Your app install numbers don’t match your ad spend because every ad platform counts installs differently, and none of them use the same attribution logic as your Mobile Measurement Partner (MMP). Ad networks report every install they can claim credit for, which often leads to double-counting across platforms. The result is that your total reported installs across Google, Meta, TikTok, and Apple Search Ads will almost always exceed what your MMP records. This article unpacks the most common reasons behind that gap and what you can do about it.
What causes the gap between ad spend and app install numbers?
The gap between ad spend and app install numbers is caused by a combination of attribution conflicts, platform-level overcounting, tracking limitations on iOS, and ad fraud. Each ad platform attributes installs to itself whenever it can, regardless of whether another platform also touched that user. Your MMP acts as the neutral source of truth, assigning each install to a single source, which is why its numbers are always lower than the sum of individual platform reports.
Several factors compound this problem. On iOS, Apple’s App Tracking Transparency (ATT) framework limits the data platforms can access, which means a significant share of installs go unattributed or are estimated using modelled data. On Android, the picture is clearer but still imperfect. Add in view-through attribution windows, cross-device journeys, and organic installs that platforms sometimes claim, and the discrepancy grows quickly.
Understanding where the gap comes from is the first step toward trusting your data again.
How does mobile attribution actually work?
Mobile attribution is the process of connecting an app install or in-app event back to the marketing touchpoint that drove it. When a user clicks an ad, a tracking link records that interaction. When the user later installs the app and opens it, the MMP matches the install to the recorded click using a device identifier or probabilistic fingerprinting, then credits that install to the correct campaign.
The MMP sits between your app and all your advertising platforms. It receives postbacks from each platform and applies a consistent attribution model to determine which touchpoint gets credit. This is why your MMP is the most reliable source for install data: it applies one set of rules across all channels, rather than each platform applying its own.
Tools like Adjust, AppsFlyer, and Branch are the most widely used MMPs in the industry. Each one integrates directly with your app via an SDK, captures install and event data at the device level, and provides a unified view of performance across all your paid and organic channels.
Why do ad platforms report more installs than your MMP does?
Ad platforms report more installs than your MMP because they each claim credit for every install that occurred after a user was exposed to their ad, regardless of what else happened in between. This is called self-attribution, and every major platform does it. When a user sees an ad on Meta, then clicks an ad on Google, and then installs the app, both Meta and Google may report that install. Your MMP, however, will only credit one of them.
There are a few specific mechanisms that drive this overcounting:
- View-through attribution: Platforms claim installs from users who saw an ad but never clicked it, based on a post-view attribution window that can range from one hour to several days.
- Extended click windows: A platform may attribute an install to a click that happened days or weeks earlier, even if other touchpoints occurred more recently.
- Organic install overlap: Some platforms attribute installs that were genuinely organic, especially when a user searched for the app by name after seeing an ad.
This is not necessarily dishonest behaviour from the platforms. It reflects how each network measures its own contribution. But it does mean you should never add up install numbers across platforms and treat that sum as your total. Your MMP total is the number to trust.
What is the difference between last-click and multi-touch attribution?
Last-click attribution gives 100% of the credit for an install to the final ad a user clicked before installing the app. Multi-touch attribution distributes credit across all the touchpoints a user interacted with before installing, based on a chosen model such as linear, time-decay, or data-driven. The difference matters because it changes which channels appear to be performing well and how you allocate budget.
Last-click is the default model used by most MMPs and ad platforms. It is simple, consistent, and easy to act on. Its weakness is that it ignores the role of earlier touchpoints, which can make upper-funnel channels like display or video look ineffective even when they contribute meaningfully to conversion.
Multi-touch attribution gives a more complete picture of the user journey, but it requires more data, more configuration, and more careful interpretation. On iOS, where user-level data is restricted due to ATT and the deprecation of the IDFA, running a full multi-touch model is increasingly difficult. Many teams use a hybrid approach: last-click for day-to-day campaign optimisation, and incrementality testing or modelled attribution for strategic budget decisions.
How can ad fraud inflate app install numbers?
