Mid-Year Predictions Check-In

Time for a pit stop and tyre change. Six months ago I published ten predictions for 2026 and promised to score them in public in December. They are worth a midpoint check-in.

Not because I want an early victory lap. The opposite, really. Re-reading my own forecast halfway through the year, I felt something I hadn't expected: not the quiet satisfaction of being right, not the sting of being wrong, but a kind of vertigo at how far the ground has already moved.

In December I called 2026 a watershed: the ridge where the industry stops transforming and starts being replaced. I said that if seven or eight of the ten landed, the watershed was real. Last year I went six and a half out of eight, and the thing I learned was that the half-marks (the calls that were right in spirit and wrong in detail) taught me more than the clean hits.

This year the half-marks are the whole story.

Here's what I want to say before I grade a single line. None of these were obvious calls. It's easy to predict that "AI will matter" and collect your applause in December. I tried to predict the mechanisms of who captures which layer, which price moves and which structure breaks. And the thing I keep running into, six months in, is that the industry didn't prove me wrong so much as it went somewhere I hadn't drawn on the map. That isn't a failure of forecasting. It's the most alive thing about this business.

So let me grade honestly. The destinations, mostly, I called. The routes the industry took to reach them are where I want to spend your time, because the routes are where the optimism lives.

Prediction 1: The Great Compression becomes structural. Tracking, and understated.

I said the Omnicom–IPG merger would eliminate the announced 4,000 roles and then quietly many thousands more, and that these jobs would not return in the next cycle. By the end of last year, IPG had already shed around 3,200 roles and Omnicom about 3,000, past the headline number before the integration had properly begun. DDB, FCB and MullenLowe, names older than most of the people reading this, no longer exist as networks. The combined group is steering towards roughly 105,000 people, down from 128,000. When I wrote The Real Count, I argued the announcement number was a floor, not a ceiling. It was.

Prediction 3: Zero-click search kills 40% of traffic. Past the post, and then some.

This is the one I'd most like to have been wrong about. I wasn't. Fewer than a third of Google searches now end in a click to anyone's website. On the queries that trigger an AI answer, the zero-click rate runs past 80%. Google's own AI Mode has crossed a billion users. Publishers are in court calling it an extinction event. In The Zero-Click Crisis I said the click would die and the citation would live. The funeral has been quicker than I expected.

Prediction 4: Google is king again. Right engine, overstated volume.

I said Google's distribution moat would let it reassert itself, that "Google is dying" would collapse as a narrative, and that OpenAI had taken its shot and missed. The engine of that call was right: Gemini 3 has led on the benchmarks that matter, AI Overviews reach two billion people, and nobody serious is writing Google's obituary now. But I oversold the ending. OpenAI didn't miss (yet), ChatGPT still leads on consumer usage and crossed 1.1 billion monthly users. What's actually forming is an oligopoly, not a restored throne. I was right about the comeback. I was wrong to make it a coronation (but I still have 6 months…)

This is the part that taught me the most.

Prediction 8: The AI free lunch ends. Right outcome, wrong mechanism.

I predicted that in the second half of the year the model providers would dramatically raise prices and that the era of cheap AI was about to end with a 300% shock. The outcome is arriving but the bill is exploding in a different way, because the free lunch is ending through metering, not pricing. Anthropic moved to per-token billing. OpenAI's own head of ChatGPT admitted an unlimited plan makes about as much sense as unlimited electricity. Uber burned its entire annual AI budget in four months. I forecast the destination (costs spiralling) but got the road completely wrong. Cheaper tokens, bigger bills. I didn't see that coming.

Prediction 6: OpenAI launches a marketing agent. The opposite happened and it half-came true anyway.

I said OpenAI would launch a $500-a-month marketing agent that quietly replaced junior teams: the Napster moment for agency services. What OpenAI actually did was start selling advertising inside ChatGPT, with Target, Audible and Adobe among the first brands. Almost the inverse of what I predicted: not a tool that does marketing, a surface that sells it. And yet the underlying call, autonomous agents collapsing the junior-team value proposition, is happening everywhere, just under other logos: Adobe, Salesforce, HubSpot, a dozen startups. The disruption is real. I simply put it in the wrong hands.

Prediction 9: Consumers control the vault. Not yet, but soon. (I've heard that one before.)

