The Slow Triage

Somewhere in the dentsu network over this past weekend, in Milan, perhaps, or São Paulo, or Kuala Lumpur, a market CEO is reading Friday's turnaround update for the third time, trying to work out what colour tag is hanging on their door.

The tags have an inventor. In the 1790s, Dominique-Jean Larrey (Napoleon's chief surgeon) looked at the carnage of a battlefield and did something no army had done before: he sorted it.

Not by rank, which was how armies had always worked. By wound.

The dangerously injured were treated first, entirely without regard to distinction; generals with minor wounds waited while privates with shattered limbs went to the front of the queue. Officers raged. Larrey kept cutting. He called the system triage, from the French trier, meaning 'to sort'.

Two hundred and thirty years later, it is essentially unchanged. It passed through the American Civil War, became doctrine by the First World War, and sits today, in laminated form, inside the major-incident plan of every contemporary hospital. When a mass casualty event occurs, the first thing medical teams still do is Larrey's sort: red tags for immediate treatment, yellow for delayed, green for the walking wounded. And black, the category nobody discusses in front of the patients, for the 'expectant'. The ones who will consume resources that could save the others.

Anyone who watched The Pitt's mass-casualty episodes saw the doctrine in action. The moment the shooting is called in, bins marked MCI are wheeled out, and inside, ready and waiting, are colour-coded slap-on wristbands. Two centuries of medical revolution, and the state of the art for the sort is still a coloured band and a ten-second decision, because the logic was already perfect. The doctrine has survived every one of those revolutions for a simple reason: its two commandments were right from the start.

Sort without sentiment. Sort fast.

On Friday, alongside its half-year results, dentsu updated its turnaround plan with a single sentence of ambition: no markets operating at a loss by the end of 2027.

Read that again. It is not a growth target. It is not a market-share target. It is a sorting instruction. Every one of dentsu's international markets now has a tag on its door.

One year ago this month, I published Time to Take the Ball and Go Home. The argument: dentsu's optimal path wasn't fixing its broken international expansion but embracing what I called the "fortress strategy" to defend and grow their impregnable position in Japan's $51.3Bn advertising ecosystem. I estimated that "even at fire-sale valuations of 6-8x EBITDA, Dentsu's international portfolio could generate $3-5 billion while eliminating operational complexity."

That same month (August 2025) dentsu appointed bankers to explore exactly that sale. In January, I covered the result in The Messy Middle On Sale: Apollo walked, Bain developed "significant reservations," and the process collapsed. The buyers, I stated at the time, "weren't walking away from a bad price. They were walking away from a business model that doesn't escape the deathly middle no matter who owns it."

This direction was vindicated so completely that the board's answer to the crisis was to promote the fortress into the global CEO seat. Takeshi Sano, the man who ran dentsu Japan, now runs dentsu, full stop. The org chart is the confession.

Before we examine what dentsu announced on Friday, be fair to the man holding the clipboard, because on Larrey's first commandment, Sano has been flawless.

He inherited a disaster: a record net loss of ¥327.6 billion (around $2.1 billion) for 2025, driven by a ¥310.1 billion ($1.9 billion) impairment on the overseas operations, a suspended dividend, a collapsed sale process, and a departed CEO.

Pause on that loss, because if you sit outside Japan it is easy to misjudge the scale of the animal that survived it. Dentsu's single-year loss of $2.1 billion is bigger than the entire stock-market value of Stagwell. Not Stagwell's loss, not Stagwell's profit; the whole company, currently priced at roughly $1.8 billion. A bleeding beast. But a beast.

Judge his response on its own terms and it is exactly what a competent field surgeon does. Guidance held, twice. Roughly 3,000 of the planned 3,400 role reductions complete, 88% executed. The drama has stopped.

And the sorting has been genuinely rank-blind. Larrey's radicalism was that the insignia meant nothing, only the wound and the odds.

Consider what dentsu triaged out first. Not a struggling creative agency in a peripheral market. The Ginza headquarters: opened in 1933, the symbol of the company's post-war rise, the most decorated asset on the books. It was assessed, tagged, and sold, and the roughly ¥30 billion gain (around $190 million) is funding the treatment of everything else in the tent. Whatever else you say about this plan, there is no sentiment in it. The generals are not being treated first.

