Prime Minister and commander of Shu Han's armies, faced a persistent strategic problem: he had exceptional strategic judgment but could not be everywhere, and Shu Han's officer corps was thinner than Cao Wei's. His solution was the Eight Formation Arrays (八阵图) — a systematic battle formation doctrine that encoded expert tactical responses to standard combat situations into rules that average unit commanders could follow. Each of the eight formations specified: the situation it was designed for, the configuration of troops, the response to specific enemy moves, and the transitions between formations. Du Fu's famous verse — "功盖三分国，名成八阵图" (His achievements covered the Three Kingdoms; his fame came from the Eight Formations) — reflects the doctrine's reputation. Lu Xun of Wu, having defeated Liu Bei at Yiling, reportedly encountered a stone formation left by Zhuge Liang at the mouth of a valley; his general entered it and became disoriented, and Lu Xun — a formidable commander himself — refused to enter, saying the formation appeared to still be operational. The formation outlasted the man who designed it, continuing to function without him.

Adopted by: Systematic doctrine as organisational force multiplier is the mechanism behind the most scalable businesses and institutions in history. Ray Kroc's insight at McDonald's (1954) was not that McDonald's food was exceptional — it was that the McDonald brothers had created an operating system that could produce consistent results with any staff. Kroc bought the franchise rights and codified the system into the Operations and Training Manual — a 75-page document (later expanded to hundreds of pages) specifying every procedure, temperature, timing, and interaction. By encoding the McDonald brothers' judgment into doctrine, McDonald's became scalable across 40,000 locations with varying staff quality. Neil Rackham's SPIN Selling (1988) encoded the sales patterns of top performers — identified through analysis of 35,000 sales calls — into teachable methodology that lifted average reps toward top-quartile results. Charlie Munger's "mental models" framework and Warren Buffett's investment checklist encode decades of investment judgment into explicit criteria that a competent analyst can apply, reducing dependence on Buffett's and Munger's specific intuition for each decision.

Impact: Organisations that depend on exceptional individual judgment face a scaling ceiling: there are never enough exceptional individuals, and those who exist are the organisation's single points of failure. Doctrine-based organisations decouple performance from individual supply. McDonald's scale is not dependent on finding millions of exceptional fast-food operators — the doctrine makes ordinary operators capable. The US Marine Corps produces effective infantry units with 18-year-olds who have had 13 weeks of training — because the tactical doctrine encodes the judgment that experienced combat soldiers develop over years, and 13 weeks is enough to learn the doctrine. The bottleneck shifts from "how many exceptional people can we find" to "how good is our doctrine and how well do we train it."

Why best: The alternative to tactical doctrine when performance depends on expert judgment is either (a) concentrating exceptional individuals at all critical points — which fails when those individuals are unavailable — or (b) accepting performance variance as permanent — which means the organisation's results vary with the supply and retention of exceptional people. Doctrine is the third option: transfer the exceptional individuals' knowledge into a system, then train the system. The system doesn't quit, get poached, or fall ill.

Sources: Chen Shou, Records of the Three Kingdoms 三国志 — "Zhuge Liang Zhuan" 諸葛亮傳 (280–290 AD); Du Fu, "八阵图" 杜甫 (~766 AD); John F. Love, McDonald's: Behind the Arches (1986); Neil Rackham, SPIN Selling (1988); Goldratt, The Goal (1984)

Steps

Step 1: Identify the expert judgment that drives superior outcomes

Doctrine cannot be built from what experts say they do — it must be built from what they actually do that produces superior results. Start with performance data, not with interviews:

• Identify the top quartile of performers in the domain (sales reps, engineers, support agents, investment analysts)
• Measure the performance gap between the top quartile and the median: is it large enough to justify the doctrine-building investment? If the top quartile is 20% better, doctrine has limited leverage; if it is 3× better, doctrine has enormous leverage
• Observe top performers in action across multiple situations — do not ask them to explain their approach; observe what they actually do. Top performers are often unable to articulate the rules behind their intuitive decisions (the knowledge is tacit, not explicit)

The judgment you are trying to encode is not what top performers believe they do. It is the pattern of decisions that, when observed empirically, produces the performance gap.

Step 2: Document decision patterns as explicit rules with situational triggers

The core discipline of doctrine building is converting tacit pattern recognition into explicit if-then rules. Tacit knowledge ("I can tell when a customer is ready to close") is not teachable; explicit rules ("if the customer has named a budget, asked about implementation timeline, and referenced their decision date in the same conversation, use the direct close") are teachable.

For each expert pattern observed:

• Identify the trigger: What situation, condition, or signal activates this response?
• Specify the response: What exactly does the expert do? Not "they communicate clearly" but "they ask three specific questions in this order"
• Identify the variations: Under what conditions does the standard response not apply?