Last year I made a blockchain prediction and it didn't land. This year, exposed and stubborn, I made it again, that consumers would start owning their own data through blockchain vaults, with a major brand leading the way. The need showed up exactly as I argued: zero-party data, the information people choose to hand over, is now described as the new loyalty currency. But it's arriving through ordinary loyalty programmes, not blockchain. Twice now I've predicted this shift, and twice the future has turned up wearing a different outfit. At some point I have to ask whether I'm right about the destination and permanently early on the vehicle or just wrong in a way I happen to find flattering.

Prediction 7: The CMO role splits into Growth and Brand. Half right, and personally awkward.

I predicted the chief marketing officer would fracture into two, a Chief Growth Officer and a Chief Brand Officer, because no one person could do both jobs well any more. The unbundling is undeniable: fewer than half of the largest companies now even use the CMO title, and roughly a third operate without one at all. But the role isn't splitting cleanly down a growth-and-brand seam. It's being pulled towards commercial, revenue, growth and the brand half is the quieter one.

And then, in May, I took a job as the full-remit CMO of Nomad Foods. Brand, growth, innovation, R&D, the lot. I spent December predicting the death of the all-in-one marketing leader. I spent the spring becoming one. Make of that what you will. I'd argue it tells you something true, that the role isn't disappearing, it's concentrating, and the people who can hold the whole thing together are becoming more valuable, not less. But I'll concede it's a funny look.

Prediction 10: The market revalues Publicis. Strategy right, market unmoved.

This was my most exposed financial call: that Wall Street would finally stop valuing Publicis like a holding company and start valuing it like the data-and-platform business it has become by tripling its market cap towards forty billion. It hasn't happened yet. Publicis sits around twenty-six billion, on roughly the same revenue multiple as ever, on a price-to-earnings ratio in the low teens. The strategy I've been mapping for a year (identity, commerce, creators, sport, and now data infrastructure with LiveRamp) is intact and compounding. The market just won't pay for it yet. This is the Publicis Paradox I wrote about, only sharper: a company doing everything right operationally, still priced as the thing it's trying to stop being. I was right about the company. I was wrong about the crowd.

Prediction 2: A sovereign wealth fund buys a holding company. Wrong target, right trend.

I predicted a Gulf fund would acquire WPP or another mid-tier holdco, folding marketing services into a portfolio of sport and entertainment. No such deal has landed on WPP, which is busy restructuring under Cindy Rose. But look at where the sovereign money actually went this year: the Public Investment Fund into a fifty-five-billion-dollar take-private of EA, Gulf funds underwriting an eighty-billion-dollar media takeover. The appetite I described is real and enormous. I aimed it at the wrong company.

Prediction 5: Adobe becomes the orchestration layer. Real, but unprovable.

I said Adobe's Agent Orchestrator would, by year-end, run more marketing execution than the top three holding companies combined. The direction is unmistakable as Adobe went all-in on orchestration this spring, and tellingly, Omnicom, Publicis and WPP are now building on top of it. But "more than the big three combined" is a claim no one can actually measure, so I'll mark it honestly: directionally right, literally unscoreable.

Tally it up and the pattern is clear. My structural calls like the compression, the search collapse and the platform power landed hard. My specific, contrarian, event-shaped bets like the acquisition, the blockchain vault and the revaluation are the ones that missed. And several of the "misses" aren't really wrong about what is happening, only about how it arrived.

Here's what I take from that, and it isn't gloomy. The reason I keep getting the mechanism wrong is that this industry is more inventive than any forecast. I predicted a price hike; it found metering. I predicted a marketing agent; it built an ad network. I predicted a blockchain vault; it reached for zero-party data. Every time I drew a straight line, the industry found a door I hadn't put on the map.

We spend a lot of time… I spend a lot of time… writing about compression as a threat. But the same restlessness that compresses is the restlessness that creates. An industry this hard to predict is an industry that is very much alive. And for the people in it, that's the good news hiding inside the disruption: a business that reinvents its own plumbing every six months is a business that keeps making room for the people who can reinvent alongside it. Creative by name. Creative, relentlessly, by nature.

Six months ago I drew a map of a country I thought I understood. Halfway through the year, the country has already moved, not away from where I pointed, but by routes I never imagined. That used to worry me. Now I think it's the most hopeful thing I know about this work: the ground keeps moving because the people standing on it refuse to.

See you in December, for the full reckoning.

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 22 June 2026 in the Brandflow newsletter on LinkedIn.