Operating Doctrine

Competitors are not uniform. A large software company earns most of its money from a narrow core: a few products, sold to a particular size of buyer, through a particular channel. Around that core sits everything else it has accumulated. Adjacent segments, smaller customers, secondary geographies, features maintained because removing them would embarrass someone. The core funds the company. The rest is held because it was won once and has not yet been worth the cost of letting go.

The claim is that the core and the surroundings are defended with entirely different intensity, and that a new entrant should therefore never open where the money is. It should open where the money is not, take positions that the incumbent will decline to contest, hold them long enough that they compound, and only then look inward. Stated in the older vocabulary: take the villages before the capital, and take them because the capital will not march out to defend a village.

Two things follow immediately, and they are the parts most often dropped. The first is that this is a sequencing rule, not a licence to be everywhere. The second is that it only works where the incumbent genuinely will not respond, which is a fact to be established rather than assumed.

The oldest form of the argument is Sun Tzu: avoid strength, strike weakness, and arrange the position so that the outcome is settled before the engagement. The most useful form is Corbett, who was writing about naval power and asked a question that translates almost directly. Can a limited object be taken and held without first defeating the enemy's main force? His answer is yes, on one condition, which is that the object be isolated from that force. Isolation is the whole of it. A peripheral position that the main force can reach is not peripheral.

Lanchester supplies the arithmetic. Under conditions of aimed fire, effective strength rises with the square of numbers, so a force half the size of its opponent is a quarter as effective in a general engagement. The conclusion is not that the smaller side loses. It is that the smaller side must never permit a general engagement, and must instead create situations where it is locally the larger force. That is a mathematical statement of why a new company should put its entire effort into one narrow segment.

Liddell Hart adds the line of least expectation, and the observation that the aim of the approach is dislocation rather than destruction: you want the opponent's position to become untenable, not to destroy the opponent. Mao supplies the shape of the campaign, base areas held securely in the countryside, space traded for time, the cities surrounded last. Boyd supplies tempo, the idea that acting faster than the opponent can decide means their response arrives against a situation that has already changed.

Clausewitz supplies the two correctives that the rest of the reading list lacks. War is the continuation of policy by other means, which is to say the object governs the effort and an effort that outgrows its object has failed even if it wins. And there is a culminating point of victory, a point past which every further advance makes the winner weaker. Both are constraints on this doctrine rather than justifications for it.

The military framing explains what to do. It does not explain why it works, and if it is left to stand alone it teaches the wrong lesson, which is that the periphery is weakly held. Usually it is not weakly held. Usually it is rationally abandoned, and the difference matters enormously.

Christensen's account is the correct one. A company with high gross margins, whose best customers are its largest, allocates resources toward those customers because doing so is correct at every individual decision. Defending a small, price-sensitive, low-margin segment against a cheaper entrant means committing senior attention and engineering to the least profitable revenue it has, and the internal case for doing so will lose to the internal case for serving the core. Walking away raises the average margin and is reported as discipline. The opening is not a gap in the enemy's line. It is a decision the enemy is making on purpose, for reasons that are sound from inside their own accounts.

Ries and Trout give the practical form: flanking and guerrilla campaigns exist precisely because a frontal attack on a leader is arithmetically hopeless, and both require the willingness to leave ground that cannot be held. Moore gives the sequencing rule that matters most in software: take one beachhead segment completely before touching the next, and choose the next because it is adjacent in the customer's world rather than adjacent on an internal map. Hamel and Prahalad document the only clean case of the full doctrine being executed deliberately, Komatsu against Caterpillar, whose internal slogan was encirclement in as many words. That case is also the warning, because it ran for two decades and was carried by a company with a protected home market and a state banking relationship.

The operative test therefore is not military at all. It is an accounting question. Would defending this segment cost the incumbent more than the segment earns them? Where the answer is yes, a position is available and can be held for years. Where the answer is no, there is no flank, only a fight against a better capitalised opponent on ground they have decided to keep.

What the above amounts to in practice, for each business the company runs. The order is load bearing. Steps taken out of order are the most common way the method fails.

1. 01

   Locate the core

   Identify which revenue actually funds the incumbent's ability to respond, and which revenue is merely on their price list. Everything after this depends on getting it right, and it is not always the product they market most loudly.

