The Paris Discount

Two shops on the same street. Same good year, same surplus sitting in the till.

One owner takes the money and pays it out. A fresh coat of paint on the sign, a tidy sum handed back to the people who own the place. You can see exactly what was done with it. The street nods approvingly. Sensible. Disciplined. A shop that knows how to reward its owners.

The other owner does something stranger. She sinks the same money into a cold store out the back, a delivery van, a loyalty scheme that won't break even for two years. None of it shows in the window. Walk past and the shop looks exactly as it did last month, only with a lighter till. The street mutters. Wasting money. Overreaching. Should have just banked it.

Then the supermarket opens across the road.

A year later, walk down that street again. One of the two shops is still trading. It's the one with the cold store, the van and the loyalty scheme. The one everybody said was wasting its money. The other is a vacant unit with a beautiful but slowly fading sign.

This is not a parable about shops. Within three months of each other this year, two of the largest advertising companies on earth each wrote a cheque for roughly the same amount, drawn on cash they already had.

Omnicom spent $2.5 billion of it. Publicis spent $2.2 billion of it.

Omnicom bought back its own shares.

Publicis bought a company called LiveRamp.

One reversed its own dilution. The other bought the data infrastructure of the agentic era.

The market cheered the first and discounted the second. That discount has a postcode, and it has a name. It's the Paris discount and it may be the most interesting mispricing in the industry right now.

On 18 February, alongside its fourth-quarter results, Omnicom's board authorised a $5 billion share-repurchase programme and immediately executed $2.5 billion of it through an accelerated repurchase, funded with cash on hand. The stock jumped around 13% that day, even though the quarter itself carried a headline accounting loss. The mechanism is worth saying out loud, because it is the whole point: buy back enough of your own shares and earnings per share rise even if the business earns no more than before. Fewer slices, same pie, a bigger number per slice. The financial press has a name for it. The buyback tailwind.

A few weeks later, Publicis spent its $2.2 billion the other way. LiveRamp, the data-collaboration layer that lets brands, retailers and publishers match and activate audience data across 25,000-plus publisher domains without exposing the raw records to each other. It is the fifth move in a sequence I've mapped here before: identity (Epsilon), commerce, creators, sport, and now the data pipes. More than €10 billion poured into acquired capability over the past decade while the company has conspicuously declined to do the one thing its rivals reach for when they run out of ideas.

A fortnight ago, Goldman Sachs initiated coverage of all three big listed holding companies. WPP: sell (with a 240p target). Publicis and Omnicom: buy. And it set out, in the dry language of a broker note, the exact puzzle this newsletter is about: that the company spending on capability is also the one the market can't quite bring itself to pay for.

Here's where the lazy version of this story goes wrong, so let me take the other side first, properly, because it's stronger than the cheerleaders admit.

A buyback is not, by itself, a failure of imagination. It is sometimes the most honest thing a management team can do. If you cannot deploy capital inside your own business at a better return than your shareholders could get elsewhere, handing it back is discipline, not surrender. The history of corporate empire-building is mostly a history of value destroyed by executives who couldn't bear to return cash and bought things they shouldn't have. Serial acquirers overpay. Integrations fail. Synergies are promised and quietly forgotten.

Publicis knows this better than most. I can personally assure you that Sapient was a hard, multi-year integration that tested the patience of everyone involved (not to mention a write-down in year 3 of the integration). Epsilon, now lionised, was met with open scepticism in 2019 as too expensive, too far from advertising, a data business a creative company had no obvious right to own. Acquired capability is a promise, not a fact, and the graveyard is full of strategic logic that didn't survive contact with the org chart.

And Omnicom's case is genuinely real. Its scale after the IPG merger is now the largest in the industry. The synergy programme is substantial. It inherited, in Acxiom, an identity asset of real value. A $5 billion buyback puts a floor under the stock and tells the market the integration is on track. None of that is nothing.

But notice the shape of the bull case. Even its most enthusiastic version comes with a warning attached: that the buyback could become, in the financial press's own phrase, a mask for underlying stagnation, and that the thing to watch is organic growth. The bull case for the buyback contains, folded inside it, the bear case for the business. Hold that thought. We'll need it.

