What Marketing Strategy Actually Is

The company that thought it had a strategy

By 2003, Kodak had what looked, on paper, like a strategy. There were binders. There was a PowerPoint deck, thick and expensively produced, that articulated a "vision of the digital imaging future" and a "bold transformation roadmap". There was an ambition, publicly stated, to become a leading player in digital photography. There were task forces, growth initiatives, and a five-year plan with milestones, owners, and quarterly review dates. From the outside, and from the inside of the boardroom, this looked like a company thinking strategically about a shifting market.

It was not a strategy. It was a collection of goals, aspirations, and initiatives stapled together into a document that described the destination without ever confronting the actual problem. Kodak had been the company that invented the digital camera, in 1975, in its own research labs. It had owned the patents. It had understood, earlier than almost any competitor, what the technology could become. And yet the company that filed for bankruptcy protection in 2012 had failed to answer the one question that any coherent strategy must answer: given what we know about where the world is going and what we can actually do, what are we going to stop doing in order to do the things that matter?

Kodak's binders talked about transformation. They did not talk about cannibalising the film business that paid for the dividends. They did not talk about choosing between being a consumer brand and being a business-imaging company. They did not talk about the uncomfortable reality that the digital future would reward companies good at software and networks, not companies good at coating silver halide onto celluloid. In the vocabulary of Richard Rumelt, Kodak had a "bad strategy" — a document full of ambition, goals, and buzzwords, produced in place of the harder work of diagnosis and choice. The failure was not that Kodak could not predict the future. The failure was that its strategy documents never contained a strategy in the first place.

This pattern is the starting problem for marketers. Most of what gets called "marketing strategy" in most organisations is not strategy at all. It is an artefact — a PowerPoint, a binder, an annual plan — that describes where the company would like to end up and lists the initiatives it hopes will get it there. It is produced at an offsite, signed off by the board, filed in a shared drive, and largely forgotten until the next offsite. The tactical plan is where the real decisions happen, often with little reference to the strategic binder that sits above it. The strategic document is, in effect, a ritual rather than an act of thought. Before we can talk about how to do marketing strategy well, we have to be honest about how thoroughly the word itself has been hollowed out.

Two ways of being wrong about strategy

There are two long-standing and opposing views about what strategy actually is, and both of them contain a serious mistake worth understanding on its own terms. The first is the planning view. In its strong form, strategy is a document produced once a year by a dedicated process — often owned by a strategy function, a consulting firm, or a strategy-and-insights team inside marketing. It is comprehensive. It covers the market, the competitors, the customers, the portfolio, the brand, the channels, and the finances. It sets objectives, cascades them into KPIs, and attaches owners and deadlines to initiatives. The planning view treats strategy as a predictable output of a formal process — the sort of process that can be scheduled, resourced, and reviewed. Its animating assumption is that careful analysis plus disciplined documentation plus annual governance equals a good strategy.

The planning view is not silly. It has real virtues. It forces an organisation to step back from the noise of daily operations and think about the longer horizon. It creates an artefact that can be shared, challenged, and referenced. It imposes a rhythm of review that, in principle, allows a firm to update its thinking as conditions change. Drucker (1954) defended something like the planning view — and his best-known dictum, that the purpose of a business is to create a customer, is itself a strategic position stated with total clarity. Porter (1996), in "What Is Strategy?", is not dismissive of planning either. His argument is not that planning is pointless. His argument is that most companies confuse planning with strategy, and confuse strategy with operational effectiveness.

The emergent view is the other extreme. Mintzberg's work, especially "The Rise and Fall of Strategic Planning" (1994) and "Strategy Safari" (1998, with Ahlstrand and Lampel), made the case that real strategy is not what sits in the binder. Real strategy is the pattern that actually emerges from a stream of decisions, adjustments, and accidents. If you read the binder of any company that succeeded over a decade, Mintzberg argued, you will not find the real story of how that company actually won. The real story is the thousand small choices — about which customers to pursue, which products to kill, which hires to make, which agency to fire — that added up, in retrospect, to something coherent. Honda's entry into the US motorcycle market in the 1960s, which Mintzberg analysed in detail, was not the deliberate plan of the strategic binder; it was a series of tactical improvisations that only looked like a strategy in the telling.