2. 02

   Find ground they cannot profitably hold

   Apply the accounting test from section three. A segment qualifies when defending it would cost the incumbent more than it earns them, and when winning it does not require them to lose anything they would notice. If we cannot state why they will decline to respond, we have not found a flank.

3. 03

   Estimate the response before committing

   Chen's criteria: how much market do we share with them, and how similar are our resources. High commonality with a much stronger opponent predicts a response we cannot survive. This step exists to cancel campaigns, and if it never cancels one it is not being run honestly.

4. 04

   Concentrate

   One segment, one buyer type, disproportionate resource, until it is finished. This is Lanchester's requirement and it is the step that a diversified group is structurally worst at obeying. Section six returns to this.

5. 05

   Hold before advancing

   A position is taken when the customers stay without being persuaded again. Retention, references, and the routines the customer has built around the product are the only things that make ground permanent. Advancing from an unheld position converts one weak presence into two.

6. 06

   Advance along the customer's path, not the map

   The next segment is the one the current customers are already adjacent to: same buyer, same procurement route, same problem one size larger. Adjacency on an internal org chart is not adjacency.

7. 07

   Set the withdrawal condition in advance

   Before the first commitment, write down the evidence that would mean this position is not available, and the date it will be checked. Written afterwards it never gets written. This is the only defence against the escalation described in section six.

Everything above is abstract until it is placed on ground. For a company that sells software to businesses and does no outbound selling, the ground is the search result page. That is where a buyer with a problem and a budget goes first, and it is the one place where the position a company holds can be measured from outside, by anyone, on any day.

Read as terrain, a search query is a settlement. The head term for a category is the capital: high volume, high commercial value, and held by a small number of domains that have been accumulating links and coverage for a decade. A narrow query, a single business function named for a single industry, is a village. It is worth less, it is reached by fewer people, and it is usually held by whoever bothered to write a page about it.

The useful property of this battlefield is that Corbett's isolation condition stops being a matter of judgement and becomes a number. For any query, the strength of the current occupants is measurable, and so is the strength of the weakest of them. That weakest result is the decision: it is the cheapest way in, and if even it is far beyond what an unknown site could reach, then the query is closed regardless of how attractive it looks. This is the difference between a flank and a wish.

Three tests follow, and they are applied in this order because each one is cheaper than the next and each can end the enquiry.

1. 01

   Is there a way in at all

   Look at who currently holds the top of the page and how much accumulated authority the weakest of them carries. If nothing there is within reach, no amount of writing changes the outcome. Published difficulty scores are not a substitute for this and have been wrong every time they were checked against the live page.

2. 02

   Is it worth taking

   Volume and commercial value, measured rather than assumed. This test exists because the first one selects against it. The queries that are easiest to win are overwhelmingly the ones nobody has bothered to contest, and a result page can be open precisely because there is no money behind it.

3. 03

   Will the need come back

   A subscription requires a problem that recurs. Demand that arrives in a single annual spike is an event, and someone who searches once and buys once is not a subscriber. This is measurable from the shape of the demand across a year, so it is measured.

Two cautions belong here rather than in the criticism that follows, because they are properties of this specific terrain rather than of the metaphor.

The first is that ranking is not the objective. It is a proxy for getting in front of a buyer, and a proxy that can be satisfied completely while the objective is not met at all. Ground that produces visits and no customers has been taken and is not worth holding, and no amount of further advance along the same line fixes it. The measure of a position is a customer, never a placement.

The second is that search is one approach and not necessarily the best one. Where the buyers for a product already gather somewhere, in a marketplace, a package registry, an integration directory, the route to them runs through that place and requires none of the accumulated authority a search result demands. That is the line of least expectation in Liddell Hart's sense: not a harder push along the obvious road, but a different road. A campaign that treats the search page as the only battlefield will miss it.

The war metaphor buys clarity about sequencing and pays for it in accuracy about almost everything else. The objections below are not hedges. Several of them are, on the evidence, fatal to the strong form of the thesis, and the honest position is that the doctrine survives only in the reduced form given in section seven.

