What Is Strategy 2026: The Two Sources That Count It

The strategy definition, from the author's own institute

The shortest useful answer comes from the page the Institute for Strategy and Competitiveness maintains at Harvard Business School, which is Michael Porter's institute. It gives three statements in sequence.

First, what strategy produces: "Strategy is the creation of a unique and valuable position, involving a different set of activities."

Second, what it costs: "Strategy requires you to make trade-offs in competing - choosing what not to do."

Third, what holds it together: "Strategy involves creating "fit" among a company's activities."

Then the page fixes the objective, which most definitions leave out: "The fundamental goal of a company is superior long-term return on invested capital (ROIC)." And it puts a condition on the metric everyone reaches for first: "Revenue growth is good only if superiority in ROIC is achieved and sustained."

The page also corrects the question itself. It identifies the wrong question as asking how to be the best, on the grounds that being the best leads to competition where everyone chases the same position and nobody earns anything durable.

One structural claim closes it, and it is the one the second source will argue with: "While both are essential, business units typically account for 90% or more of economic performance". Two levels only, corporate and business unit, and the decisions that matter sit at the lower one.

The exclusion test, which is one question long

Everything in the section above collapses into a single test a reader can apply in thirty seconds to any document handed to them.

Does it say what the organization will not do?

The reason this works is the trade-off clause. If a document commits to a unique position involving a different set of activities, then some activities are excluded by construction. A document with no exclusion in it has not chosen a position, which means it is a list of intentions that happens to be labeled strategy.

Three things fail that test routinely and all three get presented as strategy. A list of initiatives contains no exclusion, because each initiative is additive. A set of financial targets contains no exclusion, because a target says where to arrive and not what to give up. A mission statement contains no exclusion, because it describes a purpose rather than a choice.

The test is also why interviewers push where they push. Asked for a recommendation, a candidate who names the cost of it has passed the exclusion test out loud. A candidate who names three initiatives has produced a plan. On our own synthesis scale, naming the risk yourself carries 20 of 100 and answering first carries 35, and those weights are ours rather than any firm's.

One caution about the test. It tells you whether a choice was made. It says nothing about whether the choice was right, and the second source is where that question gets numbers.

What one filed letter looks like when the trade-off is written

The exclusion test is easier to trust when you have seen a document pass it. One exists, it is signed, and it was filed with a regulator, which makes it datable and citable rather than anecdotal.

Amazon's 1997 letter to shareholders, reproduced as an exhibit in a 2001 filing, writes the trade-off out in four sentences.

"Because of our emphasis on the long term, we may make decisions and weigh tradeoffs differently than some companies."

"We will continue to make investment decisions in light of long-term market leadership considerations rather than short-term profitability considerations or short-term Wall Street reactions."

"When forced to choose between optimizing the appearance of our GAAP accounting and maximizing the present value of future cash flows, we'll take the cash flows."

"At this stage, we choose to prioritize growth because we believe that scale is central to achieving the potential of our business model."

Three exclusions in there, stated as exclusions. Short-term profitability is given up. The appearance of the accounts is given up. And the fourth sentence names the phase, "at this stage", which is the part most readers skip and the part that makes it a decision rather than a creed.

Label this correctly, though. It is one company, one year, one emerging market, written by an interested party addressing its own shareholders. A shareholder letter is an act of communication as much as a record. What it demonstrates is the shape of a written trade-off, at a moment when almost every page on this query asserts that one is needed without showing what one looks like.

The study that prices strategy

The second source is a 2018 McKinsey Quarterly article built on a database of large companies, and it is the only document in our reading that attaches probabilities to strategic choices.

Its base: "From 2010 to 2014, the average company in our database of the world's 2,393 largest corporations reported $920 million in annual operating profit." That average company "used $9,300 million of invested capital, which earned a return of 9.9 percent. After investors and lenders took 8 percent to compensate for use of their funds, that left $180 million in economic profit."

We redid both operations. Dividing 920 by 9,300 gives 0.09892, so 9.9 percent, which holds. Eight percent of 9,300 is 744, and 920 minus 744 is 176, so the economic profit is 176 million dollars. The article prints 180 million, which is 176 rounded, and we say so rather than reprinting 180 as exact.

Then the distribution, which is the finding. "Companies in the top quintile capture nearly 90 percent of the economic profit created, averaging $1.4 billion annually." And the middle: "Companies in those three quintiles delivered economic profits averaging just $47 million a year."

Dividing 1,400 by 47 gives 29.8, so the top quintile averages about thirty times the middle three. That ratio is the reason the word is worth arguing about at all.

Five published moves, with the thresholds that count

The study's usable part is that it defines what counts as a big move, numerically, instead of leaving it to judgment.

Four of the five are described as asymmetric, meaning they raise the chance of rising more than they raise the chance of falling. Capital expenditure is the exception, because heavy spending also raises the odds of dropping.

Then the probabilities. "For companies in the middle quintiles, pulling one or two of the five levers more than doubles their odds of rising into the top quintile, from 8 percent to 17 percent. Three big moves boost these odds to 47 percent."

Checking the commentary against the arithmetic matters here, because the article's own phrasing invites a slip. Twice 8 is 16, so "more than doubles" holds for a move to 17. But 47 is not three times 17; dividing 47 by 8 gives 5.875, so three moves multiply the base rate by nearly six. The jump from 17 to 47 is the large one, and a reader who assumes the steps are even will get that backwards.

One worked example is published with its odds attached. A named company "reallocated 61 percent of its capital spending among its three major divisions", and "Based on the moves PCC made, its odds of rising to the top were 76 percent."

What the study says you do not control

The same article divides the outcome into parts, and the part attributable to choice is smaller than a reader hoping for agency would like.