Ad fraud inflates app install numbers by generating fake installs, clicks, or impressions that appear legitimate to ad platforms and sometimes to MMPs. Common fraud types include click flooding, where fraudsters fire thousands of clicks to hijack organic installs, and install farms, where real or emulated devices install apps repeatedly to collect CPI payouts. The result is that your reported install numbers rise while your actual user base and engagement metrics stay flat.
Fraud is particularly prevalent in programmatic and performance networks where payouts are tied directly to installs. If your install numbers look strong but your day-one retention, session depth, or in-app conversion rates are unusually low, fraud is a likely explanation.
Your MMP has fraud detection tools built in, and most offer integrations with dedicated fraud prevention partners. Enabling these tools and monitoring post-install behaviour alongside install volume is the most effective way to catch fraud early. A sudden spike in installs from a new traffic source with no corresponding improvement in downstream metrics is a clear signal to investigate.
How do you fix install tracking discrepancies in your app?
Fixing install tracking discrepancies starts with auditing your MMP setup to ensure the SDK is correctly implemented, all in-app events are firing accurately, and attribution windows are configured consistently across platforms. Most discrepancies have a technical root cause that can be identified and resolved with a structured review.
Here are the most important steps to take:
- Verify your SDK integration: Confirm the MMP SDK is installed in the correct version and that it initialises before any other tracking calls in your app.
- Audit your deep links and tracking URLs: Broken or misconfigured tracking links are a common cause of unattributed installs. Check that every campaign uses correctly structured URLs with the right parameters.
- Review your attribution windows: Make sure the click and view-through windows in your MMP match what each ad platform is using. Mismatched windows create artificial discrepancies.
- Enable SKAdNetwork on iOS: For iOS campaigns, ensure SKAdNetwork is properly configured so you receive conversion data even from users who have not consented to tracking.
- Reconcile platform data regularly: Compare MMP data against platform-reported data at least weekly. A consistent 10-20% gap is normal; anything larger warrants investigation.
- Activate fraud detection: Enable your MMP’s built-in fraud protection and review flagged installs regularly.
If you are running campaigns across Apple Search Ads, Google, Meta, and TikTok simultaneously, the complexity of managing attribution correctly increases significantly. Having a clear data governance process, with one person or team responsible for MMP hygiene, makes a meaningful difference to data quality over time.
Our app growth services are built around getting this infrastructure right from the start, so that every campaign decision is based on data you can trust. If you want an expert to review your current setup, book a free consultation and we will walk through your tracking setup together.
What level of discrepancy between ad spend and installs is acceptable?
A discrepancy of 10 to 20% between ad platform-reported installs and MMP-attributed installs is generally considered acceptable in mobile marketing. This range reflects normal differences in attribution logic, view-through claims, and organic overlap. A discrepancy above 20% is a signal that something in your tracking setup needs attention, or that fraud may be a factor.
The more important benchmark is internal consistency. If your discrepancy rate is stable week over week, you can account for it in your planning and still make reliable decisions. If it fluctuates significantly or spikes suddenly, that instability is more disruptive to your decision-making than the absolute size of the gap.
On iOS specifically, the removal of the IDFA and the limitations of SKAdNetwork mean that unattributed installs are a structural reality rather than a sign of a broken setup. Industry experience shows that iOS attribution rates have declined since ATT was introduced, and teams that benchmark their iOS performance against pre-ATT numbers will consistently feel like their ROAS doesn’t match reality. Adjusting your attribution expectations for iOS and leaning on aggregated modelling where user-level data is unavailable is now standard practice.
The goal is not a perfect match between ad spend and install numbers. The goal is a consistent, well-understood measurement framework that lets you make confident budget decisions and optimise campaigns based on reliable signals. That is what good attribution tracking delivers.
Frequently Asked Questions
Which MMP should I choose for my app — Adjust, AppsFlyer, or Branch?
The right MMP depends on your app’s scale, budget, and the platforms you run campaigns on. AppsFlyer tends to be the most feature-rich and is popular with larger teams running campaigns across many networks, while Adjust is well-regarded for its clean interface and strong fraud protection. Branch is a strong choice if deep linking and cross-platform journeys are a priority for your product. Most MMPs offer a free trial or a starter tier, so it is worth testing the one that best integrates with your existing ad networks and internal reporting tools before committing.
