The State of Paid Media in the AI Era

Paid media used to be a craft. You learned the platforms, you mastered the bid logic, you understood the auction dynamics, and the gap between a good media buyer and a bad one was measured in millions of pounds of wasted spend. That craft is mostly gone. The platforms have absorbed it. In its place is a different discipline — one most UK marketing teams are not yet structured for, and most agencies are not yet honest about.

The shift is not subtle. In 2022, a strong UK paid media practitioner could meaningfully outperform default settings on Google and Meta through manual bidding, granular audience construction, and surgical placement controls. By 2024, those advantages had narrowed sharply. By 2026, on most account types, manual override of platform automation is actively destroying value — and the data inside the platforms now proves it account by account.

The state of play in 2026

Three platforms — Google, Meta, and Amazon — now command roughly 75–80% of UK digital ad spend (with TikTok closing meaningfully on the consideration-stage budget that used to flow to Meta). All three have converged on the same operating model: a single goal-based campaign type, fed by first-party conversion signal and a deep library of creative, that handles bidding, audience selection, placement, and creative rotation autonomously.

In Google that is Performance Max plus Demand Gen. In Meta it is Advantage+ Shopping and Advantage+ Audiences. In Amazon it is the sponsored-product automation layer plus DSP's AI-driven optimisation. The platforms have not just added AI features — they have made AI the default and quietly deprecated the manual surface area that used to define expert media buying.

What's actually changing

The structural change underneath all this is that the role of the human media buyer is collapsing into three new functions, none of which look much like the 2020 job description:

• Goal architect. Defining what success actually means — which conversion events matter, what each one is worth, what the constraint is (volume, efficiency, growth, profit margin), and how the AI should weigh trade-offs. Get this wrong and no amount of platform automation will save the account.
• Creative supply lead. Producing enough variety of creative — formats, hooks, lengths, angles — to give the platform real choice. The AI cannot pick the right creative for the right audience if it only has three options to pick from. The brands winning are shipping creative variants the way newsrooms ship articles: continuously.
• Signal & measurement engineer. Making sure the conversion data flowing into the platform is clean, complete, and weighted to reflect real business value. Enhanced conversions, server- side tracking, offline upload, value-based bidding signals — all the plumbing that used to be a 'nice to have' is now the actual leverage point.

Live UK signals

A few things we are seeing repeatedly across UK accounts in Q1 2026, that we'd encourage every CMO to pressure-test in their own data:

• Accounts that migrated fully to PMax and Advantage+ between Q3 2024 and Q2 2025 are now showing 18–32% lower blended CPA than equivalent accounts that retained legacy campaign structures — but only when creative supply and conversion signal were upgraded in parallel.
• Asset-group creative diversity has emerged as the single strongest controllable variable for PMax performance. Accounts with 20+ distinct creative concepts per asset group are outperforming accounts with fewer than 10 by margins large enough to make the relationship unambiguous.
• Manual exclusions of brand search and audience layering — once considered table stakes — now frequently underperform the algorithm's own choices. We are recommending most clients turn most of these off and let the platform learn.
• TikTok Smart Performance Campaigns are quietly becoming a meaningful DR channel for UK ecommerce in considered-purchase categories (£40–£200 AOV), particularly for brands with strong UGC supply.

What this means for UK CMOs

If you lead a UK marketing function, the practical shifts to make in the next two quarters are:

1. Audit your conversion architecture before anything else. If your platforms are not receiving clean, weighted, server-side conversion data, every other optimisation effort is downstream of that fix. This is the single highest-leverage thing most UK accounts could do today and aren't.
2. Restructure creative production. Move from quarterly campaign assets to a continuous creative supply line — ideally with AI-assisted variant generation feeding human-edited final assets. Target a 5–10× increase in monthly variant volume on the same or smaller production budget.
3. Stop micromanaging the algorithm. Most manual overrides — bid adjustments, audience exclusions, placement filters — now actively underperform. Set the goal correctly, supply the signal, and audit weekly rather than tweaking daily.
4. Renegotiate your agency relationship. If your agency is still primarily charging for media-buying labour, you are paying for work the platforms now do for free. The valuable work is creative strategy, conversion engineering, and AI orchestration — make sure that's what you're actually buying.

The honest part

The uncomfortable truth for everyone in paid media — including this agency — is that the most profitable era of expert media-buying labour is behind us. The platforms have absorbed the craft we used to sell. The teams that thrive from here will be the ones who admit this, restructure around the new high-leverage activities, and stop pretending that account-management labour is still where the value sits. The brands that win will be the ones that demand this honesty from their partners.

For the broader strategic context, see The AI Marketing Revolution and The End of the Marketing Funnel. For our take on how the agency model itself has to change to survive this shift, see The Death of the Marketing Agency (As We Know It).