The Eight Formation Arrays specified eight distinct tactical situations and the specific formation response to each. The doctrine did not say "use good judgment" — it said "when the enemy presents X, deploy formation Y."

Step 3: Specify transitions — when to shift from one rule to another

Doctrine that specifies responses to individual situations but not transitions between them is incomplete. Practitioners need to know not just what to do in each situation but when to change what they are doing.

Zhuge Liang's eight formations specified not just the static configurations but the movement patterns between them — how to shift from a defensive encirclement to an offensive penetration as the battle situation changed. A sales doctrine that specifies discovery, demonstration, and closing as separate phases is not complete without rules for when to move from discovery to demonstration — what signals indicate that enough has been discovered?

Map the transition conditions: the observable signals that indicate a practitioner should shift from one mode to another. These transitions are often where expert practitioners most significantly outperform average ones — average practitioners often stay in one mode too long or shift too early.

Step 4: Test with average practitioners — the doctrine is not valid until they can execute it

The target practitioners for doctrine are not experts — they are average performers. Test the doctrine with average practitioners before finalising it:

• Give average practitioners the written doctrine without any additional coaching
• Observe where they succeed and where they fail
• Where they fail: is the failure because the doctrine is incomplete (missing a rule or a trigger), ambiguous (the rule is unclear), or simply requires knowledge the practitioner does not have (a training problem, not a doctrine problem)?

The doctrine is not validated by expert practitioners saying "yes, that's what I do." It is validated by average practitioners executing it and achieving materially better outcomes than they achieved without it. If the gap persists after average practitioners have studied the doctrine, the doctrine has not yet captured the relevant expert judgment.

Step 5: Build in the failure modes — explicit exceptions and escalation paths

Any doctrine that applies to all situations without exceptions will produce failures in the cases it does not cover. Zhuge Liang's formations had specific responses to situations they were not designed for — retreat paths, regrouping procedures, signals for when to abandon the formation and use individual judgment.

For the doctrine:

• Identify the situations where the standard rules do not apply — edge cases, novel situations, situations that require judgment beyond what the doctrine encodes
• Specify explicitly what to do in those situations: either "use judgment" (with guidance on what factors to weigh) or "escalate" (with specification of to whom and with what information)
• The failure mode specification is as important as the standard doctrine — practitioners who hit an undocumented failure mode with no guidance will either freeze or improvise badly

Step 6: Build feedback loops — doctrine must update from field results

Static doctrine decays. The situations the doctrine was designed for change; competitors adapt to predictable patterns; new edge cases emerge that were not in the original specification. Doctrine that is not updated from field experience becomes a liability — it produces consistent results in situations it was designed for and produces consistent failures in situations it was not.

Establish two feedback loops:

• Failure loop: When a practitioner follows the doctrine and fails, investigate whether the failure reveals a doctrine gap (update the doctrine) or a training gap (update the training)
• Success loop: When a practitioner deviates from the doctrine and succeeds, investigate whether the deviation reveals a doctrine gap (update the doctrine) or a tactical innovation (add to the doctrine)

The most dangerous practitioner for doctrine quality is the one who regularly outperforms by ignoring the doctrine — they may be a top performer whose tacit knowledge is not yet encoded. Investigate their deviations rather than disciplining them.

Step 7: Assign ownership and protect from entropy

Doctrine degrades without active maintenance. Assign specific ownership of the doctrine to a person whose performance is measured against the doctrine's effectiveness:

• The owner reviews doctrine quarterly against field results
• No practitioner may update the doctrine unilaterally — changes go through the owner, who validates them against performance data before incorporating them
• The owner removes rules that have stopped working, even if they were added by senior people
• Practitioners can flag potential doctrine updates; the owner evaluates them

The Eight Formation Arrays maintained their effectiveness for decades because the doctrine was explicit and transmissible — it did not depend on Zhuge Liang being alive to teach it. The doctrine's owner after Zhuge Liang's death was the system itself, transmitted through the trained commanders.

Rules

• Build doctrine from observation, not from interviews. What top performers say they do and what they actually do are often different. Observe multiple performances before converting patterns to rules.
• The doctrine is not validated until average practitioners can execute it. Expert validation is necessary but not sufficient. The doctrine must work for the people who need it, not for the people who already know it.
• Every rule must have a trigger. "Be responsive to customers" is not doctrine — it is aspiration. "Within two hours of a customer message that contains the word 'urgent' or describes a production issue, respond with acknowledgment and estimated resolution time" is doctrine.
• Specify the failure modes explicitly. Doctrine without documented edge cases and escalation paths produces confident failures — practitioners who apply the wrong rule to the wrong situation because they have no guidance on when the standard rules do not apply.
• Assign ownership and review cadence. Doctrine that is not maintained becomes stale and then counterproductive. The ownership must be specific, the review must be scheduled, and the performance data must flow back to the owner.
• Protect the doctrine from HiPPO override. High-status individuals often override doctrine based on intuition rather than evidence. Doctrine updated by authority rather than performance data ceases to encode field-validated knowledge and begins to encode the HiPPO's preferences. Changes to doctrine require performance evidence, regardless of who proposes them.