1. 01

   Markets have no geography

   Encirclement is a spatial idea and it needs contiguity, a supply line, and a route from the outer position to the inner one. None of these exist here. Winning small customers from an enterprise vendor supplies no route to their enterprise accounts, because the buying committee, the procurement process, the security review, and the switching cost are unrelated to those of the segment just won. There is frequently no path from the village to the capital at all, and believing there is one is the most expensive error available in this framing.

2. 02

   Winnability selects for worthlessness

   This is the objection the terrain in section five makes unavoidable, and it is the one the doctrine has no good answer to. The two properties that matter, how easy a position is to take and how much it is worth, run in opposite directions almost everywhere. A position is undefended most often because defending it would earn nobody anything, and a strategy that ranks candidates by weakness of opposition is therefore a strategy that ranks them, approximately, by how little they are worth. The rare cell that is both open and valuable is the entire thesis, and rare is the operative word.

3. 03

   Nothing is captured, everything is rented

   Territory taken in war stays taken until it is retaken. A subscription customer leaves in thirty days and takes the position with them. Conquest language implies a ratchet that churn simply does not provide, and it encourages counting wins that have not been kept.

4. 04

   The enemy usually is not fighting

   Section three is the load-bearing explanation, and it is not martial. When the incumbent withdraws from a segment for margin reasons, reading that withdrawal as a defeat inflicted produces two errors at once. It credits us with a capability we have not demonstrated, and it leaves us with no way to tell the difference between ground that was ceded and ground that will be contested. The second error is discovered at the worst possible moment, which is on arrival at the ground they intend to keep.

5. 05

   The zero-sum premise is usually false

   War is distributive. Most software categories are still growing, and the majority of available revenue is held by people currently using spreadsheets, email, and nothing. A frame that directs every unit of attention at named rivals points it at the smaller pool. Porter's objection is sharper still: competing to beat rivals converges on imitation, and imitation competes away the return for everyone in the category.

6. 06

   Multipoint retaliation, and we are all periphery

   Karnani and Wernerfelt describe how a firm attacked in one market strikes back in another where the attacker is exposed. A group running 15 products presents 15 surfaces to counterattack, and none of them is a capital. There is no protected core here that funds a long campaign; every position is a village. Declaring hostilities from that footing invites a form of response we are the least equipped of anyone to absorb.

7. 07

   Concentration and diversification contradict each other

   This is the deepest problem and it does not have a comfortable answer. The Lanchester argument that justifies attacking a dispersed periphery says with equal force that the attacker must concentrate. A portfolio of many products is the definition of not concentrating. Either the products are genuinely independent, in which case each must win on its own and the group narrative adds nothing to any of them, or resources are shared, in which case the group is committing the exact error the doctrine identifies in its opponents. A doctrine of concentration held by a diversified company is a doctrine held in name.

8. 08

   The language degrades the decisions

   Perceived threat narrows information processing and centralises control, which is the opposite of what a small challenger needs. And a front that has been publicly declared is a front that keeps receiving money after the evidence has turned, because withdrawal now reads as defeat rather than as reallocation. The metaphor is not free. It is paid for in judgement, and the invoice arrives late.

9. 09

   Ambition is not a strategy

   Rumelt's charge against bad strategy is that martial and motivational language substitutes for a diagnosis, a guiding policy, and coherent action. Everything on this page above section four is at permanent risk of being exactly that. The test is whether the doctrine has ever caused a decision not to be taken. If it only ever authorises action, it is not doing any work.

Strip the metaphor and a smaller, more defensible method is left, and it is the one actually in use.

Enter where the established supplier's economics make defence irrational for them, because that is what buys the years needed to become competent. Concentrate everything on one narrow segment until it is genuinely held, because concentration is the only advantage a small entrant can manufacture. Expand along paths the customer already walks. Treat retention, not acquisition, as the measure of whether ground has been taken. Decide the abandonment condition before committing, and honour it.

Read that way, the interesting object is not the competitor at all. It is the customer who is currently served badly or not at all, and the reason no established company finds it worth serving them properly. The sequencing rule remains useful. The hostility is decoration, and on the evidence in section six it is decoration that costs more than it returns.

The word retained internally is campaign, not war. A campaign has an object, a budget, a culminating point, and an end. That is the part of the military inheritance worth keeping.