The emergent view is not silly either. It contains an important empirical observation: most of the formal strategy documents produced in the world bear only a passing resemblance to what their companies actually did. And it contains a useful warning: any strategy process that produces a beautiful binder and then outsources the doing to a different group of people is likely to produce a beautiful binder and not much else. Mintzberg's five Ps — strategy as plan, ploy, pattern, position, and perspective — were designed precisely to make room for the idea that strategy could legitimately be discovered through action as well as designed through analysis.

The strong form of each view, however, is a mistake. The strong form of the planning view gives you Kodak — a binder without a diagnosis. The strong form of the emergent view gives you the opposite failure mode: a company whose "strategy" is whatever pattern the accountants are able to reconstruct after the fact, with no intentional choice at the centre of it. Neither produces strategy in the sense that Rumelt, Porter, or Ritson would recognise. To get there, we need a more useful definition of the word.

Rumelt's kernel: diagnosis, guiding policy, coherent actions

Richard Rumelt's "Good Strategy Bad Strategy" (2011) is the most useful thing written about strategy in the last two decades, partly because its central move is so unglamorous. Strategy, Rumelt argues, is not a goal, an ambition, or a motivating statement. It is not a list of priorities. It is not a set of values. It is a specific, three-part structure he calls the kernel. A good strategy contains a diagnosis of the situation, a guiding policy that responds to the diagnosis, and a set of coherent actions that carry out the guiding policy. Everything else is optional. If any of these three are missing, what you have may still be valuable — a useful plan, a stirring ambition, an operational roadmap — but it is not a strategy.

The diagnosis is the hardest of the three and the one that most "strategy" documents skip. A good diagnosis names the critical feature of the situation. It reduces the overwhelming complexity of reality to a simplified understanding of what is actually going on and where the central difficulty lies. Theodore Levitt's "Marketing Myopia" (1960) — arguably the most important single article in the history of marketing — was itself a diagnostic act: Levitt's insight was that the railroad companies had diagnosed their situation wrong, defining themselves by product (railroads) rather than by customer need (transportation). Every strategic failure begins with a bad diagnosis, and the most common bad diagnosis is the one Levitt identified: mistaking what you make for what you do. It is, in Rumelt's words, a judgement about which aspects of the situation matter most. Diagnosis is not research. Research is the gathering of inputs. Diagnosis is the interpretive act that says: here is what the research is telling us, here is the critical constraint, here is where we will focus. A firm can have thousands of data points and no diagnosis. It can also — more rarely, but it happens — have relatively thin data and a genuinely sharp diagnosis that organises everything that follows.

The guiding policy is the overall approach chosen to deal with the obstacles identified in the diagnosis. It is not a specific action. It is a direction — a set of principles that, once adopted, rules a large class of actions in and another large class of actions out. Porter's generic strategies (cost leadership, differentiation, focus) are guiding policies in this sense: once you have chosen differentiation as your guiding policy, a range of tactical decisions about price, channel, product, and brand make sense in one way and do not make sense in another. A guiding policy is not the strategy. It is the spine of the strategy. The coherent actions are the specific, concrete, resource-committing moves that carry the guiding policy into reality. And the test of coherence is simple: do the actions reinforce each other, or do they cancel each other out?

A worked example makes this concrete. Take a mid-sized consumer goods brand that has been losing share to private label in a saturated category. A bad-strategy response looks like this: "Our ambition is to return to growth. Our strategy is to launch innovative new products, reinvigorate the brand, and improve operational efficiency, while delivering shareholder value." That sentence is entirely aspirational. It has no diagnosis. It has no guiding policy, because every possible action is still permitted. It has no coherent actions, because the "strategy" has not narrowed the action space at all.

The Rumelt-kernel response looks like this. Diagnosis: the brand's share loss is not a general brand-health problem but a specific failure of distinctiveness in the mainstream tier, where private label now matches quality at a significantly lower price. The premium tier, where distinctiveness remains intact, is stable and profitable. Guiding policy: concentrate resources on defending and extending the premium tier, retreat from direct price competition in the mainstream, and accept a smaller, more profitable portfolio. Coherent actions: discontinue three weak mainstream SKUs, redirect their marketing budgets into the two premium lines with highest margin, reposition the brand architecture around premium leadership, negotiate with retailers for improved premium shelf space in exchange for accepting smaller mainstream facings. That is a strategy. It is not a good or bad strategy in the abstract — it might be wrong — but it is a strategy in the sense that it has done the work of diagnosis, commitment, and coherent action that the kernel requires.

