Thomas Petit (Independent Growth Consultant) unpacks with Shamanth Rao (Founder at RocketShip HQ - Mobile UA Agency) the changes that ATT brought to the Apple Search Ads landscape and how to adapt.
Apple designed a new measurement framework called AdService, which was rolled out in March/April 2021, with an attribution that does not rely on IDFA anymore.
Apple Search Ads’ attribution doesn’t depend on the IDFA but more importantly doesn’t depend on the ATT consent status, which means ASA needs to be handled slightly differently.
The payload when a user is coming from one of your ASA campaigns is pretty granular: you can get keyword-level data and you can get retention and other post-install metrics.
If a user has opted-in to ATT (and you get it early enough), Apple gives you the exact timestamps of the click (advanced payload).
In most cases, ATT and the new ASA attribution don’t have major implications and you can keep your old campaign structure. There are still some inventory price differences.
Thomas still likes to have ad groups (regardless of how they group keywords - meaning, LTV, etc.) “duplicated” so that he ends up with 3 ad groups targeting New Users LAT Off, Returning Users LAT off and All Users. It’s a more complex structure (and there is audience overlap), but the conversion and ARPU or retention performance can be entirely different depending on the targeting. It also helps if/when deciding from where to pull back spend.
Users you don’t target if you add an 18+ criteria: underaged users, some blocked users based on device (e.g. education) but more importantly users that deactivated “Apple Ads personalization”.
The assumption is that only a tiny portion of users opt-out of Apple Ads personalization (the setting is not easy to find), but nobody really knows: unlike with LAT, we can’t monitor it and it has not been shared.
The LAT On % was always going up. With ATT, users that were LAT On have been automatically opted-out of Ads personalization. The Ads personalization % might go down as new devices are introduced, because less people will be opted out “by default”.
Limiting targeting only to new and returning can lower down your potential reach, but if you’re strictly focused on performance (or get bad performance when targeting All Users) then it might make sense to not target All Users
Custom Product Pages will allow you to customize the experience: show different listings depending on new or returning users, depending on user intent and USP.
Apple designed a new measurement framework called AdService, which was rolled out in March/April 2021, with an attribution that does not rely on IDFA anymore.
Apple Search Ads’ attribution doesn’t depend on the IDFA but more importantly doesn’t depend on the ATT consent status, which means ASA needs to be handled slightly differently.
The payload when a user is coming from one of your ASA campaigns is pretty granular: you can get keyword-level data and you can get retention and other post-install metrics.
If a user has opted-in to ATT (and you get it early enough), Apple gives you the exact timestamps of the click (advanced payload).
In most cases, ATT and the new ASA attribution don’t have major implications and you can keep your old campaign structure. There are still some inventory price differences.
Thomas still likes to have ad groups (regardless of how they group keywords - meaning, LTV, etc.) “duplicated” so that he ends up with 3 ad groups targeting New Users LAT Off, Returning Users LAT off and All Users. It’s a more complex structure (and there is audience overlap), but the conversion and ARPU or retention performance can be entirely different depending on the targeting. It also helps if/when deciding from where to pull back spend.
Users you don’t target if you add an 18+ criteria: underaged users, some blocked users based on device (e.g. education) but more importantly users that deactivated “Apple Ads personalization”.
The assumption is that only a tiny portion of users opt-out of Apple Ads personalization (the setting is not easy to find), but nobody really knows: unlike with LAT, we can’t monitor it and it has not been shared.
The LAT On % was always going up. With ATT, users that were LAT On have been automatically opted-out of Ads personalization. The Ads personalization % might go down as new devices are introduced, because less people will be opted out “by default”.
Limiting targeting only to new and returning can lower down your potential reach, but if you’re strictly focused on performance (or get bad performance when targeting All Users) then it might make sense to not target All Users
Custom Product Pages will allow you to customize the experience: show different listings depending on new or returning users, depending on user intent and USP.
Apple designed a new measurement framework called AdService, which was rolled out in March/April 2021, with an attribution that does not rely on IDFA anymore.
Apple Search Ads’ attribution doesn’t depend on the IDFA but more importantly doesn’t depend on the ATT consent status, which means ASA needs to be handled slightly differently.
The payload when a user is coming from one of your ASA campaigns is pretty granular: you can get keyword-level data and you can get retention and other post-install metrics.
If a user has opted-in to ATT (and you get it early enough), Apple gives you the exact timestamps of the click (advanced payload).
In most cases, ATT and the new ASA attribution don’t have major implications and you can keep your old campaign structure. There are still some inventory price differences.