"we found that endowment determines about 30 percent and trends another 25 percent." Endowment is where you start, in revenue, debt and past investment in research. Trends are the industry and geography you are exposed to.

Adding those gives 55 percent. The tempting next step is subtraction, and it is wrong. The article says the levers explain "more than 80 percent of the up-drift and down-drift in corporate performance", which is a statement about drift rather than about position, and the two are not the same quantity. So we do not write that moves account for 45 percent. That subtraction would be arithmetic performed on two different measurements.

The industry claim is blunter: "about 50 percent of your position on the curve is driven by your industry". And it comes with a count that makes the point better than the percentage does. "There are 12 tobacco companies in our research, and 9 are in the top quintile. Yet there are 20 paper companies, and none is in the top quintile."

Nine out of twelve is 75 percent. Zero out of twenty is zero percent. Same study, same window, same measure of economic profit, and the choice of industry did more than anything a paper company's management could have decided.

One figure in the article is second-hand and has to be labeled. It reports that "80 percent of executives believe their product stands out against the competition-but only 8 percent of customers agree", and attributes it in a footnote to a 2007 book by two other authors. It is a citation rather than a finding of the study, and it is quoted here as one.

Where the two sources disagree, and why both are right

Putting the two documents side by side produces an apparent contradiction that dissolves on inspection, and understanding why is more useful than either source alone.

The first row is a genuine difference of measure. Return on invested capital is a rate, so a small business earning 50 percent on a small base beats a large one earning 12 percent on a huge base. Economic profit is a rate multiplied by capital employed, so the large one wins. The study makes this explicit itself, comparing a 12 percent return on 136 billion dollars of capital against a 50 percent return on 2.6 billion. Neither measure is wrong and they rank companies differently, so this page publishes both rather than electing one.

The second row is not a contradiction at all. Porter's claim is about where economic performance is generated, at the business unit rather than the corporate center. The study's claim is about what explains a company's position on a distribution, which includes the portfolio choice of which industries to be in. Stacking the two without saying so manufactures a conflict that neither author is having.

Four steps to test any strategy document

In order, and the first two are the ones that fail most documents.

1. Find the exclusion. Read for what the organization says it will not do. No exclusion means no position was chosen, whatever the document is called.
2. Find the objective and check the metric. Porter's institute puts the goal at return on invested capital and states that revenue growth counts only if superiority in that return is achieved and sustained. A document whose only target is growth has not met its own test.
3. Check whether any move crosses a threshold. Reallocating half of capital expenditure across a decade is a published bar. Moving 5 percent is an intention. The five thresholds above give you numbers to hold the document against.
4. Ask what the industry contributes. Roughly half of a company's position is attributed to the industry it sits in. A document that plans to outperform without saying anything about which industries it is in has left out the larger term.

A fifth step belongs to the reader rather than the document. Check the date. The probabilities above come from a window ending in 2014 and nobody has republished them on a recent window, so treat them as an order of magnitude rather than a current forecast.

What our own drills score in a strategy answer

No firm publishes a rubric for any of this. The weights below are ours, set by us and read from our own code.

The Risk row is where this subject and our scale meet exactly. The exclusion test asks what a document gives up. Twenty of a hundred points on our synthesis scale go to naming that yourself, which also removes the interviewer's best follow-up question.

The synthesis drill targets that scale directly, the structure drill covers the diagnosis that precedes it, and how an answer is assessed sets out each weight with the reasoning behind it. Neither drill teaches strategy content, and the two sources above are free to read in full.

What neither source publishes

Five gaps, and the first one is the reason to treat the numbers as an order of magnitude.

Whether the probabilities still hold is unknown. The study covers 2010 to 2014 and nobody has rerun it on a later window. A ten-year odds figure computed on a window that closed twelve years ago is perishable, and this page says so rather than quietly presenting it as current.

The 2,393 companies go unnamed, so no reader can reproduce the calculation. That is the largest limitation on the whole study, and the article does not address it.

Neither source says anything about companies that are not large and listed. Both datasets are big public companies, so nothing here is evidence about a private business or a small one.

Nobody publishes a threshold at which a plan earns the word strategy. We do not invent one, which is why this page gives an exclusion test that a reader can verify rather than a bar we made up.

And the word itself does not predict who sells it. One economic consultancy describes its own work as guiding "corporations through critical business strategy and performance-related issues" while also advising on litigation, and its peer publishes a behavior-based first round and a research presentation rather than a strategy case. A firm with the word in its offer does not necessarily run the exercise the word implies.

For what the categories of firm actually are and which documents separate them, the types of consulting firm keeps only the ones a filing supports. For the frameworks that operationalize the definition above, the Porter five forces page works one example end to end.

What changed on this strategy page, and when

1 October 2026. This replaces a 227-word page. Its definition was defensible and it had no source under any of it.

The working definition on the old page, about where to compete and how to win given limited resources, has been replaced by the definition Porter's own institute publishes, in his institute's words. That is the same idea with an owner attached, and the owner matters on a page whose whole subject is a contested word.

Gone: the assertion that a strategy requiring you to do everything is not one. The idea survives, sourced, through the trade-off clause and the exclusion test. Gone too: the flat claim that a list of initiatives, a set of targets and a mission statement are routinely presented as strategy and contain no choice. The second half is now derived from the trade-off clause rather than asserted, and the word routinely is gone, because we have not counted documents.

Added in this version: the definition and objective from the author's own institute, one filed shareholder letter where the trade-off is written out and labeled for what it is, a study of 2,393 companies with the economic profit figures recomputed by hand and the 176 against 180 rounding stated, five moves with their published numeric thresholds, the odds table with the warning that the steps are uneven, the shares attributed to endowment and trends with an explicit refusal to subtract them, nine out of twelve against zero out of twenty, and the two places where the two sources measure different things.