The Decentralized Swarm: Why App Promoters Use 50 Ad Accounts per Brand

The conventional wisdom of Facebook advertising in the current era is simple: consolidate. Meta's machine learning requires significant data density to optimize delivery. Operators are routinely instructed to minimize their active campaigns, merge their ad sets, and pool all their budget into a single ad account to let the algorithm do the heavy lifting.

When we observe the broader market, operators are largely playing by Meta's rules. We analyzed a general cohort of citations[0] Facebook ads targeting Italy over a 120-day window. In this sample, we found exactly citations[1] brands operating across just citations[2] ad accounts. This represents a highly consolidated structure of approximately 769 ads per account.

However, when you filter the data for a specific objective, the entire consolidation playbook is violently thrown out the window.

The App Promotion Swarm

We built a second cohort containing citations[3] Facebook ads in Italy, restricting the sample to the "App Promotion" objective over a 90-day window.

The structural difference is staggering. While the general market consolidates, the app promotion market shatters into a decentralized swarm. The citations[3] ads in this cohort belong to just citations[4] brands. Yet, those citations[4] brands are pushing media through an incredible citations[5] distinct ad accounts.

To put that into perspective, the general market averages 1.3 ad accounts per brand. The app promotion cohort averages nearly 52 ad accounts per brand, with a mere 6.6 ads running per account.

Why would a mobile app developer need 52 ad accounts to run their user acquisition campaigns? The data points to three distinct operational tactics.

1. The Affiliate Network Architecture

Most major app developers do not rely solely on their internal media buying teams. They outsource a massive portion of their user acquisition to affiliate networks, paying on a strict Cost Per Install basis.

Rather than giving hundreds of affiliates access to a central corporate ad account, each affiliate uses their own infrastructure. If an app has 50 different independent media buyers running traffic on their behalf, they will naturally appear across 50 different ad accounts. The fragmented nature of the cohort is a direct reflection of a decentralized workforce.

2. Risk Mitigation Through Redundancy

Mobile app creatives routinely test the boundaries of platform policies. Whether it is aggressive gameplay footage that does not perfectly match the actual app experience, or bold claims about in-app rewards, app promotion ads are frequently flagged and disabled.

If a brand uses a single, consolidated ad account, a policy violation can freeze their entire user acquisition pipeline. By distributing their citations[3] ads across citations[5] ad accounts, operators neutralize the threat of platform bans. If an ad account is restricted, the swarm barely notices. The budget simply shifts to the other active accounts.

3. The Micro-Budget Burnout

The most revealing difference between the two cohorts lies in their reach. The general cohort achieves a median total reach of citations[6] users per ad. The app promotion cohort reaches just citations[7] users before the ad is killed.

Why are the ads dying so fast? The median spend for the 25th, 50th, and 75th percentiles in both cohorts is locked at exactly citations[10] EUR. This is a clear signature of automated media buying rules. However, while general ads might pause and restart, the app promotion ads are churned and burned.

With an average of just six ads per account, operators are likely spinning up a fresh ad account, launching a handful of creatives with a citations[10] EUR budget threshold, and killing anything that fails to deliver immediate installs. The microscopic reach of citations[7] users indicates that the vast majority of these ads fail the initial test and are never turned back on.

The Mirage of Cheap Inventory

At first glance, the decentralized swarm appears to be incredibly cost-efficient. The app promotion cohort registers a median CPM of citations[9] EUR, compared to citations[8] EUR in the general market.

However, operators must be careful not to mistake a statistical anomaly for a bidding advantage. Out of the citations[3] app promotion ads analyzed, only a citations[11] registered enough sustained delivery to record a valid CPM. Meta's reporting requires a baseline volume of impressions before it will calculate a CPM metric. Because the swarm kills its ads after reaching just a few hundred users, they never cross the threshold.

The citations[9] EUR CPM is a mirage generated by a single surviving ad, not a reflection of the overall market.

The data makes one thing absolutely clear. While Meta continues to push for consolidation and machine-learning optimization, a subset of performance operators is actively resisting. They are choosing fragmentation, redundancy, and brute-force testing over algorithmic trust.