These are constraints on the method, and they are not negotiable against any objective it might serve.

• No competitor is named as an adversary in any public document, including this one.
• No product is marketed by disparaging another company's product. Comparison is made on documented, checkable facts or it is not made.
• No campaign targets an individual person at any company.
• Nothing described here changes an obligation to a customer. Contracts, refunds, and support are governed by the published terms and nothing on this page qualifies them.
• No competitive objective justifies misrepresenting what a product does, what it costs, or who is billing for it.

Each entry states the one claim taken from the work. The two counterweights, Porter and Kim and Mauborgne, argue against the frame used here and are listed for that reason.

Strategy and war

Sun Tzuc. 5th c. BC

The Art of WarAvoid what is strong, strike what is weak. Position so the outcome is settled before the engagement begins.

Carl von Clausewitz1832

On WarWar is policy carried on by other means, so the political object governs the effort. Also: the culminating point of victory, past which further advance weakens the winner.

Julian Corbett1911

Some Principles of Maritime StrategyA limited object can be taken and held without defeating the enemy's main force, provided the object is isolated from it. The closest thing in the literature to what we are actually doing.

Frederick Lanchester1916

Aircraft in Warfare: The Dawn of the Fourth ArmUnder aimed fire, effective strength rises with the square of numbers. Concentration compounds, so the smaller side must never accept a general engagement and must always create local superiority.

B. H. Liddell Hart1954

Strategy: The Indirect ApproachMove along the line of least expectation. Dislocate the opponent's position before attempting to decide anything by force.

Mao Zedong1937 to 1938

On Protracted War; On Guerrilla WarfareBuild secure base areas in the countryside, trade space for time, and surround the cities from the countryside. The direct historical antecedent of the thesis on this page, and the reason we should be honest that it is not a business idea in origin.

John Boyd1986

Patterns of ConflictTempo is itself a weapon. Operating inside the opponent's decision cycle makes their response arrive against a situation that no longer exists.

Strategy and competition

Bruce Henderson1976 and 1967

The Rule of Three and Four; Brinkmanship in BusinessStable markets settle into a small number of viable competitors. A challenger's opening is created by the leader's unwillingness to pay the price of an all-out response.

Al Ries and Jack Trout1986

Marketing WarfareFlanking and guerrilla forms exist because frontal attack on a leader is arithmetic suicide. Fight where the leader is not, and be willing to leave ground you cannot hold.

Gary Hamel and C. K. Prahalad1989 and 1994

Strategic Intent; Competing for the FutureKomatsu's stated intent was to encircle Caterpillar, segment by segment, over two decades. The documented case of this doctrine working, and evidence that the timescale is decades rather than quarters.

Geoffrey Moore1991 and 1995

Crossing the Chasm; Inside the TornadoTake one beachhead segment completely before touching the next, and choose the next because it is adjacent in the customer's world, not because it is adjacent on our map.

Michael Porter1996

What Is Strategy?A counterweight. Competing to beat rivals produces imitation and mutually destructive rivalry; strategy is a distinct position, not a contest. The strongest argument against the frame this page uses.

Clayton Christensen1997 and 2003

The Innovator's Dilemma; The Innovator's SolutionThe incumbent abandons the low end because abandoning it raises their margins. Asymmetry of motivation, not weakness of defence, is what makes the periphery winnable.

Aneel Karnani and Birger Wernerfelt1985

Multiple Point CompetitionA firm attacked in one market retaliates in another where the attacker is exposed. A diversified group offers more surfaces to counterattack than a focused one.

Ming-Jer Chen1996

Competitor Analysis and Interfirm RivalryWhether a rival responds is predictable from market commonality and resource similarity. Before attacking, estimate the response rather than assuming one.

W. Chan Kim and Renee Mauborgne2005

Blue Ocean StrategyA second counterweight. In most growing categories the money is in non-consumption, so a competitor-facing frame points attention at the smaller pool.

Richard Rumelt2011

Good Strategy / Bad StrategyStrategy is a diagnosis, a guiding policy, and coherent action. Ambition and martial language are the common substitutes for all three, and this page is at permanent risk of being one of them.