AI has stopped being a novelty in media planning. Most marketing organizations already use it, and the conversation has shifted from discovery to delivery. Campaign briefs now treat agentic planning as a baseline rather than an experiment, and clients are no longer asking whether agencies test AI. They want to know what it has achieved and where the roadmap goes next.
That is a healthier conversation, but it creates a new pressure: adoption no longer earns credit on its own. The bar has moved from experimentation to proof, and the infrastructure for providing that proof has not kept pace.
The measurement problem behind the proof
Nowhere is this more visible than in incrementality. It is one of the most requested measures in advertising, yet the standards, audit frameworks and accreditation behind it are still maturing. Brands, agencies and platforms each have their own definition of “what really works” — with different baselines and assumptions.
- The bar has moved: brands now judge AI by delivered outcomes and multi-year roadmaps, not pilots.
- Incrementality is under-standardised: every partner measures differently, so comparisons are messy.
- Ease of buying is being confused with value: inventory outside easy pipes gets ignored even when it performs.
When ease of access looks like value
AI is also revealing how much valuable media sits outside the paths the industry has made easiest to use. Affinity has spent two decades working across privacy-first browsers, device manufacturers and enterprise-grade native ecosystems such as OEM, CTV and smart devices. These are meaningful consumer touchpoints, but much of that inventory has historically been difficult to reach through conventional programmatic workflows.
The result is a structural gap: walled gardens have become the default partly because they are simple to buy. When ease of buying becomes the filter, advertisers can leave incremental performance on the table. The same scrutiny should apply to pricing. CPM tells you the cost of a thousand impressions, not whether those impressions were worth anything. The industry keeps using it because it is a shared language, not because it is the best measure of value.
Productivity is not the same as value
Productivity gains are internal: hours saved, campaigns turned around faster, reporting that once took a day now taking ten minutes. These improvements are real and increasingly visible. But they are not value. Value is whether the work changed the business: demand that would have been missed, incremental sales, better customer economics, and access to audiences the old planning process never surfaced.
What the shift means for marketing teams
For marketing and agency leaders, the useful response is to change the questions asked in reviews and briefs. Instead of “are you using AI?”, the questions become “what did it deliver?” and “what is the roadmap for the next few years?” The same applies to measurement: asking for incrementality reports is not enough if the definitions and baselines are not comparable. Media plans should also be checked for inventory that was easy to buy rather than genuinely high-performing.
As AI gets embedded across planning, creative and operations, the technology will become less of a differentiator and more like infrastructure. What will separate companies is the ability to demonstrate, with evidence, that outcomes changed — not simply that workflows accelerated. AI makes it easier to move quickly. It does not remove the need to prove the decisions were right. The real opportunity is to create more value from the same investment.
Source: Digiday
