MMP for Subscription Apps: Why Attribution That Stops at Install Costs You Money
Your Meta dashboard says CPI is $3. Your Google dashboard says CPI is $4.50. Meta looks like the better channel.
But here is the number neither dashboard shows you: cost per subscriber. If 10% of Meta installs start a trial and 30% of those trials convert to paid, that $3 install actually costs $100 to produce a subscriber. If 20% of Google installs start a trial and 45% convert to paid, that $4.50 install costs $50 per subscriber.
Meta looks cheaper by CPI. Google is cheaper by half on the metric that actually matters.
This gap is invisible without a Mobile Measurement Partner (MMP). Platform dashboards show installs. Subscription revenue happens days or weeks later — outside the platform's attribution window. Without a system that connects install source to subscription outcome, you are making budget decisions on the wrong number.
Key Takeaways
- CPI is the most visible UA metric and the worst predictor of subscription profitability. A $3 CPI channel can cost $100/subscriber while a $4.50 CPI channel costs $50/subscriber. The cheaper channel by CPI is twice as expensive by CPS.
- Ad platforms cannot track the full subscription funnel. Install attribution is platform-native. Trial-to-subscribe conversion happens days later — and each platform reports from its own perspective, double-counting conversions across channels.
- An MMP applies one attribution model across all channels. Every conversion is counted once. The full funnel — Install → Trial → Subscribe — is visible by channel.
- Privacy changes made platform data less reliable. ATT and SKAdNetwork reduced deterministic tracking. For subscription apps, where conversion is delayed, the gap between platform-reported data and reality is wider than for any other app category.
- Airbridge Core Plan is built for this problem. Funnel, Retention, and Revenue reports by channel — with RevenueCat/Adapty integration. Start with 15K free attributed installs.
The $50 Gap Hiding in Your Dashboard
Subscription app economics run on one formula:
CPS = CPI ÷ (install-to-trial % × trial-to-paid %)
This formula reveals why CPI alone is misleading. Two channels with different CPIs can produce the opposite conclusion when you look at subscriber cost:
Channel A looks 33% cheaper by CPI. Channel B is 50% cheaper by CPS.
This is not a hypothetical edge case. It is the standard reality for subscription apps. Different channels attract users with different intent levels, different onboarding behavior, and different willingness to pay. CPI captures none of this. CPS captures all of it.
The problem: calculating CPS by channel requires connecting three data points — install source, trial event, and subscription event — across a timeline that can span days or weeks. No single ad platform has this data. The install belongs to Meta. The trial event lives in your app. The subscription lives in RevenueCat or Adapty. An MMP is the system that connects them.
Deep dive: CPI vs Cost per Trial vs Cost per Subscription: The Three Metrics That Actually Matter
What Ad Platforms See — and What They Miss
Each ad platform measures conversions inside its own ecosystem, using its own attribution rules. This creates three problems for subscription apps:
1. Attribution stops at install (or early events)
Meta can see that a user clicked an ad and installed your app. It can see a trial start if you send the event back. But the subscription — which happens 3, 7, or 30 days later — falls outside the platform's attribution window. Meta reports the install as a success. Whether that install became a paying subscriber is invisible.
2. Each platform claims credit for the same user
A user sees your TikTok ad on Monday, your Meta ad on Wednesday, and installs on Thursday. Both platforms report the install. Both count it as their conversion. Your Meta dashboard says 1,000 installs. Your TikTok dashboard says 800 installs. Your actual install count is 1,200. The overlap inflates total reported installs by 50%.
3. Retargeting absorbs credit for organic users
Retargeting campaigns show the highest conversion rates in platform dashboards — because they target users who were already going to convert. Without cross-channel deduplication, retargeting gets over-credited while upper-funnel channels look unprofitable and get cut.
The result: budget decisions based on platform dashboards favor cheap installs (CPI), over-credit retargeting, and cannot distinguish channels that produce subscribers from channels that produce downloaders.
What an MMP Adds: One Attribution Model Across All Channels
An MMP sits between ad platforms and your app. It collects attribution data from all channels, applies one consistent model, and produces a single source of truth.
For subscription apps, this changes three things:
- Deduplication. Every conversion is counted once. A user who saw TikTok and Meta ads before installing is attributed to one source — not counted twice.
- Full-funnel visibility. The MMP tracks Install → Trial → Subscribe as a connected sequence, not three isolated events. Conversion rates at each stage are visible by channel.
- Subscription data connected to install source. The MMP links billing data (RevenueCat, Adapty) back to the campaign that drove the install — so you can calculate CPS by channel. This is the connection that platform dashboards structurally cannot provide.
Why Privacy Changes Made This Worse
Apple's App Tracking Transparency (ATT) and SKAdNetwork reduced the data that ad platforms receive. Deterministic user-level tracking dropped significantly — platforms now rely more on modeled conversions and aggregated reporting.
For most app categories, this made measurement harder. For subscription apps, it made measurement structurally broken. Here is why:
SKAN reports conversions within a limited postback window. Subscription conversion happens days or weeks after install — often after the SKAN window has closed. The platform receives the install signal but never sees the subscription. The gap between what the platform reports and what actually happened is wider for subscription apps than for any other app category.
An MMP with SKAN support and configurable attribution rules bridges this gap by combining SKAN data with deterministic signals and applying consistent logic across channels — so subscription events that fall outside platform windows are still attributed to their source.
The Three Questions an MMP Answers for Subscription Apps
The value of an MMP for subscription apps comes down to three questions that platform dashboards cannot answer:
Question 1: Which channels produce paying subscribers?
Platform dashboards show install volume and CPI. An MMP shows cost per subscriber by channel — the metric that actually determines whether a campaign is profitable.
How Core Plan answers this: The Actuals Report shows Install, Start Trial, and Subscribe events with full attribution across Meta, Google, Apple Search Ads, and TikTok. The Revenue report connects subscription revenue to acquisition source. You can see not just which channel drove the install, but which channel drove the revenue.
Question 2: Where do users drop off — and does it vary by channel?
Blended conversion rates hide channel-specific problems. Your overall trial-to-paid rate might be 30% — but if Meta users convert at 15% and Google users convert at 50%, you are averaging two very different outcomes.
How Core Plan answers this: The Funnel report shows conversion rates at each stage — Install → Trial → Subscribe — segmented by channel. If Meta traffic drops off at trial start while Google traffic drops off at subscription, you know the problem is different for each channel and requires a different fix.
Question 3: Are subscribers from different channels retaining differently?
A channel that produces subscribers who churn after one month is not the same as a channel that produces subscribers who renew for a year. CPS alone does not capture this — you need retention data by acquisition source.
How Core Plan answers this: The Retention report shows cohort retention by channel. You can compare 30-day, 60-day, and 90-day retention rates for subscribers acquired from Meta versus Google versus TikTok — and shift budget toward channels that produce subscribers who stay.
What You See Without an MMP vs With Airbridge Core Plan
Stop Optimizing for CPI. Start Measuring What Matters.
Subscription app economics are simple: acquire subscribers for less than they are worth. But platform dashboards show the install. Billing platforms show the subscription. No single system shows both — unless you have an MMP.
That full-funnel view is what turns isolated metrics into decisions. Which channels to scale. Which to cut. Where the funnel breaks. Whether subscribers retain.
Core Plan is built for this. Full-funnel attribution across Meta, Google, Apple Search Ads, and TikTok. Funnel, Retention, and Revenue reports by channel. Native RevenueCat and Adapty integration. Standard events from day one.
Airbridge Core Plan
