The median mobile app running structured ad-network tracking on Linkrunner runs exactly one network. The top 25 per cent run two. Only the top decile run more than three.
That contradicts almost every conference talk you have heard on channel diversification. It also contradicts most agency pitches. And it is worth understanding why before you sign up for a fifth ad network and another set of pixels to maintain.
This post lays out the benchmark, explains why the dominant pattern is fewer networks not more, and offers two decision rules for when adding a network actually pays off.
The Benchmark: How Many Ad Networks Apps Actually Run
Ad network mix is the count of distinct paid acquisition channels a mobile app actively runs attributed campaigns against in a given period. It is a structural indicator of how diversified a team's user-acquisition strategy is in practice rather than on paper.
Headline figures from the Linkrunner cohort of projects with at least one ad-network-linked campaign, as of 29 June 2026:
- Median: 1 network
- P75: 2 networks
- P90: around 3 networks
- Maximum: 7 networks
- Mean: 1.6 networks
- Projects in cohort: around 115
Methodology note: campaign-level cohort with explicit ad_network_id. This captures projects running structured ad-network tracking rather than UTM-only attribution. The defensible framing is "projects running structured ad-network tracking", not "all Linkrunner customers".
The number is consistently lower than the conference-deck consensus, which assumes the median growth team runs four to six networks. That assumption comes from talks given by teams at the top of the distribution. The middle of the distribution looks very different.
Why Most Apps Run Fewer Networks Than They Think
The dominant pattern is not "we only need Meta". It is "we keep one network at scale, run a second as a hedge, and the third lives in someone's open tab".
What drives the gap between paper diversification and real diversification:
- Operational reality of running a network well. Each platform needs creative variants formatted to its specs, campaign structures the algorithm respects, bid strategies the team understands, and a measurement workflow that closes the loop. Most teams underestimate the marginal cost of channel number three or four.
- Attention is finite. Teams that scale two networks well consistently beat teams that dabble in six. The difference shows up in creative refresh cadence, not in dashboard breadth.
- The optimisation cost curve doubles roughly every additional network. A second network adds one set of duties. A fourth network adds three.
Our budget allocation framework for multi-channel app growth covers how to split spend once you have committed to a multi-network operating model. It also covers the diminishing-returns thresholds where adding the next network stops paying off.
When Adding a Network Actually Pays Off
Adding a network is the right call when the existing networks are showing measurable saturation, not when a competitor mentioned a new channel on LinkedIn.
Signals that justify the additional operational cost:
- CPI rising 3 weeks running while creative volume is held constant. This is the cleanest saturation signal because it isolates channel exhaustion from creative fatigue.
- Audience frequency above 4 with CTR declining at the same time. The algorithm has run out of fresh audience to show your ads to.
- A specific underexploited audience that an emerging network reaches at lower cost. For example, Jio Ads for Indian Tier-2 and Tier-3 reach.
Before committing budget, run an incrementality test on the new network rather than relying on its self-reported attribution. Our incrementality testing guide walks through the geo-lift and holdout designs that work for mobile apps. If creative fatigue is the underlying driver, the why your best-performing campaigns stop working after 30 days covers the refresh patterns that can extend the channel you already have before you reach for a new one.
When Adding a Network Costs You Money
Adding a network destroys margin when the existing channels still have headroom. Three diagnostic signals to check first:
- You have not exhausted Meta or Google's tail audiences. Most apps run Meta and Google on a narrow lookalike or keyword list and call that the network's ceiling. It usually is not.
- Your creative team produces fewer than 8 fresh assets per week. Adding a network without enough creative inventory means the new network runs the same tired assets for longer.
- Your attribution stack cannot dedupe credit across networks. Two networks claiming the same install is worse than one network claiming all of them.
A pattern worth flagging from attribution audits: teams running structured tracking on a single network consistently report cleaner attribution math and faster decision cycles than teams running structured tracking on four or more. The dashboard becomes simpler. The reallocation conversations become shorter. Fewer arguments about which network actually drove the conversion.
The Two-Network Operating Pattern
The P75 of the Linkrunner cohort runs two networks. That is the operating shape most growth-stage apps converge on.
