McDonald's Corporation

On 15 July we wrote that McDonald's is not a restaurant company but a royalty-and-rent stream on other people's sales, that it sat at a 52-week low out of boredom rather than arithmetic, and that the right thing to do was buy the landlord and add below $250. Eight weeks and one earnings report later, the honest thing to do is grade that call in public — because a verdict that cannot be checked is a slogan. Here is the ledger.

Two of seven wrong, both on the same subject: we underestimated how much the value strategy depends on the goodwill of the people who fund it. Nothing we said about the architecture was wrong — Part IV will show the landlord's margin barely noticed a quarter in which the tenants' customers stayed home. But the architecture only earns what the tenants execute, and in Q2 a third of them declined. That is why the moat dial moves from 9 to 8 in this report, and why the global score drops from 7.8 to 7.3 while the verdict does not change. Price fell 4%; quality fell a notch; the trade is roughly where it was, now with the add-zone in reach.

The long version lives in our July report: the brothers invented fast food as a process, Ray Kroc turned it into a franchise system, and Harry Sonneborn turned the franchise system into a real-estate empire — "We are not technically in the food business. We are in the real estate business." What matters here is the pattern the history keeps repeating, and the new chapter.

The lesson is the one 2003 taught: this company's crises are self-inflicted and operational, never architectural. Overbuilding in 2002, over-pricing in 2021–23, over-promoting in 2026 — each time the cure was the same: stop, simplify, let the system's economics do the work. The new chapter fits the pattern exactly. That is comforting about the business and slightly less comforting about the people running it, which is where the management dial goes in Part VIII.

FY2025 reported revenue by segment, each dollar sitting on roughly five dollars of system-wide customer spending:

Read Q2 as a stress test of the architecture, because that is what it was. Third-party foot-traffic data put U.S. visits down 4.5% year on year, worse than the industry's −3.0% (approx. — Placer.ai). Comparable sales still printed +0.8%, because the check rose more than the guest count fell. And consolidated operating income rose 3% to $3.34B on revenue of $7.10B (press release) — a 47% operating margin in the quarter the company itself called a failure of execution. This is the Sonneborn architecture doing exactly what it was built to do: royalty and rent are percentages of the tenants' sales, not of their profits or their traffic. Menu inflation lifts the landlord's take; a value war that thins the franchisee's margin does not thin corporate's.

Which is also the political problem in one sentence. Since 2020 menu prices are up roughly 40% (approx. — industry estimates); the landlord captured the upside of every price increase, and is now asking the tenants to fund the discounts that win the traffic back. Corporate co-funded the Extra Value Meal relaunch "through early 2026" (approx. — dated press reports) and then stopped. From the franchisee's chair, that is the whole dispute: you got the inflation; we get the value war. The digital layer cuts the same way. About 220 million 90-day-active loyalty members drove more than $40B of trailing system sales, up 20% (press release) — the largest first-party customer dataset in food service — and Q2 showed its flip side: when the company pulled the app's national offers to launch a plain under-$3 menu, the deal-trained customer simply stayed home.

In July we scored the moat 9, "narrower only than Coca-Cola's, and made of land." The land is still there. What Q2 exposed is that the moat has an execution layer — two thousand-odd owner-operators — and that layer is a counterparty, not an asset. When a third of it declines to follow the price list, the brand's promise to the customer ("value, everywhere, the same") is broken by the very people who deliver it, and no amount of corner-ownership fixes that quarter. Morgan Stanley's phrase from its 2025 downgrade — "pricing power has eroded" — is too strong for the landlord and about right for the tenant. Score: 8. Wide, structural, and — for the first time in our coverage — visibly dependent on a negotiation.

In July the central question was appetite, in two forms — the squeezed wallet and the GLP-1 molecule. Both are still there. But the quarter added a third, more immediate one, and management stated it better than we could: "We don't have a strategy problem. We simply didn't execute."

What actually happened, in the company's own accounting. In Q2 the U.S. business rolled out an everyday-affordable menu of roughly ten items under $3, and at the same time pulled back national digital offers and the "Buy One, Add One for $1" deal. Kempczinski called the swap "a bad trade." The CFO attributed about two-thirds of the U.S. traffic shortfall to that trade and about one-third to a FIFA World Cup campaign that "underperformed versus our expectations" (approx. — Q2 call). Restaurant teams were, in the CEO's words, overwhelmed by too many deployments — a K-pop promotion, McValue changes, the new menu, a beverage platform in May, the World Cup in June — with campaigns shifting every two to four weeks. And underneath all of it: only 60–65% of the system priced the way corporate recommended, with some operators marking small items up.