Porter, Mintzberg, and the ground truth

Rumelt's kernel is powerful because it reconciles what Porter and Mintzberg each got right. Porter (1996) insisted that strategy is fundamentally about trade-offs — about doing different things or doing things differently from competitors, and accepting that you cannot be everything to everyone. Operational effectiveness, he argued, is not strategy. Benchmarking competitors and matching their best practices produces a race to the middle, not an advantage. What produces advantage is a distinctive activity system — a set of interlocking choices that together create a position competitors cannot easily copy, because copying one element in isolation does not work. The Rumelt kernel incorporates this: the guiding policy is precisely the choice of position, and coherent actions are the interlocking activities that produce it.

Mintzberg's contribution is the recognition that no plan survives first contact with reality, and that the best organisations treat their strategy as a live thesis rather than a filed document. The five Ps argument is that strategy can legitimately mean several different things — a plan you make, a ploy against a competitor, a pattern that emerges from actions, a position in the market, a perspective shared by the firm. Real strategy, in Mintzberg's telling, is typically some mixture of the deliberate and the emergent: you set out with a plan, you learn from execution, you update your plan, you keep moving. Rumelt is compatible with this. The kernel does not require omniscience; it requires clarity about where you think the central difficulty lies and what you are choosing to do about it today, with a readiness to rediagnose when the evidence demands.

What the academic debate can obscure is that, for the working marketer, the useful synthesis was already sitting in Mark Ritson's Marketing Week columns and his Mini MBA teaching. Ritson's triptych — diagnosis, strategy, tactics — is essentially Rumelt translated for the marketing practitioner, with Porter's trade-off discipline built in. You begin with diagnosis: a rigorous, multi-month assessment of the market, the customer, the competitors, the brand, and the company. You move to strategy: a small number of choices about where to play and how to win, built on the diagnosis. You end with tactics: the specific 4P or 7P decisions that carry the strategy into the market. Ritson's argument, made repeatedly in his columns and teaching from roughly 2015 onwards, is that most marketers collapse the first two stages into each other, skip straight to tactics, or produce a strategy that is really just a dressed-up list of tactics. The triptych is a forcing function against that mistake.

The effectiveness evidence supports the triptych. McKinsey's strategy-execution research, replicated in various forms from 2018 onwards, has consistently found that around seventy per cent of strategies fail in execution — which is to say that the link between the binder and the market is broken somewhere in the middle. The IPA Effectiveness Databank, Les Binet and Peter Field's long-term case data, tells a more hopeful version of the same story: campaigns grounded in a clear strategic diagnosis and a defensible guiding policy produce long-run effects that are an order of magnitude larger than campaigns produced without that discipline. The binder alone does not work. Tactics alone do not work. The diagnosis-strategy-tactics chain is what works, and the reason it works is that it forces the coherence Rumelt describes and the emergent learning Mintzberg demands.

Three companies illustrate the payoff. Apple's strategy after Steve Jobs returned in 1997 was not a traditional plan; it was a working hypothesis — that consumer technology would reward elegant, integrated hardware-software-services experiences in a small number of categories, and that Apple should retreat from its sprawling product line to focus on that. The initial coherent actions (killing Newton, killing most of the Mac line, focusing on the iMac and then the iPod) were a strategic commitment in exactly Rumelt's sense. The subsequent twenty years of emergent learning — from the iPod to the iPhone to the App Store to Services — were Mintzberg's updating-in-action. Tesco's turnaround under Dave Lewis from 2014 to 2021 began with an explicit diagnosis (too many product lines, a damaged balance sheet, a loss of price credibility against Aldi and Lidl) and a guiding policy (simplify the portfolio, restore trust on price, exit ventures that did not earn their place) that then generated years of coherent action. And LEGO's 2004 turnaround under Jørgen Vig Knudstorp began — famously — with Knudstorp telling the board that the company had "too many products, in too many categories, for too few customers", a diagnosis so clear it did most of the work the guiding policy and coherent actions needed to do next.