There is even a defensible logic to keeping the wounded network alive at break-even rather than closing it outright. Toyota for example still needs serving in a hundred markets, and a network that provides that service without consuming capital is a client asset for fortress Japan, not a strategic embarrassment. A calmer balance sheet in 2027 sells Merkle better than a distressed one in 2026.

If you were hired to stop the bleeding, this is what stopping the bleeding looks like. The problem is Larrey's other commandment: sort fast.

Note what Friday's headlines told you: dentsu returned to profit. Net revenue up 3.7%. A ¥46 billion interim profit (roughly $290 million). Turnaround on track. Next story.

That profit swing is a building, a currency, and an absence.

Roughly ¥29.6 billion of it is the Ginza gain. A further slice is the weaker yen. The rest is simply the non-recurrence of last year's impairments. Strip out the one-offs and the underlying business is a flatline: organic growth of 0.3% for the half and negative 0.2% in the second quarter, with the Q2 operating margin down 40 basis points. The half-year margin improvement the release trumpets was earned entirely in Q1. All three international regions shrank. Full-year guidance, reiterated on Friday, projects revenue up 3.9% but underlying operating profit down 3.6%. The dividend remains suspended.

We are at the end of another earnings cycle, and the pattern held. The major holding companies have now reported, and the trade press, tired of copy-pasting from earnings release headlines, have moved quickly back to covering new ads, the logo refreshes and the creative directors changing jobs. More than 20,000 people will lose their roles across the five majors this year. The trajectory buried in these numbers is what decides that, and sadly almost nobody reads past the first line.

Larrey's system was not gentle. At Borodino he reportedly performed two hundred amputations in twenty-four hours, one roughly every seven minutes, under fire, without anaesthetic. The brutality had a purpose, and the purpose was the entire moral logic of triage: speed is what keeps the yellow tags from turning black. Larrey knew that a soldier operated on within the first hour usually lived, and the same soldier left until evening usually didn't. Sorting without speed isn't triage. It's just a queue outside a tent where the treatable deteriorate into the expectant.

Dentsu's triage runs to the end of 2027.

Closing a market costs money for severance, leases and client transitions and after a ¥327.6 billion loss with the dividend suspended and distributable reserves still deep in negative territory, those costs can only be paid as fast as Japan's transfusions cover them. The fortress is not just the profit engine; it is the blood supply. A record first-half from dentsu Japan with thirteen consecutive quarters of growth and margins above 30% is what funds every bandage in the tent. The pace of the triage is set by how much blood the donor can spare per quarter.

And while the sort proceeds at transfusion speed, the salvageable continue to decay, because an unsold agency network is a melting asset in a way a field hospital's patients, grimly, are not even a match for. The value walks out of the door every evening and is under no obligation to return in the morning. Clients read earnings calls too, and a network in visible run-off invites exactly the client caution that turns a managed decline into an accelerating one. The parts with genuinely retained value, Merkle above all, the media operations and pockets of creative, are salvageable today. Every quarter of managed quiet, more yellow tags darken.

Dentsu knows this, which is why the most revealing sentence in its communications this year is not in Friday's results at all. It is the position, held since February, that while the group-wide sale is dead, it remains open to "divesting certain businesses at the local level". Translated from the corporate: the whole was unsellable, but the parts might still fetch something if we sell them before the tags turn black.

Which returns us to that market CEO, still reading.

Larrey's system has survived for two hundred and thirty years because its two commandments only work together. Sorting without sentiment but without speed is not a field hospital; it is a waiting room with excellent record-keeping. By the end of 2027 we will know which one dentsu built. If the sort accelerates, such as if Merkle and the media assets are invested in or sold while they still command value, and the expectant markets are closed with Larrey's speed rather than a CFO's, then Sano will have executed the fortress strategy by the only route the market left open, and this period will read as the disciplined middle chapter of a genuine turnaround.

But if the quiet persists, like if the flatline is presented as stability for another four quarters while the salvageable decay at transfusion speed, then what dentsu announced on Friday was not triage at all. And the difference between a field hospital and a hospice is not the calm. Both are calm. The difference is whether anyone is still being treated.

The results being boring is not the same as the story being over. Sometimes the absence of news is the news.

So, a question for those of you who know these businesses from the inside, either agency-side, client-side, or from within the network itself: if you were holding dentsu's tags, which parts of the international business get green, and which get black?

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 17 August 2026 in the Brandflow newsletter on LinkedIn.