Let me be precise about the word "discount," because the obvious version of it is wrong and a sharp reader will catch it.

The obvious version says Publicis trades cheaply next to Omnicom. It doesn't. The entire sector has been written off. Advertising trades at roughly 8.8 times forward earnings against the S&P 500's 22.5. The market gave up on the category somewhere around 2023, when AI first put a question mark over the whole model, and it has not come back. On the headline multiple, Omnicom is, if anything, the cheaper of the two.

So the discount is not Publicis-versus-peer.

The discount that matters is Publicis versus itself.

Over the past three years, Publicis has grown its earnings per share by around 12% a year and its share price by around 5% a year. The business compounded. The stock didn't keep up. This is a company posting its twentieth consecutive quarter of growth, guiding to 4–5% organic this year, carrying the best AI-readiness assessment of the three in Goldman's own scoring, throwing off a free-cash-flow yield near 11%. And analyst price targets sit well above where it trades, from Goldman's €110 up towards €131, against a price in the high-€80s. The best operator in a written-off sector, not being paid like one.

There are exactly two of them. Omnicom enjoys both. WPP has lost both. Publicis lacks one and has refused the other.

The first is to be in the right index.

Omnicom is listed in New York and sits in the S&P 500. That is not a neutral fact. State Street alone owns 13.8% of Omnicom, disclosed explicitly on a passive basis, bought not because anyone judged the advertising business, but because it's a line in an index a fund is obliged to track. Passive funds now own roughly 24% of the entire S&P 500, and the big three index managers are the largest single shareholder in 88% of the companies in it. A meaningful slice of Omnicom is owned by capital that has never read an earnings call and never will. It is bought, mechanically, every time someone's pension auto-contributes in a rising American market. Its beta is a sleepy 0.78. There is a permanent bid under the stock that has nothing to do with how the business is doing.

Now look at WPP, where the same machine runs in reverse. In December, WPP was relegated from the FTSE 100, replaced by British Land, ending a membership it had held since 1998. A company worth around £24 billion in 2017 is worth around £3 billion today. And relegation does not just bruise the ego, it flips the index machine into selling. The trackers that had to hold it now have to let it go. WPP's cash generation is now too weak to even attempt the buyback escape hatch its American rival just used; Goldman flagged the free-cash-flow outlook directly. The cushion is gone, the lever is gone, and the private-equity names are circling. The exposed middle.

And Publicis sits in Paris, in the CAC 40. There is real passive support in a French blue-chip, but European passive penetration is a fraction of America's, and there is no automatic transatlantic bid waiting under the stock. Its beta is an even lower 0.55. It is, in market-structure terms, the most exposed of the three to the actual opinion of actual investors. It has no index machine doing its buying for it.

This is where Sadoun has drawn the line in permanent marker. Omnicom reached for it: $2.5 billion to shrink the share count. Publicis has spent a decade reaching the other way, more than €10 billion into capability rather than into its own equity. It declines the mechanical lift. On purpose.

So stack it up. Omnicom: a passive bid and a buyback tailwind. WPP: neither left. Publicis: no automatic bid, and a buyback it won't take. The discount isn't a mystery. It's the arithmetic of those choices.

Sadoun put it more memorably than I can. In a message to staff this spring, reported across the trade press as Publicis posted that twentieth quarter of growth, he said WPP and Omnicom wanted to "squeeze to please Wall Street." Squeeze the people, through layoffs. Squeeze the shares, through buybacks. Squeeze the assets, by putting them up for sale. Publicis, he said, was doing the polar opposite.

Back to the two shops.

The supermarket is open now. It's called Meta, and Google, and Amazon, and it would like very much to sell to your customers without you in the middle. One shopkeeper has a full till today and a freshly painted sign. The other has a cold store, a delivery van, and a reason to still be on the street next year.

The street still admires the fresh paint. It usually does. The street is usually about a year behind.

The Paris discount is what you pay for being early. And as discounts go, it's the one worth having.

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 8 June 2026 in the Brandflow newsletter on LinkedIn.