How IDFA Changes in iOS 14 Will Impact Retention Strategies

💎  x
9

Andy Carvell (Co-founder of Phiture - growth consulting firm) outlines the impact of privacy changes in iOS 14 on retention and engagement strategies, especially how the reduced reliance on IDFAs impacts tooling and targeting. He also talks about what to look forward to in iOS 14, and using successful strategies from his push notification playbook to increase IDFA opt-ins.

Type:
Podcast
Publication date:
September 24, 2020
Added to Growth Gems on:
October 11, 2020
💎 #
1

Using "scoped identifiers" from your analytics/CRM tools for internal tracking purposes will be ok. Example: Amplitude uses an Amplitude id for each user, Braze id, etc. You can also use a more "universal" identifier like email address, phone number, etc. if you collect that information.

05:08
💎 #
2

Regular "direct" deep linking for notifications and emails that get users to the right place in the app won't be affected.

08:45
💎 #
3

Deferred deep linking when you don't know who is going to click on that link and if they have the app installed or not will be less effective. You will need to rely on a probabilistic approach to get people to the right part of the app after the install or "match" with the initial link, which might be problematic in some use cases.

09:33
💎 #
4

In general segmenting users based on behavior is not going to change because you'll be doing your segmentation in your analytics/CRM platform and using an internal id within your ecosystem.

14:10
💎 #
5

Geofencing is not going to be possible without getting explicit exact location targeting permissions. Example: the Burger King campaign.

14:51
💎 #
6

Challenges related to CDPs depend on your current configuration: if you're leveraging CDPs to retarget people on programmatic ad networks then you'll be highly impacted.

18:58
💎 #
7

If you have a CDP you're probably well position to adapt to the IDFA changes, as CDPs have integrations for multiple types of use cases.

19:40
💎 #
8

It's early to say how it will impact app engagement, but iOS widgets are really cool and look a lot more usable. It's been possible on Android for a long time but iOS is doing it in a nice way.

22:19
💎 #
9

Tip for IDFA consent win-back: go back to loyal users (still in the app after 2 or 3 weeks but have opted out) and ask them to reconsider. If they say yes you can deep link to the settings.

28:26
This is premium content. Upgrade your plan in your profile to get access.
💎 #
1

Using "scoped identifiers" from your analytics/CRM tools for internal tracking purposes will be ok. Example: Amplitude uses an Amplitude id for each user, Braze id, etc. You can also use a more "universal" identifier like email address, phone number, etc. if you collect that information.

05:08
💎 #
2

Regular "direct" deep linking for notifications and emails that get users to the right place in the app won't be affected.

08:45
💎 #
3

Deferred deep linking when you don't know who is going to click on that link and if they have the app installed or not will be less effective. You will need to rely on a probabilistic approach to get people to the right part of the app after the install or "match" with the initial link, which might be problematic in some use cases.

09:33
💎 #
4

In general segmenting users based on behavior is not going to change because you'll be doing your segmentation in your analytics/CRM platform and using an internal id within your ecosystem.

14:10
💎 #
5

Geofencing is not going to be possible without getting explicit exact location targeting permissions. Example: the Burger King campaign.

14:51
💎 #
6

Challenges related to CDPs depend on your current configuration: if you're leveraging CDPs to retarget people on programmatic ad networks then you'll be highly impacted.

18:58
💎 #
7

If you have a CDP you're probably well position to adapt to the IDFA changes, as CDPs have integrations for multiple types of use cases.

19:40
💎 #
8

It's early to say how it will impact app engagement, but iOS widgets are really cool and look a lot more usable. It's been possible on Android for a long time but iOS is doing it in a nice way.

22:19
💎 #
9

Tip for IDFA consent win-back: go back to loyal users (still in the app after 2 or 3 weeks but have opted out) and ask them to reconsider. If they say yes you can deep link to the settings.

28:26
This content is available to Growth Gems premium members only.
Apply Now
💎 #
1

Using "scoped identifiers" from your analytics/CRM tools for internal tracking purposes will be ok. Example: Amplitude uses an Amplitude id for each user, Braze id, etc. You can also use a more "universal" identifier like email address, phone number, etc. if you collect that information.