Examples

McDonald's Operations and Training Manual (1958): Ray Kroc's insight was that the McDonald brothers' Speedee Service System — a kitchen layout, preparation sequence, and customer interaction protocol that produced a consistent product in 30 seconds — could be codified and transmitted to any operator. The manual specified everything: fryer temperature (335°F for fries), portion sizes (1.6 oz patty), cleaning schedules, greeting scripts. By encoding the McDonald brothers' judgment into doctrine, McDonald's could hire ordinary people and train them to produce the McDonald brothers' results. The doctrine was the business — not the food, not the brand, but the system. McDonald's scaled to 40,000 locations on the strength of a manual that any franchisee could learn.

SPIN Selling — sales doctrine from empirical observation: Neil Rackham and the Huthwaite Research Group analysed 35,000 sales calls over 12 years to identify what top performers actually did differently. The result was SPIN Selling: four question types (Situation, Problem, Implication, Need-Payoff) and the sequence in which top performers used them. The doctrine did not emerge from asking top performers what they did — it emerged from observing thousands of calls and identifying patterns that correlated with successful outcomes. When organisations trained average reps on the SPIN methodology, they achieved measurable improvements in close rates — not because the doctrine was clever, but because it encoded what actually worked, derived empirically rather than anecdotally.

US Marine Corps infantry doctrine: The USMC produces effective infantry units from recruits with 13 weeks of basic training. This is possible because decades of combat experience have been encoded into tactical doctrine: when to advance, when to suppress, how to communicate under fire, how to assess and treat casualties, how to transition between formations. The doctrine does not replace judgment for novel situations — it handles the 80% of combat situations that follow recognisable patterns, freeing the Marine to use judgment for the 20% that do not. The doctrine is updated after each major engagement — failures are analysed and doctrine is revised. The system produces consistent results from consistently average practitioners because the expert knowledge is in the doctrine, not in the individual.

Engineering incident runbooks: A technology company's senior engineers develop intuition for diagnosing and resolving production incidents that takes years to develop. Without doctrine, incidents that junior engineers cannot resolve wait for senior engineers — creating a bottleneck on the senior engineers' availability. With doctrine (runbooks): each known failure mode is documented with symptoms, diagnostic steps, resolution procedures, and escalation conditions. Junior engineers following the runbook resolve 70% of incidents that previously required senior engineers. The senior engineers' time is freed for the 30% of incidents the runbook does not cover — the novel failures that require genuine expert judgment. The doctrine has not replaced the senior engineers; it has replaced the need for senior engineers to handle routine failures repeatedly.

Common Mistakes

Building doctrine from expert interviews rather than observation: Asking top performers to explain their approach produces their rationalised account of what they believe they do — which often differs significantly from what they actually do. The pattern that produces the performance gap is often invisible to the expert themselves, because it operates at the level of intuition rather than conscious reasoning. Observe multiple performances before converting patterns to rules; verify the rules against performance data before finalising.

Writing principles instead of rules: "Be empathetic with customers," "prioritise long-term value," "communicate proactively" are principles, not doctrine. They are inarguably correct and entirely unactionable. Doctrine specifies what to do, when to do it, and what signal indicates it is time to do it. If the practitioner reading the doctrine still has to exercise significant judgment to know what to do in the next ten minutes, the doctrine has not been written yet.

Validating with experts instead of average practitioners: Showing the doctrine to top performers and getting their approval is the wrong validation test. Top performers can follow any doctrine because they already know what to do — the doctrine confirms their existing knowledge. The test is whether average practitioners, following only the doctrine, achieve materially better outcomes. If they do not, the doctrine has not yet captured the relevant expert judgment.

Leaving edge cases undocumented: Practitioners who hit an undocumented edge case with a confident rule-following mindset will either freeze (waiting for guidance that does not arrive) or apply the wrong rule (the closest-seeming rule in the doctrine). Both are worse than having no doctrine at all. Document the edge cases, even if the guidance is "use judgment" — knowing when to deviate from the doctrine is as important as knowing when to follow it.

Allowing doctrine to stagnate: Doctrine written once and never updated encodes the judgment of experts from the time it was written, against the competitive landscape as it existed then. As situations change, competitors adapt, and edge cases accumulate, static doctrine becomes increasingly inadequate. The doctrine's usefulness decays at the rate at which the field changes. Assign ownership, schedule reviews, and build the feedback loops before the doctrine is deployed — not after it starts failing.