What it looks like in practice:
- Primary network carries 65 to 80 per cent of paid spend. Mature creative pipeline, automated bid management, daily review cadence.
- Secondary network carries the remaining 20 to 35 per cent. Tighter creative variant set, weekly review cadence, used to hedge primary network volatility.
- Allocation is reviewed monthly, not weekly. Frequent reallocation introduces variance that hides the underlying signal.
Dashboard cuts worth setting up:
- Incremental CPI by network, computed against a holdout when possible
- Dedupe rate between networks (how often the same user shows up in both networks' installs)
- Share of organic, tracked alongside paid mix to catch cannibalisation
The beyond Facebook and Google guide covers the emerging channels worth considering for that secondary slot if you are still running a Meta-only setup.
The Top-Decile Pattern: Three to Seven Networks
The top 10 per cent of the cohort run three or more networks, with the maximum at seven. This pattern is real and it works, but only when specific infrastructure is in place.
What top-decile teams have built first:
- Event taxonomy that is platform-agnostic. One canonical event schema feeding postbacks to every network, not bespoke implementations per platform. Publicly named customers operating at this scale, including Matiks running on 130+ Meta campaigns and Jumbo Gaming tracking 200+ campaigns through Linkrunner, lean on a single shared event model.
- Postback dedupe configured. Last-click rules at the MMP layer prevent two networks from claiming the same install.
- Per-network attribution windows tuned to platform mechanics. A single 7-day click window applied uniformly across networks usually punishes one of them.
- Creative pipeline producing 15 or more variants per week. Without volume, more networks just means slower refresh.
The how to scale mobile app campaigns from Rs5L to Rs50L monthly walks through the stage gates where adding a network becomes operationally feasible.
How to Validate Your Own Position on the Distribution
The most useful self-check this week:
- Count active campaigns with an explicit ad-network identifier in the last 30 days
- Count distinct networks across those campaigns
- Compare against the benchmark: median 1, P75 2, P90 around 3
Decision rule:
- Above the median with rising blended CPI: consolidate. Pick the two networks with the strongest incremental performance and pause the rest for one quarter.
- At the median with falling blended CPI: stay where you are. You are not under-diversified, you are well-optimised.
- At the median with rising CPI and saturation signals: add one network at a time, run an incrementality test, give it a quarter.
Tools like Linkrunner expose the per-network breakdown view that surfaces active campaign counts, distinct networks, and incremental CPI alongside the cross-customer benchmark, which avoids stitching exports by hand. If you live in raw exports, the budget allocation framework covers the SQL cuts you need.
FAQ
How many ad networks does the median mobile app actually run?
The median app running structured ad-network tracking on Linkrunner runs one network. The 75th percentile runs two and the 90th percentile runs around three. The cohort is around 115 projects with explicit ad-network campaign tracking as of June 2026.
Is running more ad networks always better for growth?
No. Adding a network when existing channels still have headroom usually destroys margin because the optimisation cost compounds faster than the incremental volume. Validate saturation before adding.
When should I add a third or fourth ad network?
When CPI is rising three weeks running on existing networks at constant creative volume, audience frequency on the primary network sits above 4 with declining CTR, and your team can produce 15 or more fresh creative variants per week. Run an incrementality test before committing budget.
What does ad-network mix look like for high-spend apps?
The top decile of the Linkrunner cohort runs three to seven networks, with most clustering at three to four. These teams have already invested in platform-agnostic event taxonomy, postback dedupe, and high-volume creative pipelines.
How do I count active networks accurately when reporting?
Filter to campaigns active in the last 30 days with an explicit ad-network identifier, then count distinct networks. UTM-only campaigns without a network identifier should be excluded because they typically represent organic, influencer, or unstructured traffic rather than paid platforms.
What to Do Next
The highest-value action this week: count your distinct ad networks against the benchmark and decide whether you are over-diversified, well-positioned, or genuinely saturating. The honest answer for most teams sits closer to the median than they expect.
If you would like to see your ad network mix alongside incremental CPI and the cross-customer benchmark in one view, request a demo from Linkrunner and we will walk you through the per-network breakdown for your project.