05:08
💎 #
2

Regular "direct" deep linking for notifications and emails that get users to the right place in the app won't be affected.

08:45
💎 #
3

Deferred deep linking when you don't know who is going to click on that link and if they have the app installed or not will be less effective. You will need to rely on a probabilistic approach to get people to the right part of the app after the install or "match" with the initial link, which might be problematic in some use cases.

09:33
💎 #
4

In general segmenting users based on behavior is not going to change because you'll be doing your segmentation in your analytics/CRM platform and using an internal id within your ecosystem.

14:10
💎 #
5

Geofencing is not going to be possible without getting explicit exact location targeting permissions. Example: the Burger King campaign.

14:51
💎 #
6

Challenges related to CDPs depend on your current configuration: if you're leveraging CDPs to retarget people on programmatic ad networks then you'll be highly impacted.

18:58
💎 #
7

If you have a CDP you're probably well position to adapt to the IDFA changes, as CDPs have integrations for multiple types of use cases.

19:40
💎 #
8

It's early to say how it will impact app engagement, but iOS widgets are really cool and look a lot more usable. It's been possible on Android for a long time but iOS is doing it in a nice way.

22:19
💎 #
9

Tip for IDFA consent win-back: go back to loyal users (still in the app after 2 or 3 weeks but have opted out) and ask them to reconsider. If they say yes you can deep link to the settings.

28:26
This is premium content. Upgrade your plan in your profile to get access.
This content is available to Growth Gems premium members only.
Apply Now
You need to be logged in the community to see the discussion below. You can also head over to
this topic in the community

Impact of privacy/IDFA changes on retention strategies

At a high-level, it shouldn't be a huge impact: most of the tools that you are using use "scoped identifiers". The SDK in the mobile app assigns an id to a user the first time and is able to identify that user when it comes back.

However some internal tools might have been built using the IDFA, in which case the impact will be strong.


You can still use other kinds of ids

[💎 @05:08] There are other ways to create an id and assign it to a person for analytics purposes. You don't have to use the IDFA and "scoped identifiers" from your tools will be ok to use. Example: Amplitude uses an Amplitude id for each user, Braze id, etc. You can also use a more "universal" identifier like email address, phone number, etc. if you collect that information.

So the IDFA changes are not going to impact your retention strategies unless you or tools in your stack are currently using the IDFA (in which case it needs to be reworked).


You of course still have to be in compliance with privacy rules like GDPR, etc.

Deep linking

[💎 @08:45] Regular "direct" deep linking for notifications and email that get users to the right place in the app won't be affected.

[💎 @09:33] Deferred deep linking when you don't know who is going to click on that link and if they have the app installed or not will be less effective. You will need to rely on a probabilistic approach to get people to the right part of the app after the install or "match" with the initial link, which might be problematic in some use cases.

Examples:

  • "invite a friend", content shares, etc. → if you can't make the match then it's mostly fine to result to a generic experience.
  • "auto-logging" or "get $X for your invite" requires to have an exact match which will not be available


Are tools like Branch that plan on using fingerprinting too optimistic? If that's the case the efficiency of that kind of virality techniques will be impacted.


Segmenting users still possible

[💎 @14:10] In general segmenting users based on behavior is not going to change because you'll be doing your segmentation in your analytics/CRM platform and using an internal id within your ecosystem.

What's going away is to segment users across multiple apps.

[💎 @14:51] Geofencing is not going to be possible without getting explicit exact location targeting permissions.

  • Broad location-based like state/county or anything within several miles radius will be fine
  • Something like the Burger King campaign in a McDonald's radius will not be possible anymore

Using CDPs might bring challenges

CDPs work as central hubs. But they seem pretty well set up because they're usually thinking about things like privacy (e.g. mParticles is not collecting IDFA by default anymore).

You should still be able to use these CDPs, and they are saying that most integrations will not be affected.


[💎 @18:58] Challenges related to CDPs depend on your current configuration: if you're leveraging CDPs to retarget people on programmatic ad networks then you'll be highly impacted.


[💎 @19:40] If you have a CDP you're probably well position to adapt to the IDFA changes, as CDPs have integrations for multiple types of use cases.


The IDFA changes are going to push more account creations for users so you can build a unique identifier (like email) and then ask for consent for other activities like retargeting.


