When media buying scales from 10 campaigns to 1,000, the number of decisions doesn’t just grow, it outgrows the time people have to make them. In models that rely on human review, every new campaign adds coordination, cost and operational lag. But in an autonomous model, campaign count stops being the limit, and what starts to matter is the quality of the boundaries the system decides within.
In this article
- What grows when the number of campaigns grows
- How each media buying model scales
- What scaling costs in each model
- When coordination cost outweighs human review
- What governs scale in an autonomous model
- How Mainkore handles scale
- FAQ
What grows when the number of campaigns grows
Scaling media buying multiplies four things at once: decisions per day, signals per decision, combinations between channels and review cycles.
Each new campaign adds its own bids, budgets and creatives to watch. Each new channel adds signals that should inform decisions elsewhere. And every combination of campaign and channel is a place where budget could be better allocated, which means the number of possible decisions grows faster than the number of campaigns. Review cycles grow too, because someone has to look at all of it.
How each media buying model scales
Each media buying model scales through something different: people, a platform, or a decision layer across channels.
| Model | How it scales | What grows with each new campaign | Where the limit sits |
| In-house team | With people | Review hours | Team size and hours in a day |
| Traditional agency | With people, billed externally | Hours and fees | Agency team capacity and review cycle |
| DSP or programmatic | Inside the platform | Setup, rules and exception checks | The platform’s own environment |
| Smart bidding | Inside the ecosystem | Configuration per campaign | The ecosystem’s boundary |
| Autonomous AI media buying | Across channels | Objectives and creative, not review of each decision | The quality of the boundaries it works within |
None of these limits is a failure of the people involved. They’re the point where each model’s architecture stops stretching.
What scaling costs in each model
In models that depend on review, the cost of scaling grows with the number of decisions, not with the value of the work.
An in-house team’s cost grows roughly with headcount: reviewing 1,000 campaigns properly needs meaningfully more people than reviewing 10. An agency’s cost follows the same curve, billed as more hours and a bigger retainer. Programmatic and smart bidding avoid most of that headcount, since bidding itself doesn’t need a person per campaign, but the configuration, rules and exception checks around them still tend to grow campaign by campaign. What connects all three is coordination cost: the effort of keeping many decisions aligned.
When coordination cost outweighs the value of human review
Human review stops paying for itself when the time it takes to coordinate a decision is longer than the window in which that decision is still useful.
At that point the team doesn’t decide worse for lack of talent. It decides later, for lack of hours. That delay is operational lag, and it’s easier to measure than it sounds. Three questions help: how long passes between a signal appearing and a budget changing, how many campaigns each person can review properly in a week, and which decisions routinely wait for the next meeting. If the answers are growing faster than the results, scale is already costing more than it shows. Speed of reallocation is the other side of this, covered in how AI reallocates advertising budgets.
What governs scale in an autonomous model
In an autonomous model, what limits scale isn’t the number of campaigns, it’s the quality of the boundaries the system operates within.
Spend ceilings, brand rules and channel eligibility are defined once and apply the same way at campaign 10 and at campaign 1,000. That’s what lets the review load stop growing with volume. It also means a vague boundary scales just as efficiently as a good one. Our view is that at scale, boundary design becomes the main job of the people running media, and the place where their judgment has the most leverage. How those boundaries are built is explained in how to govern AI in paid media.
How Mainkore handles scale
Mainkore is autonomous advertising decision intelligence that has operated across more than 12,000 campaigns, evaluating 200+ variables per decision, in around 20ms, 24/7.
Adding a campaign still means defining its objective, budget and creative. What doesn’t grow with each campaign is the review of every allocation decision, because those decisions are made continuously inside boundaries set up front. That’s the specific claim worth testing when you compare models: not which one performs best at 10 campaigns, but what your cost and your lag look like at 1,000.
FAQ
The ability of a media buying model to handle more campaigns, channels and decisions without decision quality, speed or cost per decision getting worse. What changes at scale is structural, not only a question of volume.
The time and effort needed to keep many buying decisions aligned across people, campaigns and channels. In review-based models, it grows with the number of decisions.
No. It moves the team’s work from reviewing individual decisions to designing the boundaries those decisions follow, and to the strategy a system is never given.
Scalability is one of the eight criteria in our media buying models guide. Want to see what your cost curve would look like at 1,000 campaigns?
