By Mike Sharp, Operations Director at Launch
That time of year is here. The time when I confidently predict the future of paid media, only to probably look back next year and wonder what on earth I was drivelling on about.
I have tried to do this every year I have worked in agency land, and I’d say my track record is mixed at best.
Looking at 2026 feels different though, mainly because the robots are here and doing half of our jobs now. So how hard can it be to predict…?
Hitting the performance plateau hard
We are already seeing early indications of some brands hitting a growth plateau, but going into 2026, this will affect every advertiser who isn’t evolving their approach. Accounts relying solely on bottom-of-funnel search tactics will reach their limits. You will not be able to solve a volume problem through further optimisation.
The most successful brands in 2026 will be those willing to accept lower ROAS on certain campaigns in favour of new user acquisition. Investment in social, YouTube, and display advertising will become necessary to encourage users to search for the brand later. This may not be new, but the increased urgency will be. After many years of hearing this from Google, 2026 may be the time when increased allocation to YouTube becomes essential.
Google goes AI mode mad
At some point in 2026, Google could be poised to make its most radical pivot yet by establishing AI Mode as the default search experience. This was hinted at by Google’s lead product manager for AI products recently, though quickly clarified and sort of back-tracked-on.
This would effectively retire the traditional list of blue links in favour of a conversational, answer-first interface that addresses the challenge posed by OpenAI’s Atlas browser. This shift would accelerate the zero-click era, in which users get comprehensive answers directly on the search results page, potentially leading to a sharp decline in organic traffic to external websites.
Does this mean search is obsolete? Certainly not; it is simply evolving. Google is unlikely to abandon the goose that has laid so many golden eggs. AI Mode ads are already being tested in the wild and newly developed GEO tactics will address potential shortcomings.
Attribution is dead (or maybe just lying)
Tracking a user’s journey across multiple touchpoints over several weeks has always been difficult. With tighter privacy regulations and browser restrictions, the “perfect path to purchase” view is now already a fantasy. Relying on just platform dashboards in 2026 means navigating a map with half the roads missing.
There is a solution, but it’s not a new tracking pixel; it’s a shift in mindset from attribution to incrementality. We will see a wholesale return to media mix modelling (MMM) and geo-lift experiments. The question will no longer be “Did Google Analytics track this sale?” but rather “Would this sale have happened without this spend?” By systematically darkening channels in specific regions to measure lift, brands will trade the illusion of precision for the reality of growth.
It feels like a modern day return to John Wanamaker’s famous quote: “Half my advertising is wasted; the trouble is I don’t know which half.”
Brand and performance finally merge
For 15 years, I’ve seen brand teams and performance teams fighting, or at least siloed and not working closely together. The narrative has it that the brand team makes the pretty pictures that build emotion but don’t sell, while the performance team churns out ugly, high-frequency ads that convert but eventually annoy.
In 2026, the silos will have to stop. As third-party data vanishes, the algorithms powering Meta and Google require a full-funnel view to function. They need the broad intent signals generated by brand campaigns to effectively model and find high-value users for performance conversion. You can no longer harvest demand you haven’t created. The winners of 2026 will be the teams that treat brand as a strategic input for the algorithm, not just a line item for “awareness.”
The evolution of experimentation (and why you still need humans)
AI is increasingly being integrated into experimentation across all areas, but the human approach will remain essential in 2026.
On one side, we have synthetic users. Brands are now using AI to simulate customer journeys and narrow down 50 bad ideas to three good ones in seconds. It’s faster than a focus group and cheaper than a survey.
But – and this is the part people forget – AI is also a hallucination engine, not an oracle. It is not by default a reflection of reality, and you cannot blindly trust it. It can’t cope well with niche audiences and behaviours. It assumes everyone acts rationally. If you rely solely on synthetic data, you are betting your budget on a guess.
That is why the smart money is flowing into platforms that can still offer the human approach. Look at OpenAI’s $1.1 billion acquisition of Statsig. They know that AI can generate the idea, but you still need a rigorous framework to test that idea on real humans. Even Google announced this week it was resurrecting its long-mourned optimisation tool (hidden inside Google Ads). They know we need to validate their AI’s guesses with actual traffic to stop the models from eating themselves.
So, you know. Looks like 2026 is set to be just as predictable as every other year.