Thomas still likes to have ad groups (regardless of how they group keywords - meaning, LTV, etc.) “duplicated” so that he ends up with 3 ad groups targeting New Users LAT Off, Returning Users LAT off and All Users. It’s a more complex structure (and there is audience overlap), but the conversion and ARPU or retention performance can be entirely different depending on the targeting. It also helps if/when deciding from where to pull back spend.
Users you don’t target if you add an 18+ criteria: underaged users, some blocked users based on device (e.g. education) but more importantly users that deactivated “Apple Ads personalization”.
The assumption is that only a tiny portion of users opt-out of Apple Ads personalization (the setting is not easy to find), but nobody really knows: unlike with LAT, we can’t monitor it and it has not been shared.
The LAT On % was always going up. With ATT, users that were LAT On have been automatically opted-out of Ads personalization. The Ads personalization % might go down as new devices are introduced, because less people will be opted out “by default”.
Limiting targeting only to new and returning can lower down your potential reach, but if you’re strictly focused on performance (or get bad performance when targeting All Users) then it might make sense to not target All Users
Custom Product Pages will allow you to customize the experience: show different listings depending on new or returning users, depending on user intent and USP.
Notes for this resource are currently being transferred and will be available soon.
The LAT cohort was always very peculiar with Apple Search Ads.
Content from 2017 (cf. The Mysterious LAT, and How to Tame It – with Thomas Petit) was still valid until 3 months ago.
Before: ads shown to LAT-on people, but post-install performance was not measured and conversions were going to organics.
Now we need consent to get IDFA, and this is going to break.
[💎@04:40] Apple designed a new measurement framework called AdService, which was rolled out in March/April 2021, with an attribution that does not rely on IDFA anymore.
For the first time, we can measure the impact of people who were LAT On before.
[💎@05:30] Apple Search Ads’ attribution doesn’t depend on the IDFA but more importantly doesn’t depend on the ATT consent status, which means ASA needs to be handled slightly differently.
Other changes: search tab, China, etc.
In Apple’s vocabulary, we’re not tracking ASA users but measuring campaign performance. What Apple defines as tracking is mixing information from your company with the information from another company.
Users see ad impressions in the store. When a new user opens the app for the first time, the measurement framework pings an Apple Server to check if this new user comes from an ASA campaign. The payload can come back empty (-> user not from an ASA campaign) or come back with information (-> user from an ASA campaign).
[💎@10:30] The payload when a user is coming from one of your ASA campaigns is pretty granular: you can get keyword-level data and you can get retention and other post-install metrics.
[💎@11:06] If a user has opted-in to ATT (and you get it early enough), Apple gives you the exact timestamps of the click (advanced payload).
In the past, you could implement the iAd framework by yourself or use an MMP. It’s still the case with AdServices. So if you don’t have an MMP you can implement it yourself, but there’s no real advantage (or drawback).
People used to make estimations and predictions on LAT users. Some would bid on it, some wouldn’t.
Most people have kept their old campaign structure.
[💎@16:40] In most cases, ATT and the new ASA attribution don’t have major implications and you can keep your old campaign structure. There are still some inventory price differences.
At the campaign level it’s pretty much the same: splitting campaigns by brand, generic, competitors and discovery.
[💎@17:28] Thomas still likes to have ad groups (regardless of how they group keywords - meaning, LTV, etc.) “duplicated” so that he ends up with 3 ad groups targeting New Users LAT Off, Returning Users LAT off and All Users. It’s a more complex structure (and there is audience overlap), but the conversion and ARPU or retention performance can be entirely different depending on the targeting. It also helps if/when deciding from where to pull back spend.
Before, you would not capture about 30-40% of users if you were not targeting LAT On users.
[💎@23:51] Users you don’t target if you add an 18+ criteria: underaged users, some blocked users based on device (e.g. education) but more importantly users that deactivated “Apple Ads personalization”.
When it comes to managing your campaigns, Apple Ads personalization is the new LAT.
[💎@25:30] The assumption is that only a tiny portion of users opt-out of Apple Ads personalization (the setting is not easy to find), but nobody really knows: unlike with LAT, we can’t monitor it and it has not been shared.
[💎@26:40] The LAT On % was always going up. With ATT, users that were LAT On have been automatically opted-out of Ads personalization. The Ads personalization % might go down as new devices are introduced, because less people will be opted out “by default”.
[💎@27:28] Limiting targeting only to new and returning can lower down your potential reach, but if you’re strictly focused on performance (or get bad performance when targeting All Users) then it might make sense to not target All Users.