Widgets

[💎 @22:29] It's early to say how it will impact app engagement, but iOS widgets are really cool and look a lot more usable. It's been possible on Android for a long time but iOS is doing it in a nice way.


Andy also liked privacy features like the orange signal when an app is capturing video, etc.


Asking for IDFA permission/consent

For users that have explicitly given their consent you will still be able to do what you were doing before.

You can optimize your pre-dialog (soft prompt) or even win-back users later on in the journey, which is especially interesting for high value users.

From push notifications requests we know that we can always get some uplift from the baseline:

  • Ask for opt-in (system dialog, native prompt) at a happy moment - works to some extent
  • Use a pre-permission (in-app dialog) where you ask users before showing them the system dialog - widely successful and you should keep testing that messaging.
  • [💎 @28:26] Win-backs: go back to loyal users (still in the app after 2 or 3 weeks but have opted out) and ask them to reconsider. If they say yes you can deep link to the settings.

This is premium content. Upgrade your plan in your profile to get access.

Impact of privacy/IDFA changes on retention strategies

At a high-level, it shouldn't be a huge impact: most of the tools that you are using use "scoped identifiers". The SDK in the mobile app assigns an id to a user the first time and is able to identify that user when it comes back.

However some internal tools might have been built using the IDFA, in which case the impact will be strong.


You can still use other kinds of ids

[💎 @05:08] There are other ways to create an id and assign it to a person for analytics purposes. You don't have to use the IDFA and "scoped identifiers" from your tools will be ok to use. Example: Amplitude uses an Amplitude id for each user, Braze id, etc. You can also use a more "universal" identifier like email address, phone number, etc. if you collect that information.

So the IDFA changes are not going to impact your retention strategies unless you or tools in your stack are currently using the IDFA (in which case it needs to be reworked).


You of course still have to be in compliance with privacy rules like GDPR, etc.

Deep linking

[💎 @08:45] Regular "direct" deep linking for notifications and email that get users to the right place in the app won't be affected.

[💎 @09:33] Deferred deep linking when you don't know who is going to click on that link and if they have the app installed or not will be less effective. You will need to rely on a probabilistic approach to get people to the right part of the app after the install or "match" with the initial link, which might be problematic in some use cases.

Examples:

  • "invite a friend", content shares, etc. → if you can't make the match then it's mostly fine to result to a generic experience.
  • "auto-logging" or "get $X for your invite" requires to have an exact match which will not be available


Are tools like Branch that plan on using fingerprinting too optimistic? If that's the case the efficiency of that kind of virality techniques will be impacted.


Segmenting users still possible

[💎 @14:10] In general segmenting users based on behavior is not going to change because you'll be doing your segmentation in your analytics/CRM platform and using an internal id within your ecosystem.

What's going away is to segment users across multiple apps.

[💎 @14:51] Geofencing is not going to be possible without getting explicit exact location targeting permissions.

  • Broad location-based like state/county or anything within several miles radius will be fine
  • Something like the Burger King campaign in a McDonald's radius will not be possible anymore

Using CDPs might bring challenges

CDPs work as central hubs. But they seem pretty well set up because they're usually thinking about things like privacy (e.g. mParticles is not collecting IDFA by default anymore).

You should still be able to use these CDPs, and they are saying that most integrations will not be affected.


[💎 @18:58] Challenges related to CDPs depend on your current configuration: if you're leveraging CDPs to retarget people on programmatic ad networks then you'll be highly impacted.


[💎 @19:40] If you have a CDP you're probably well position to adapt to the IDFA changes, as CDPs have integrations for multiple types of use cases.


The IDFA changes are going to push more account creations for users so you can build a unique identifier (like email) and then ask for consent for other activities like retargeting.


Widgets

[💎 @22:29] It's early to say how it will impact app engagement, but iOS widgets are really cool and look a lot more usable. It's been possible on Android for a long time but iOS is doing it in a nice way.


Andy also liked privacy features like the orange signal when an app is capturing video, etc.


Asking for IDFA permission/consent

For users that have explicitly given their consent you will still be able to do what you were doing before.

You can optimize your pre-dialog (soft prompt) or even win-back users later on in the journey, which is especially interesting for high value users.

