In this week’s edition
Thematic Strategy: How to tell a structural change from an expensive distraction
The Art of Leadership: Three bad quarters are not proof the strategy is wrong
Operating at Scale: Stop building the same fault into every site you run
Read time: 7 minutes
THEMATIC STRATEGY
How to tell a structural change from an expensive distraction
In July, Gallup reported that 11% of American adults said they were currently taking a GLP-1 medication for weight loss, against 3% in 2024. The survey ran across 5,065 adults.
A driver is a change happening outside your firm, whether you respond to it or not. Some look interesting but, in reality, are fads.
I use three tests to distinguish a driver from a fad, and a candidate has to pass all three.
Structural shift. Has it grown for ten years or more, at a rate well above GDP growth, through an economic cycle?
Cross-industry relevance. Does it reach well beyond your own industry? A change that reaches six industries gives you six ways to apply it and a position you can hold for a decade.
Actionability. Can you specify the capital and the hires that follow from it? This is where most candidates are eliminated.
Now run GLP-1 drugs through that.
The first test is closer than the headlines suggest. The consumer surge is about three years old, which on its own would not qualify, but the drug class has been growing since the mid-2000s and the science underneath it for longer.
It clears the second test comfortably. Sylvia Hristakeva and Jura Liaukonyte of Cornell University’s Dyson School, publishing in the Journal of Marketing Research in December 2025, tracked around 150,000 American households through Numerator’s transaction data. Within six months of one household member starting the medication, grocery spending fell 5.3%, with savoury snacks down about 10% and limited-service restaurants down about 8%. That reaches grocery, packaged food and restaurants, and from there into clothing and insurance.
Then actionability. “GLP-1 drugs are changing consumer behaviour” commits you to nothing and will appear in a hundred annual reports this year. “Reformulating our top twenty lines for higher protein and smaller portions” is a strategic decision.
Leadership teams that run this exercise usually find one or two real drivers buried under a longer list of priorities that feel strategic and are connected to nothing outside the firm.
Put this to yours, in these words:
Name the change outside this business that our largest investment this year will leverage.
If you have a clear answer, you have a top priority.
THE ART OF LEADERSHIP
Three bad quarters are not proof the strategy is wrong
In 2003, John Graham and Campbell Harvey of Duke University, with Shiva Rajgopal, then at the University of Washington, put a blunt question to 401 financial executives. Would you reject a project you knew would create value, if taking it meant missing this quarter’s earnings consensus?
55% said they would.
These were not careless managers. Their interviews found people who disliked the system and felt compelled to play it, on the view that refusing would cost them their positions.
I share this because it describes a failure that is usually misdiagnosed. When a company abandons a sound strategy early, the post-mortem blames the analysis. The plan was wrong, or the assumptions were optimistic. The survey suggests something else. A majority of senior financial executives will knowingly forgo value because the cost of holding the course impacts results now, and the benefit arrives long into the future.
That is a failure of patience. Abandon a long-horizon investment, and you can point to an immediate improvement in the numbers, and you will be called decisive for it. Hold the course, and you have continued spending and a claim about the future that nobody can verify.
Abandonment is rarely described as a loss of nerve. It gets described as discipline, or as a sensible reallocation towards things with clearer returns.
Here is the distinction that decides it.
A strategy that is failing has had its underlying premise tested and beaten. Something you said would be true turned out not to be. A strategy that is costing is producing exactly the unattractive numbers you said it would produce, in exactly the interval you said it would produce them. Three bad quarters, in a period the strategy always said would be bad, is evidence only that the strategy is proceeding as expected.
The trouble is that the two feel identical while you are living through them. Both hurt. Both produce a persuasive case for stopping, built from real numbers. And you will be under pressure to decide at precisely the moment you are least able to tell them apart.
So decide earlier. When you commit, write two sentences and circulate them.
The cost. What the numbers will do, and for how long. We expect margin to fall for six quarters and recover in the seventh.
The premise. What has to be observable, by when, and what happens if it is not. We expect the first two customer wins by the end of year two. If we have none by then, the premise has failed and we stop. Falling margin on its own is not a reason to stop, because we have said here that we expect it.
That takes an hour, and it converts every later argument from a contest of nerve into one of fact.
Take your longest-horizon commitment and ask your leadership team, in these words:
What would have to happen for us to stop this?
If nobody can answer, you do not have a patient strategy. You have an unresolved argument that will be settled by whichever quarter happens to be bad enough.
OPERATING AT SCALE
Stop building the same fault into every site you run
More than 23 million vehicles were recalled in the United States in the first half of this year, on National Highway Traffic Safety Administration figures. That is more than double the same period in 2025.
In the second quarter, 11.1 million vehicles were recalled across 99 separate campaigns. That is around 112,000 vehicles for each defect.
One manufacturer accounts for most of it. Ford issued 55 campaigns covering more than 12 million vehicles in six months, and in February it recalled 4.3 million American vehicles over a software error in the integrated trailer module that could disable trailer brakes and exterior lights. Ford reported 407 incidents that may have been related, and no known crashes. It began sending the correction over the air three weeks later.
A trailer module common to the F-150, the Expedition, the Navigator, the Maverick and the E-Transit is a considerable achievement. It is cheaper to build and support, and it is why the repair could be completed without anyone driving to a dealer. It also means one engineer’s mistake was present, identically, in 4.3 million vehicles.
Standardisation is right, and I would make the same choice. It converts a large number of small local failures into a small number of enormous ones. That trade is worth making, but it has to be costed.
So before you make anything identical across your estate, work through four checks.
Population. How many units will carry this component, this process or this code once the change is complete? Write that number down. It is how many can fail at once.
Detection. How would you find out it is wrong, and how long would that take? A fault common to every site produces nothing to compare, because the site that would have looked different now looks the same.
Remedy. Can you correct all of them at once, or does each one have to be visited? Ford could send most of them a file. If your correction needs a person to attend each location, your exposure is the population multiplied by the travel.
Reversal. Can you put the previous version back, and has anyone done it? If nobody has ever run the rollback, you do not know whether it works.
Detection is the one to build first, because it is the only one of the four that has to exist before the fault does.
So put this to the person who owns the standard:
If this went wrong everywhere at once, which report would show it first, and when did we last see that report move?
None of this tells you whether the change is a good idea. A large population count is not an argument against standardising, and firms that refuse to standardise because the exposure frightens them end up with hundreds of small failures they never count and cannot fix. The four checks tell you what you are buying.
The recall figures need care too. Recalls have risen partly because faults are being found, and partly because an over-the-air correction is cheap enough that a manufacturer will now issue a recall it would once have absorbed quietly. A rising number is not straightforwardly a falling standard.
Sources
Thematic Strategy: Gallup, GLP-1 usage survey, July 2026; Sylvia Hristakeva and Jura Liaukonyte, Cornell University, Journal of Marketing Research, December 2025
The Art of Leadership: John Graham, Campbell Harvey and Shiva Rajgopal, National Bureau of Economic Research working paper 10550, 2004; Duke University’s Fuqua School of Business on the authors’ interviews, February 2004; Ørsted’s impairment on Ocean Wind 1 and 2, reported by CNBC, November 2023
Operating at Scale: National Highway Traffic Safety Administration recall data, compiled in BizzyCar’s second-quarter 2026 recall report; Ford, recall 26C10 statement, February 2026