A/B testing tool coming soon. People using Creative Sets for A/B testing will switch to it. Creative Sets will be less used to it.
Custom Product Pages will show different listings depending on the campaign.
[💎@29:40] Custom Product Pages will allow you to customize the experience: show different listings depending on new or returning users, depending on user intent and USP.
CPP works based on the URL, so it not only requires iOS 15 but will also require changes for ASA. So probably not this year.
Not clear: will privacy thresholds apply for CPP from campaigns on FB/Google?
The logic of advertisers shifting FB/Google to ASA makes sense, but most people were already maxing out what they could get on search ads. If your ads are not relevant to the user search, you won’t make an impact.
Things have stayed pretty flat on ASA. You can’t move from $10k to $100k.
Google search on the web has much more long tail keywords, and that’s why it’s easier to scale.
The LAT cohort was always very peculiar with Apple Search Ads.
Content from 2017 (cf. The Mysterious LAT, and How to Tame It – with Thomas Petit) was still valid until 3 months ago.
Before: ads shown to LAT-on people, but post-install performance was not measured and conversions were going to organics.
Now we need consent to get IDFA, and this is going to break.
[💎@04:40] Apple designed a new measurement framework called AdService, which was rolled out in March/April 2021, with an attribution that does not rely on IDFA anymore.
For the first time, we can measure the impact of people who were LAT On before.
[💎@05:30] Apple Search Ads’ attribution doesn’t depend on the IDFA but more importantly doesn’t depend on the ATT consent status, which means ASA needs to be handled slightly differently.
Other changes: search tab, China, etc.
In Apple’s vocabulary, we’re not tracking ASA users but measuring campaign performance. What Apple defines as tracking is mixing information from your company with the information from another company.
Users see ad impressions in the store. When a new user opens the app for the first time, the measurement framework pings an Apple Server to check if this new user comes from an ASA campaign. The payload can come back empty (-> user not from an ASA campaign) or come back with information (-> user from an ASA campaign).
[💎@10:30] The payload when a user is coming from one of your ASA campaigns is pretty granular: you can get keyword-level data and you can get retention and other post-install metrics.
[💎@11:06] If a user has opted-in to ATT (and you get it early enough), Apple gives you the exact timestamps of the click (advanced payload).
In the past, you could implement the iAd framework by yourself or use an MMP. It’s still the case with AdServices. So if you don’t have an MMP you can implement it yourself, but there’s no real advantage (or drawback).
People used to make estimations and predictions on LAT users. Some would bid on it, some wouldn’t.
Most people have kept their old campaign structure.
[💎@16:40] In most cases, ATT and the new ASA attribution don’t have major implications and you can keep your old campaign structure. There are still some inventory price differences.
At the campaign level it’s pretty much the same: splitting campaigns by brand, generic, competitors and discovery.
[💎@17:28] Thomas still likes to have ad groups (regardless of how they group keywords - meaning, LTV, etc.) “duplicated” so that he ends up with 3 ad groups targeting New Users LAT Off, Returning Users LAT off and All Users. It’s a more complex structure (and there is audience overlap), but the conversion and ARPU or retention performance can be entirely different depending on the targeting. It also helps if/when deciding from where to pull back spend.
Before, you would not capture about 30-40% of users if you were not targeting LAT On users.
[💎@23:51] Users you don’t target if you add an 18+ criteria: underaged users, some blocked users based on device (e.g. education) but more importantly users that deactivated “Apple Ads personalization”.
When it comes to managing your campaigns, Apple Ads personalization is the new LAT.
[💎@25:30] The assumption is that only a tiny portion of users opt-out of Apple Ads personalization (the setting is not easy to find), but nobody really knows: unlike with LAT, we can’t monitor it and it has not been shared.
[💎@26:40] The LAT On % was always going up. With ATT, users that were LAT On have been automatically opted-out of Ads personalization. The Ads personalization % might go down as new devices are introduced, because less people will be opted out “by default”.
[💎@27:28] Limiting targeting only to new and returning can lower down your potential reach, but if you’re strictly focused on performance (or get bad performance when targeting All Users) then it might make sense to not target All Users.
A/B testing tool coming soon. People using Creative Sets for A/B testing will switch to it. Creative Sets will be less used to it.
Custom Product Pages will show different listings depending on the campaign.
[💎@29:40] Custom Product Pages will allow you to customize the experience: show different listings depending on new or returning users, depending on user intent and USP.
CPP works based on the URL, so it not only requires iOS 15 but will also require changes for ASA. So probably not this year.