How does SKAdNetwork work, and why does it make iOS attribution so difficult?
SKAdNetwork (SKAN) is Apple’s privacy-preserving attribution framework that allows ad networks to receive conversion data without accessing any user-level identifiers. Instead of reporting individual install events, SKAN sends aggregated, delayed postbacks to the ad network after a conversion value is registered — with no device-level data attached. This makes it impossible to build a full user journey or run traditional multi-touch attribution on iOS, and the delay in postbacks (which can be up to several days) means you are always working with lagged data. The practical implication is that iOS campaign optimisation requires a higher tolerance for uncertainty and a greater reliance on aggregated signals and modelled data.
What should I do if my MMP is showing a large number of unattributed installs?
A high volume of unattributed installs usually points to one of three issues: broken or missing tracking links in your campaigns, a misconfigured or outdated MMP SDK in your app, or a structural attribution gap caused by iOS ATT opt-outs. Start by auditing your tracking URLs for every active campaign to confirm they are correctly formatted and routing through your MMP. Then check your SDK version and initialisation order within the app. If the unattributed installs are concentrated on iOS, a portion of them are likely organic or privacy-attributed users that cannot be matched to a specific campaign — this is expected behaviour post-ATT, and probabilistic or modelled attribution can help recover some of that visibility.
Can I run multi-touch attribution on Android if iOS is too restricted?
Yes, Android still supports user-level attribution through Google’s advertising ID (GAID), which means you can run more complete multi-touch models on Android traffic without the same structural limitations that exist on iOS. This makes Android a useful environment for testing attribution models and understanding the full user journey before applying aggregated insights to your iOS strategy. Keep in mind that Google has announced plans to introduce its own Privacy Sandbox for Android, which will eventually restrict user-level tracking there too, so building comfort with modelled and aggregated attribution now is a worthwhile investment regardless of platform.
How do I know if my app install campaigns are affected by click flooding fraud?
Click flooding is one of the harder fraud types to detect because the clicks themselves look real — the volume is just artificially inflated to increase the chance of claiming credit for organic installs. The clearest signal is an unusually long time-to-install distribution from a specific traffic source: if a large proportion of attributed installs are happening many hours or days after the click, that source may be flooding clicks to capture organic conversions. Your MMP’s fraud dashboard will typically flag this pattern, and you should also look for sources with very high click volumes but low click-to-install rates alongside suspiciously strong downstream metrics that don’t match the network’s typical quality profile.
Should I use the same attribution window across all my ad platforms?
Ideally, yes — using consistent attribution windows across all platforms and matching them to what is configured in your MMP is the single most effective way to reduce artificial discrepancies. When your MMP uses a 7-day click window but a platform is reporting on a 28-day click window, installs that fall outside your MMP’s window will appear in the platform report but not in your MMP data, creating a gap that looks like a tracking problem but is actually a configuration mismatch. Review the default windows each platform applies and align them as closely as possible with your MMP settings, keeping in mind that tighter windows generally reduce overcounting but may also lower attributed volume.
What is incrementality testing, and when should I use it instead of standard attribution?
Incrementality testing measures the true causal impact of your advertising by comparing outcomes between a group that was exposed to your ads and a holdout group that was not. Unlike standard attribution, which assigns credit based on touchpoints, incrementality tells you how many installs would not have happened without your campaign — the actual lift your spend is generating. It is most valuable when you are making large budget allocation decisions, evaluating a new channel, or when you suspect your last-click attribution is over-crediting a particular platform. Running incrementality tests requires enough traffic volume to create statistically significant test and control groups, so it is typically more practical for teams spending at meaningful scale rather than those in early growth stages.
Related Articles
- Should you advertise your app before or after launch?
- How does app store optimization help advertise your app?
- What platforms should you use to advertise your mobile app?
- How do you advertise an app using Apple Search Ads?
- What makes a good app store icon?
- What is the ideal character count for app store metadata?
- How do you write an app store description that converts?
- What is the difference between an app subtitle and short description?
- What is localized keyword research for app stores?
- What is app store reputation management?
EN 
NL 