Predictive Analytics and SKAdNetwork | How to Predict LTV in iOS 14.5+

Previously, there were a number of ways to assess ad campaign performance for iOS apps. The IDFA, a unique device identifier for advertisers, helped track user–ad interactions and post-install in-app activities.

With the introduction of ATT in iOS 14.5, over 69% of iPhone and iPad owners started denying access to their IDFAs. The restrictions on IDFA transfer have limited the mobile analytics options, with ad campaign efficiency now having to rely on innovative solutions for confidential and accurate attribution.

One of them is LTV forecasting.

In this article, we will explain LTV’s significance for your business, discuss SKAdNetwork’s impact on the mobile ad market, and talk about various LTV forecasting solutions for the new reality.

What is LTV?

Lifetime Value (LTV) is the revenue generated by a user up until the moment they stop using the app.

For example, if a user viewed $USD 3 worth of ads in the first month, made a $USD 4 purchase in the second month, and rarely launched the app in the third month, making no purchases and finally uninstalling the app altogether, that user’s LTV would be equal to $USD 7.

Why Measure a User’s LTV?

Users installing apps by responding to ads have a Customer Acquisition Cost (CAC). If you spend $USD 1,000 on an advertising campaign that brings in 100 users, then the CAC is $USD 10.

Your campaign would be ineffective and loss-making if the users’ average LTV is lower than the average CAC.

Why Forecast LTV?

LTV forecasting is important for an app’s long-term development. It helps predict when your marketing spend will pay off – in a week, a month, or a couple of years.

LTV forecasting allows app owners to confidently grow their business and avoid using ad channels that would result in a poor LTV : CAC ratio.

LTV forecasts can also help you estimate your ad campaign ROI when traditional mobile analytics tools are unavailable.

Why Traditional Mobile Analytics Tools No Longer Work

We'll go back a bit to explain how Apple’s new confidentiality policy has affected the mobile ad market.

Attribution before iOS 14.5

Prior to April 26, 2021, attribution for iOS devices was based on the IDFA, an identifier generated by the device for the benefit of all stakeholders:

  • Users received personalized ads.
  • Advertisers took advantage of retargeting and lookalike audience strategies.
  • Analytics platforms measured event-based metrics, created funnels, and handled attribution.
  • App owners had a better understanding of how users interact with their apps.
attribution before iOS 14.5


Our research shows that even before the upgrade to iOS 14.5, about 20% of users limited IDFA access on their devices, but it did not create major problems for the entire mobile ad, analytics, and attribution industry in terms of data collection and processing.

Attribution after iOS 14.5

On April 26, 2021, Apple released the iOS 14.5 update, putting into effect a new privacy policy, ATT, and its own attribution framework – SKAdNetwork.

Now, when you launch an app for the first time on iPhone/iPad, the system automatically requests user consent for data transfer. According to myTracker data, 69.1% of global iOS users deny access to their IDFAs. 

Ad networks are no longer able to track impressions and user–ad interactions with a breakdown by device, which has reduced the accuracy of user attribution by analytics platforms. The attribution process as we used to know it is no more, although its individual components still exist.

Apple has come up with a solution in the form of SKAdNetwork, which helps measure the number of conversions and LTV while not disclosing the link between the ad and the user it helped to acquire. As a result, it is now much more difficult for developers and marketing experts to adequately assess an ad campaign’s ROI.

How myTracker works with SKAdNetwork data
How myTracker works with SKAdNetwork data


This affects the entire industry, as advertisers cut their advertising budgets for iOS, causing changes in monetization schemes for apps and games in the App Store. Hence, ad monetization for apps becomes more aggressive, and these ads might not even correspond to a user’s interests.

You can find more answers about attribution after iOS 14.5 in our FAQ or in our blog post on the 10 most popular questions about the iOS 14.5 update.

How LTV Forecast Facilitates the Use of SKAdNetwork

Today, analytics platforms are no longer able to attribute individual users to a specific source or campaign, but they can still predict a user LTV for various periods.

LTV forecasting can help you gauge the impact of updates on an app, evaluate the efficiency of monetization models and optimize ads, even without individual user data.

LTV Forecast for iOS 14.5+

Currently, there are two types of LTV forecasting: using standard SKAdNetwork mechanisms to send Conversion Values or through predictive analytics.

Predicting LTV with SKAdNetwork Conversion Value functionality

Conversion Value is a number between 0 and 63 sent by SKAdNetwork to app owners. Based on this data, you can create a table matching LTVs to CVs.

An LTV–CV table example

Each number from 0 to 63 can be viewed as corresponding to a range:

  • CV = 0 at install;
  • CV = 1 for a $USD 0–10 purchase;
  • CV = 2 for a $USD 10–20 purchase;

etc.

Based on LTV, retention, and a couple of simple formulas, you can forecast an approximate ad campaign revenue.

However, this solution has its drawbacks:

  • It is hard to implement on your own.
  • Not all ad networks can use this data to optimize ads (due to the lack of consistent principles and a decoding table).
  • It can’t be altered and adapted to different ad campaigns, as it applies to the entire app.
pros and cons of skadnetwork conversion values

LTV Forecast Using Predictive Models

A predictive model is a set of algorithms based on machine learning and designed to recreate attributions using existing data:

  • Payments by organic users.
  • Payments by users who gave ATT consent and were clearly attributed to a particular ad via their IDFAs.
  • The number and content of CV postbacks sent by Apple.

Using machine learning models is the quickest and easiest way to evaluate the performance of ad campaigns without having to ask for developers' help. 

Predictive models don’t identify users. They seek to recreate the attribution picture to predict LTV and help developers and marketing teams evaluate an ad campaign’s ROI.

LTV forecasts work at a partner level, which is sufficient to evaluate ad campaigns soon after launch without having to waste your budget. The prediction horizons can be for 30, 60, 90 to 180 days.

Check out our blog to learn more about how to set up a SKAdNetwork-based LTV forecast in myTracker.

Solutions for Working with SKAdNetwork on iOS 14.5+

  • SKAdNetwork support in the SDK
  • Collection of data on conversions via an API
  • Fraud protection
  • Predictive analytics
  • Consent reports

All our solutions to keep attribution confidential and accurate under the ATT framework are available to you as part of myTracker’s optimum (free) plan.

Contents