What this means in practice: three diagnostic questions

If strategy is a decision problem under uncertainty, and the Rumelt-Ritson synthesis is the right way to think about it, then the working marketer has a concrete task. Whenever you are asked to read, write, or defend a "marketing strategy" document, you should put it through three diagnostic questions. Each question targets one part of the kernel, and each one is designed to reveal the places where strategy has been replaced by aspiration.

The first question is the diagnosis question. What does this document say is the central problem we are trying to solve, and do I believe the diagnosis is right? A good strategy document tells you, in one or two sentences, what the critical feature of the situation is. It names the constraint. It is specific enough to be wrong. If you cannot find the diagnosis — if the document jumps straight from "market context" to "strategic priorities" without telling you why those priorities and not others — you are reading an aspirational document, not a strategy. Ritson's standing complaint about most marketing strategy decks is that they skip this step entirely, producing a set of priorities that look like choices but are actually a list of things the team would like to do.

The second question is the trade-off question. What does this document say we are going to stop doing, and what are we deliberately choosing not to pursue? Porter's core insight is that strategy is trade-off, and a document that promises everything to everyone is, by definition, not strategy. If the plan includes growing in every segment, launching in every category, investing in every channel, and improving on every KPI, then the strategy has not been made. The hard choices have been deferred. A useful strategy document contains sentences starting with "we will not" — and the "will not" list is at least as important as the "will do" list. If you cannot find the "will not" list, ask the author to write one. Watch carefully what happens.

The third question is the coherence question. Do the actions in this document reinforce each other, and would an external reader be able to see the guiding policy from the actions alone? This is the Rumelt coherence test applied backwards. Take the list of initiatives. Cover up the executive summary. Would you be able to infer the guiding policy from the actions? If the actions look like a portfolio of unrelated initiatives — some brand investment here, some performance media there, a new-product push, a pricing review, a CRM overhaul — without an organising thread that makes them all part of one argument, then the strategy has failed the coherence test. Coherent actions need not all sit inside the same function or the same P. What they need is to be explicable by the same underlying policy, so that moving forward on one of them makes the others more likely to succeed rather than less.

These three questions do not tell you whether a strategy is right. They tell you whether it is a strategy at all. Many documents that are celebrated as bold strategic visions turn out, under the questions, to be collections of goals and initiatives with no diagnosis, no trade-off, and no coherence. The honest response to that finding is not to reject the document but to go back and do the work that was skipped. Rediagnose. Make the trade-off. Rewrite the actions until they reinforce each other. This is, in a very practical sense, what strategic thinking feels like when it is being done rather than performed.

The Both/And core

Marketing strategy is not a choice between a deliberate plan and an emergent pattern. It is both at once. The planning view is right that strategy requires deliberate thought, explicit documentation, and a decision architecture that can be defended to a board and communicated to a team. The emergent view is right that any plan written today will be wrong in some of its details by next quarter, and that the organisations that win are the ones that treat their strategy as a live thesis rather than a filed artefact. Neither view, in its pure form, produces what marketers and their firms actually need.

The synthesis is that strategy is a decision problem under uncertainty, and the Rumelt-Ritson kernel is the disciplined way of framing it. You commit to a diagnosis of where the critical difficulty sits. You adopt a guiding policy that rules in a small number of actions and rules out many more. You carry it out through coherent actions that reinforce each other. And then, because the world moves, you expose the whole structure to execution, collect the evidence of what worked and what did not, and update the diagnosis when the evidence demands it. Deliberate and emergent — both, not either.

This is why the evidence-based marketer is suspicious of strategy documents that have no trade-offs and equally suspicious of firms that claim not to need documents at all. The binder without a diagnosis is not strategy. The pattern without a thesis is not strategy either. Strategy is the disciplined reconciliation — a working hypothesis about how to win, sharp enough to be wrong, flexible enough to be updated, and concrete enough to drive the tactical decisions that follow. Everything in the rest of F12 — the diagnosis, the planning cycle, the objectives, the STP core, the budget, the tactical bridge, the execution quality, the measurement, and the capstone — is built on this foundation. Get this first thing right, and the rest of marketing becomes explicable. Get it wrong, and no amount of tactical brilliance downstream will rescue you.