The Sharded App Install Strategy: Why Services Brands Are Deploying 264 Ad Accounts for App Promotion

Every major ad platform tells you the same thing regarding app promotion: consolidate your campaigns, feed the algorithm data, and let machine learning find your best users. The days of hyper-segmented ad groups are supposed to be over. If you want app installs, you are told you need volume and patience.

But when we look at how certain services brands actually deploy their capital in the real world, the data contradicts the standard platform playbooks.

We recently analyzed a cohort of 562 app promotion ads run by services brands over a 90-day window [1]. What we found was not a consolidated, algorithmic powerhouse. Instead, we found a shattered, decentralized network of micro-campaigns that fundamentally breaks every rule of modern user acquisition.

Let us break down the architecture of this anomaly.

The Sharded Ad Account Strategy

The most glaring signal in this cohort is the infrastructure used to deploy the ads. The 562 app promotion ads belong to just four distinct brands [2]. However, those four brands are not running their user acquisition out of four central ad accounts.

They are deploying these ads across 264 distinct ad accounts [2].

That is an average of 66 ad accounts per brand, pushing app install campaigns simultaneously. In a normal environment, a brand might use two or three accounts to separate billing entities or distinct geographic regions. Using hundreds of separate accounts for app promotion points to a very specific, aggressive structural necessity.

Why decentralize to this extreme?

1. The Franchise Model: Many service apps, such as ride-hailing, food delivery, or local home services, rely on franchisees or city-specific operators. Instead of corporate headquarters funding user acquisition from a central pool, each local operator is responsible for driving downloads in their own territory, using their own localized ad account.
2. Affiliate Networks: Some brands offload app promotion to massive affiliate networks. The brand pays a fixed bounty per install, and individual affiliates spin up their own ad accounts to hunt for cheap conversions. If an account gets flagged or ad costs rise, the affiliate simply abandons it and spins up another.
3. Bypassing Algorithmic Averages: Centralized accounts blend costs. A brand might get cheap installs in rural areas but pay a premium in city centers. By isolating campaigns into hyper-local ad accounts, operators force the platform to treat each micro-market as an independent auction.

The Rigid Micro-Budget Cap

You might expect these hundreds of localized accounts to be spending heavily to build local momentum. They are not.

Across the entire cohort, the median spend per ad sits rigidly at 23.78 Euros [3]. Furthermore, the 25th percentile and 75th percentile for spend are exactly the same: 23.78 Euros [3].

This uniformity is a massive signal. This is not organic pacing where the algorithm slowly ramps up spend based on performance. A flat distribution exactly at this price point across hundreds of ads means these operators are deploying rigid, automated budget caps.

When an ad hits that specific threshold, the system shuts it down.

The Reality of Ghost Reach

When you cap spend at such a low threshold and shatter your delivery across 264 accounts, your actual reach becomes microscopic.

The ads in this cohort achieved a median total European reach of just 82 users [4]. The median daily reach was zero [4].

Think about what that means for an app promotion campaign. The creative is being shown to fewer than 100 people before the plug is pulled. This is the definition of ghost reach. On a macro scale, these campaigns do not exist.

But in the context of hyper-local services, reaching exactly 82 highly targeted people might be the entire point.

• Hyper-Radius Targeting: If a local delivery app is trying to acquire users within a single postal code or a specific university campus, reaching 82 qualified local students is far more valuable than reaching 10,000 users statewide.
• Burn-and-Churn Creative: Operators might be testing hundreds of slight creative variations featuring local slang, specific street names, or regional landmarks. They blast it to a tiny cohort. If someone clicks and installs, they might scale it in a different account. If not, they kill it instantly.

What Operators Can Learn

If you manage user acquisition for an app, your dashboard is likely begging you to increase budgets and consolidate your audiences. The platforms want broad targeting because it makes their internal inventory management easier and maximizes liquidity in the auction.

But this cohort proves that a completely different playbook exists in the wild.

For services brands with a localized footprint, standard app promotion rules do not always apply. If your lifetime value is tied heavily to local density, broad algorithmic targeting will waste your money acquiring users in dead zones.

The Takeaways for Your Strategy:

• Do not fear fragmentation if the business model demands it. If your service requires extreme geographic density, segmenting by local ad accounts ensures your budget is not hijacked by cheaper, but useless, out-of-bounds impressions.
• Automate your kill switches. The operators in this cohort do not manually pause ads. They have strict rules cutting off spend. If you are testing high volumes of local creative, build automated rules to cull the losers before they drain your daily budget.
• Accept small numbers. A reach of 82 people looks like a failure on a corporate dashboard. But if those 82 people live on the exact street where your service just launched, it is highly efficient marketing.

The next time your platform representative tells you to consolidate your accounts to optimize for app installs, remember the 264 accounts quietly grinding out localized installs one tiny micro-test at a time. The algorithm is a tool, not a master.