Social Rocket
More and more advertising budget is being managed through self-service platforms, where AI influences pricing, delivery, and optimization. For marketers, this creates a new problem: these systems help you buy reach, but they also pursue the business goals of the platform that operates them.
Marketing Dive cites Gartner forecasts that by 2028, more than 70 percent of global advertising spend and around 80 percent of US advertising spend will run through self-service systems in which AI materially influences media buying. The central advice: use automation, but do not hand budget decisions over to the platform unchecked.
Why platform AI is not neutral
An advertising system does not optimize in a vacuum. Google, Meta, Amazon, and other platforms simultaneously sell the inventory their algorithms recommend. As a result, the interests of buyer and platform can diverge: advertisers want the most efficient results possible, while the platform wants to increase revenue and utilization.
This does not mean automated recommendations are inherently bad. It does mean that suggestions such as budget increases, additional placements, or broader target audiences should not automatically be treated as objective advice.
The practical counter-approach is simple: define campaign goals in advance, check recommendations against your own KPIs, and test changes in a controlled way. How far platforms now automate operational work is also covered in our article on TikTok Ads with Agentic Hub and Symphony.
Fewer variables, better control
Another problem is measurement across multiple platforms. Each provider works with its own attribution models, data windows, and optimization logic. As a result, two dashboards can evaluate the same campaign differently.
Instead of automating as many channels as possible at once, it can make more sense to limit the number of variables and first invest where costs and business results can be clearly connected. Especially with larger budgets, finance, analytics, or independent measurement partners should be involved in the evaluation.
The current dispute over Google's advertising technology shows how concentrated parts of this market already are. More background is provided in our article on the US ruling against breaking up Google's adtech business. The new global media model from PepsiCo and Publicis also shows how strongly data, technology, and buying are now being merged.
What marketers should concretely check
Before an automated budget decision, at least four questions should be answered: What business goal should the campaign achieve? Which metric actually reflects that goal? Can the platform plausibly show the connection between spend and results? And can the result be verified outside the platform's dashboard?
Be particularly cautious with new campaigns that have little data. There, automated systems can quickly draw far-reaching recommendations from a few signals. The same applies to fully automated cross-channel setups, where media planning, budget allocation, and optimization are handed to one system at the same time.
AI does not replace media judgement
Productive use therefore lies between two extremes. Rejecting all automation means forgoing real efficiency gains. Delegating planning, buying, and performance control entirely to the same platform, on the other hand, means losing independent oversight.
For agencies and in-house teams, this shifts the work from operating to evaluating: fewer manual clicks, but more responsibility for goal definition, experiment design, and measurement. Modern retail media measurement works similarly, for example with Ace Hardware's weather-triggered programmatic approach.