It's been over a month since Apple's iOS 14.5 release. The myTracker team has gathered data on actual user experience with ATT and is sharing it now, providing answers to your most pertinent questions.
Summary: according to myTracker data, only 16.9% of users globally have upgraded to iOS 14.5 or above as of 31 May 2021.
We’ve put together a chart showing that the system is being adopted fairly smoothly. We expected to see a sharp spike in week 3 after the release, but this did not happen, and we are seeing the number of devices with the new system installed grow in small steps.
Historically, iOS updates enjoyed a 60% adoption rate within a month of their official release, but this time we see a figure of below 20%. We can only guess why. Perhaps this is how it was originally intended by Apple, to give the industry and the market time to smoothly adapt to changes and avoid the shock.
Summary: according to our data, as of 31 May 2021, the global average figure is 21%.
There are two main approaches to measuring the ATT opt-in rate:
Approach 1: “All divided by all”. In this case, all ATT metrics are used: tracking authorized/denied, requested/not requested, or impossible to request. And proceeding from all these metrics, the opt-in percentage is calculated. This option used in some of the measurements has recently caused quite a stir in the market, as it yields a very low rate.
To us, calculation approach 2 seems more reasonable and correct. It is based on the number of users who saw the ATT prompt and either authorized or denied tracking. All our above calculations were based on this approach.
You can find out how many of your users have opted to authorize or deny tracking in myTracker's ATT status report:
Another thing to remember is that besides the opt-in/opt-out, there are also restricted users for whom data is unobtainable because they are either minors or using corporate devices whose IDs may be inaccessible. According to our data, the share of such users is about 12%.
Summary: most numbers you see on the Internet do not take into account mutual opt-ins.
True, online materials and webinars often mention 30–40% opt-in rates, which seems good enough and we could work with that. However, it is rare that somebody would mention in these materials that correct IDFA attribution and analytics by advertising source require mutual opt-ins, i.e. the consent to sending IDFA data needs to be obtained both in your app and in the advertising platform's app (publisher’s app).
Let's consider Facebook as an example. If your user has opted to allow their data processing in your app but has not given consent in the Facebook app, you will not see this user included in your attribution report; and vice versa, too.
This is a really important point to bear in mind, but somehow it is often overlooked.
Summary: you can do this using the Bayes rule, but you are not going to like the result.
To calculate the actual percentage of users you'll be able to see with IDFA, you can apply the Bayes theorem.
→ Let’s assume that X app has a very high opt-in rate of 60%.
→ Facebook is their main ad purchase source. Facebook's opt-in rate is 10%.
→ Let’s assume that 100% of users who have opted in on Facebook will also choose to do so in the X app.
Therefore, even with a high opt-in rate in the app but a low opt-in rate on, say, Facebook, you get only 16% of real users shown in your attribution data for Facebook based on their IDFA.
Unfortunately, the actual user IDFA opt-in rate is low.
Summary: fewer postbacks, limits on campaigns, timers, duplicate installs, and no Conversion Value reporting for small data.
Many networks transmit aggregate data instead of postbacks, and thus app owners, marketing experts, and analysts are still unable to get granular and accurate data in SKAdNetwork campaigns.
This means that no matter how detailed your Conversion Value settings are and how many parameters have been added, some of the networks will be unable to optimize towards these events when purchasing. The only option here is to wait until these processes get deployed by the networks.
9 for FB/Instagram and 11 for TikTok per app. On the whole, there is a limit of a total of 100 campaigns in all ad networks per app.
All events are fed to analytical systems no earlier than 24 hours after the occurrence and are reset when new ones occur in the Conversion Value scheme. This is yet another factor posing challenges for precise analytics.
Marketers now have to develop two separate strategies: for IDFA and for SKAdNetwork users.
Some of the networks count repeat installations as new, which creates the risk of paying for the same user twice.
It transpired that Apple has a certain threshold as to the amount of data for which it starts sending CV postbacks. This means that if target events rarely occur in your app, CV postback will only be sent after a certain amount of data is collected.
Summary: a broader Apple functionality and new tools provided by MMP.
This should resolve the latter issue from the question above. In the next iOS version, Apple is expected to determine the IP address itself, thus anonymizing its users even more. This may provide a basis for giving developers and marketers an even broader range of analytical tools.
This is a story about the app-to-web attribution which is so far available for Safari browser and iOS devices only, but soon may be added for Chrome, Mozilla, and other browsers too. This feature will be of most interest to e-commerce projects, where users interact both with the app and the website.
Analytical systems, such as myTracker, are working at full power and have already started implementing new alternative tools and techniques. Therefore, in the near future, we can expect new products to be launched or existing alternative attribution tools provided by MMP systems to be improved.
You may have read about the importance of predictive analytics in SKAdNetwork. If you've been meaning to try it for a while, now is exactly the time when you should start implementing it in your projects. Using predictive analytics, one can get an idea of which way to optimize their advertising or product without relying on attribution breakdowns.
Summary: use all available tools and experiment.
When buying traffic, you should use a mix of SKAdNetwork and IDFA, while they are still available.
Make sure that you’ve adjusted the settings to report exactly the events that matter. Many analysts love tracking users’ every step, but this is no longer an option with ATT, so you need to carefully select events that will let you optimize your campaign via CV postbacks in the short term.
While uncertainty still exists, you may want to improve your ASO and post more high-quality content on your app.
In myTracker, predictive analytics with LTV forecasting is an automatic feature – you only need to add the app and adjust the payment settings. This way, you will be able to see accurate forecasts of revenue generated by a certain user cohort with a 30, 60, 90, and 180-day breakdown.
Summary: if used smartly, this screen can help you increase IDFA opt-in rates.
The pre-prompt screen is the main and most efficient way of making an impact. Correct settings of your ATT opt-in screen are key for the number of IDFA you will get. Please find below some of the tips we’ve put together from our partner apps’ experiences.
Summary: there are useful settings examples provided by Apple and several hundred more that are publicly available.
We've covered the tricks of creating efficient pre-prompt screens in iOS 14.5+ in our blog article.
One of the most helpful examples is provided in the Apple manual itself. It has it all: a friendly introduction, individual offers, short-term wins, nice design, and concise text.
Check out the examples of ATT screen settings in apps around the world in the largest open base at attprompts.com.
Summary: it's too early to relax
Less than 20% of devices have switched to iOS 14.5+ so far. All of the figures are prone to change, so, for now, it is important to keep an eye on it and leverage analytical services more.
myTracker offers flexible solutions for working with iOS 14.5+ that will maintain ATT-based attribution confidentiality and precision, protect from all types of fraud and help you quickly adapt to the existing environment.