From push notifications requests we know that we can always get some uplift from the baseline:

  • Ask for opt-in (system dialog, native prompt) at a happy moment - works to some extent
  • Use a pre-permission (in-app dialog) where you ask users before showing them the system dialog - widely successful and you should keep testing that messaging.
  • [💎 @28:26] Win-backs: go back to loyal users (still in the app after 2 or 3 weeks but have opted out) and ask them to reconsider. If they say yes you can deep link to the settings.

This content is available to Growth Gems premium members only.
Apply Now

Impact of privacy/IDFA changes on retention strategies

At a high-level, it shouldn't be a huge impact: most of the tools that you are using use "scoped identifiers". The SDK in the mobile app assigns an id to a user the first time and is able to identify that user when it comes back.

However some internal tools might have been built using the IDFA, in which case the impact will be strong.


You can still use other kinds of ids

[💎 @05:08] There are other ways to create an id and assign it to a person for analytics purposes. You don't have to use the IDFA and "scoped identifiers" from your tools will be ok to use. Example: Amplitude uses an Amplitude id for each user, Braze id, etc. You can also use a more "universal" identifier like email address, phone number, etc. if you collect that information.

So the IDFA changes are not going to impact your retention strategies unless you or tools in your stack are currently using the IDFA (in which case it needs to be reworked).


You of course still have to be in compliance with privacy rules like GDPR, etc.

Deep linking

[💎 @08:45] Regular "direct" deep linking for notifications and email that get users to the right place in the app won't be affected.

[💎 @09:33] Deferred deep linking when you don't know who is going to click on that link and if they have the app installed or not will be less effective. You will need to rely on a probabilistic approach to get people to the right part of the app after the install or "match" with the initial link, which might be problematic in some use cases.

Examples:

  • "invite a friend", content shares, etc. → if you can't make the match then it's mostly fine to result to a generic experience.
  • "auto-logging" or "get $X for your invite" requires to have an exact match which will not be available


Are tools like Branch that plan on using fingerprinting too optimistic? If that's the case the efficiency of that kind of virality techniques will be impacted.


Segmenting users still possible

[💎 @14:10] In general segmenting users based on behavior is not going to change because you'll be doing your segmentation in your analytics/CRM platform and using an internal id within your ecosystem.

What's going away is to segment users across multiple apps.

[💎 @14:51] Geofencing is not going to be possible without getting explicit exact location targeting permissions.

  • Broad location-based like state/county or anything within several miles radius will be fine
  • Something like the Burger King campaign in a McDonald's radius will not be possible anymore

Using CDPs might bring challenges

CDPs work as central hubs. But they seem pretty well set up because they're usually thinking about things like privacy (e.g. mParticles is not collecting IDFA by default anymore).

You should still be able to use these CDPs, and they are saying that most integrations will not be affected.


[💎 @18:58] Challenges related to CDPs depend on your current configuration: if you're leveraging CDPs to retarget people on programmatic ad networks then you'll be highly impacted.


[💎 @19:40] If you have a CDP you're probably well position to adapt to the IDFA changes, as CDPs have integrations for multiple types of use cases.


The IDFA changes are going to push more account creations for users so you can build a unique identifier (like email) and then ask for consent for other activities like retargeting.


Widgets

[💎 @22:29] It's early to say how it will impact app engagement, but iOS widgets are really cool and look a lot more usable. It's been possible on Android for a long time but iOS is doing it in a nice way.


Andy also liked privacy features like the orange signal when an app is capturing video, etc.


Asking for IDFA permission/consent

For users that have explicitly given their consent you will still be able to do what you were doing before.

You can optimize your pre-dialog (soft prompt) or even win-back users later on in the journey, which is especially interesting for high value users.

From push notifications requests we know that we can always get some uplift from the baseline:

  • Ask for opt-in (system dialog, native prompt) at a happy moment - works to some extent
  • Use a pre-permission (in-app dialog) where you ask users before showing them the system dialog - widely successful and you should keep testing that messaging.
  • [💎 @28:26] Win-backs: go back to loyal users (still in the app after 2 or 3 weeks but have opted out) and ask them to reconsider. If they say yes you can deep link to the settings.