Why "execution, not strategy" is both true and insufficient. True, because nothing in the quarter argues against the model: the franchised margin held, the international markets that ran the playbook (Germany, Australia, UK) grew, and the beverage platform — CosMc's drinks brought home — is running checks ~50% above the daily average with more than half its traffic after lunch (approx. — call). Insufficient, because "execution" at McDonald's means the franchisees, and the company's chosen remedy is a stick: pricing non-compliance now enters business reviews that govern growth and renewal eligibility. That is the sentence that turns a marketing mistake into a legal risk. The tenants have counsel, a Bill of Rights, and a survey claiming 40% of them would not meet the company's own renewal financials (approx. — Restaurant Business).

Our read, updated: the wallet question is cyclical (consumer sentiment near record lows this summer — approx.); the execution question is fixable and already being fixed, but the fix runs through the franchisee relationship, which is now the single most important variable in the stock; and the molecule question is secular and honestly unanswerable — the company's silence on it this quarter is a choice, not evidence. The growth dial comes down to 5 because the unit target slipped a year and July went negative. None of the three questions threatens the royalty percentage, the seniority of the rent, or the corners.

Two patterns. First, the arena moved against McDonald's for one quarter, and it moved on the tenant layer — menu, price, promotion — where rivals can and do compete. Nobody competed with the royalty-and-rent layer, because nobody can. Second, and more useful for an owner: Wendy's is what a value war looks like without the landlord model. Same discounts, same squeezed customer, a franchised system too — and the outcome was written-off franchisee receivables, withdrawn guidance and a halved dividend, in the same summer McDonald's prepares its fiftieth consecutive raise. Burger King's revival is the one to watch: a rival with real momentum, a remodel budget, and a $5 meal of its own. The chicken war (Chick-fil-A, Raising Cane's, Popeyes) continues in the background, where MCD's answer is McCrispy, Snack Wraps and Spicy McNuggets (1 Sep).

Integrity: the board that clawed back $105M from its own CEO in 2021 is the same board; the current CEO's public admission of a failed quarter is the behaviour Buffett asks for. Owner mentality: weaker than July on two facts we did not have then — a chairman-CEO who sold $17.5M of stock at the top and a pay ratio above a thousand to one. Neither is a scandal; both are the institutional imperative in miniature. Capital allocation is unchanged and simple: H1 2026 returned ~$2.6B of dividends plus ~$1.3B of buybacks (approx. — 10-Q), with ~$10.3B of authorisation left; shares fell from 723M to 711M over 2023–25 (approx.). The one thing we would ask of Investor Day on 23 September is not a target but a truce: a value programme the tenants agree to fund, priced so that both layers of the system make money. Dial: 7, from 8.

Two broken screens, named. Our feed prints a price-to-book of −176× and a return on equity of −561%. Both are arithmetic on a negative number and both are meaningless — the mirror image of the lesson we taught at Altria, where equity of −$3.5B produced a P/B of −43×. McDonald's has paid its shareholders more than it ever retained, and the real estate securing its $54.8B of debt sits on the books at cost — land bought in the 1960s at 1960s prices. Negative book value here is a receipt for fifty years of distributions, not a hole. Use ROIC (17.6%) and interest cover (7.8×) instead; they work. One improvement on July's feed: the Altman-Z of 4.6 is computed on populated retained earnings of $72B, so unlike Deere's it can be trusted.

The dividend is safe; the honest finding is in question six. The raise has slowed from double digits to five percent in three years, and management's own guidance is "mid-single digits" going forward (approx.). That is the same decay we found at Rexford (31% → 1%) in miniature, and the opposite of Chubb and Altria, which re-accelerated. It matters for the valuation: a 2.9% yield growing 5% is a different security from a 2.9% yield growing 10%, and the fiftieth raise — which we expect on 22 October — will tell you which one you own. Our guess is ~$1.95–1.97 a quarter, roughly +5%. A raise below 5% would be the first real crack in the King's crown; above 6% would say the board sees the traffic coming back.

In July we savoured a first: a DCF above the price — $311 against $265 — and we said the model deserved attention because a fifty-year royalty stream is its natural habitat. Eight weeks later the same model says $281. The business did not lose $30 a share of intrinsic value in eight weeks; consensus earnings for 2027 moved from ~$14.22 to ~$13.98 (approx.), about 2%. What moved was the model's inputs — a lower starting price, trimmed estimates, a different discount rate — and that is the lesson: a DCF is a mirror with a lag. It still says the stock is 11% below intrinsic value; we would put the range wider, and we would stop quoting it to two decimals. Our own arithmetic: owner earnings of ~$13.4 a share on a 20–22× multiple for a wide-moat, asset-light-in-practice, 5%-growth franchise gives $270–295. At $253 the price sits 6–14% below that band — a margin of safety that is real but thin, and thinner than a Buffett-tier 20–30% discount for a business of this certainty would ask. The street's ~$314 sits in the same neighbourhood as July's $343 once you notice the targets simply followed the price down.