Not clear: will privacy thresholds apply for CPP from campaigns on FB/Google?
The logic of advertisers shifting FB/Google to ASA makes sense, but most people were already maxing out what they could get on search ads. If your ads are not relevant to the user search, you won’t make an impact.
Things have stayed pretty flat on ASA. You can’t move from $10k to $100k.
Google search on the web has much more long tail keywords, and that’s why it’s easier to scale.
The LAT cohort was always very peculiar with Apple Search Ads.
Content from 2017 (cf. The Mysterious LAT, and How to Tame It – with Thomas Petit) was still valid until 3 months ago.
Before: ads shown to LAT-on people, but post-install performance was not measured and conversions were going to organics.
Now we need consent to get IDFA, and this is going to break.
[💎@04:40] Apple designed a new measurement framework called AdService, which was rolled out in March/April 2021, with an attribution that does not rely on IDFA anymore.
For the first time, we can measure the impact of people who were LAT On before.
[💎@05:30] Apple Search Ads’ attribution doesn’t depend on the IDFA but more importantly doesn’t depend on the ATT consent status, which means ASA needs to be handled slightly differently.
Other changes: search tab, China, etc.
In Apple’s vocabulary, we’re not tracking ASA users but measuring campaign performance. What Apple defines as tracking is mixing information from your company with the information from another company.
Users see ad impressions in the store. When a new user opens the app for the first time, the measurement framework pings an Apple Server to check if this new user comes from an ASA campaign. The payload can come back empty (-> user not from an ASA campaign) or come back with information (-> user from an ASA campaign).
[💎@10:30] The payload when a user is coming from one of your ASA campaigns is pretty granular: you can get keyword-level data and you can get retention and other post-install metrics.
[💎@11:06] If a user has opted-in to ATT (and you get it early enough), Apple gives you the exact timestamps of the click (advanced payload).
In the past, you could implement the iAd framework by yourself or use an MMP. It’s still the case with AdServices. So if you don’t have an MMP you can implement it yourself, but there’s no real advantage (or drawback).
People used to make estimations and predictions on LAT users. Some would bid on it, some wouldn’t.
Most people have kept their old campaign structure.
[💎@16:40] In most cases, ATT and the new ASA attribution don’t have major implications and you can keep your old campaign structure. There are still some inventory price differences.
At the campaign level it’s pretty much the same: splitting campaigns by brand, generic, competitors and discovery.
[💎@17:28] Thomas still likes to have ad groups (regardless of how they group keywords - meaning, LTV, etc.) “duplicated” so that he ends up with 3 ad groups targeting New Users LAT Off, Returning Users LAT off and All Users. It’s a more complex structure (and there is audience overlap), but the conversion and ARPU or retention performance can be entirely different depending on the targeting. It also helps if/when deciding from where to pull back spend.
Before, you would not capture about 30-40% of users if you were not targeting LAT On users.
[💎@23:51] Users you don’t target if you add an 18+ criteria: underaged users, some blocked users based on device (e.g. education) but more importantly users that deactivated “Apple Ads personalization”.
When it comes to managing your campaigns, Apple Ads personalization is the new LAT.
[💎@25:30] The assumption is that only a tiny portion of users opt-out of Apple Ads personalization (the setting is not easy to find), but nobody really knows: unlike with LAT, we can’t monitor it and it has not been shared.
[💎@26:40] The LAT On % was always going up. With ATT, users that were LAT On have been automatically opted-out of Ads personalization. The Ads personalization % might go down as new devices are introduced, because less people will be opted out “by default”.
[💎@27:28] Limiting targeting only to new and returning can lower down your potential reach, but if you’re strictly focused on performance (or get bad performance when targeting All Users) then it might make sense to not target All Users.
A/B testing tool coming soon. People using Creative Sets for A/B testing will switch to it. Creative Sets will be less used to it.
Custom Product Pages will show different listings depending on the campaign.
[💎@29:40] Custom Product Pages will allow you to customize the experience: show different listings depending on new or returning users, depending on user intent and USP.
CPP works based on the URL, so it not only requires iOS 15 but will also require changes for ASA. So probably not this year.
Not clear: will privacy thresholds apply for CPP from campaigns on FB/Google?
The logic of advertisers shifting FB/Google to ASA makes sense, but most people were already maxing out what they could get on search ads. If your ads are not relevant to the user search, you won’t make an impact.
Things have stayed pretty flat on ASA. You can’t move from $10k to $100k.
Google search on the web has much more long tail keywords, and that’s why it’s easier to scale.
