Table of Contents

  1. Executive Summary
  2. What changed
  3. Observed Impact
  4. What it means
  5. Media partner & platform changes and recommendations

Executive Summary

  1.  70% of iOS users worldwide and 75% of users in the US have adopted version 14.5 or
  2. On average, 45% of iOS 14.5+ users are opting in to being tracked by apps. The opt-in rate varies by type of app. Social media apps have the lowest opt-in rate at 34%
    worldwide and 28% in the US
  3. Considering adoption and opt-in rates, approximately 13% of mobile device users
    worldwide and 30% of mobile device users in the US are untrackable
  4. Addressable audience sizes such as retargeting pools have decreased by 10% – 30% on average
  5. With future releases of iOS and anticipated 3rd party cookie deprecation, these impacts will grow over time. User-level audience and measurement approaches will fade away and modeled solutions will take their place.

What Changed:

In April 2021 Apple released iOS 14.5 which introduced App Tracking Transparency (ATT). This allows users to control which apps can track their activity across other companies’ apps and websites for advertising or sharing with data brokers. This permission is enabled at the app level.

What’s coming next

iOS 15 will be released this fall, and will come with even more features designed to enable user control of their privacy. It will include:

  • App Privacy report – Users will see how apps are using the permissions granted to track them, which third-party domains tracking is shared with, and how recently the tracking was shared.
  • Intelligent Tracking Prevention – obfuscates device ip addresses
  • Mail Privacy Protection – obfuscates user email addresses

These features are expected to accelerate the impacts we’re seeing from 14.5. Identity graphs that use cookie IDs, MAIDS, ip addresses, and email addresses will lose scale, making it difficult to understand identity and reach addressable audiences across channels and devices.

Observed Impacts

As of early August, adoption of iOS 14.5 has reached 70% worldwide. Adoption in the United States is 75%. For iOS users that have upgraded to 14.5, approximately 45% are allowing apps to track their activity on average. This varies from app to app.

Globally, iOS users make up 1B out of 3.8B mobile device users. With 70% adoption rate of iOS 14.5 and opt-in rates of 45%, about 388M or 10% of users are unidentifiable on average across apps. In the US, 65M or 25% of users are unidentifiable. It’s a bit worse for social media apps.

Users are allowing social media apps to track them at lower rates than other app categories. This means that globally, 508M or 13% of users are unidentifiable. In the US, 77M or 30% of users are unidentifiable.

What it Means

When advertisements are served in app, the app records an impression. If the user has opted in to tracking, the user’s mobile ad device ID (MAID) is recorded with the impression. If the user goes on to take other actions – tap an ad, visit another website, make a purchase – all of those events are recorded and associated with the MAID, enabling ad tech platforms to attribute user actions with advertising exposure. When a user asks an app not to track them, all of those events are still recorded, but the MAID is not, breaking the association of all those actions with a specific user. The absence of MAIDs has implications in two main areas: 1) audience targeting and 2) measurement.

Audience Targeting

It is important to note that this has been playing out for some time now. Apple Safari and Firefox both already block third-party cookies by default, so addressable audiences in the browser were already limited to about 60% of users in the US. Now addressable audiences in app are facing the same impact.

Across various brand and campaigns, we’re seeing a range of 10% – 30% reduction in size of audiences
such as:
● Retargeting pools
● Audiences defined with 3rd party data behavioral data
● Audiences defined with mobile location data


Diminished visibility of the user journey. For iOS users who have opted out of being tracked, the advertising touch points through their path cannot be connected. This means that frequency can’t be measured, and multi-touch attribution models that use user-level data will not be able to connect media touchpoints to actions for these audiences.

Loss of attribution. For iOS users who have opted out of being tracked, since their MAID will be stripped and not passed during ad serving and subsequent actions on a brand’s website, the media exposure cannot be connected to the activity. This doesn’t mean visits, conversions, or other actions are not happening, it just means that the advertising the user engaged with won’t get credit (attribution).

Platform Optimization Algorithms. Advertising campaigns running programmatically across social platforms and DSPs that are leveraging the platform’s optimization algorithm to optimize to conversions or other events are at risk. The nature of these algorithms is to analyze converting users and find similar users who are likely to convert to serve ads to. For opted-out users, this analysis and therefore ability to find similar users is disrupted.


SKadNetwork API is how Apple will be allowing app marketers to measure the results of their marketing in a privacy-friendly way. It allows marketers to measure conversion rates of app install campaigns on mobile devices for iOS apps without passing along user-level data. It also provides conversion data to advertisers without revealing personally identifiable information (PII) or device identifiers. It appends
attribution parameters upon ad click, so when the app is installed and opened for the first time, it sends an install postback to the ad network including information such as campaign ID.

Under SKAdNetwork, all data will be anonymous. Outside of that, the following data will be visible:
● Publisher ID (source), Campaign ID, Conversion Value
● Campaign – Clicks, Spend, etc.
● Info from Opt-In Users – IDFA
● App Data – Login, Event Interaction
● Install data (first time or redownload) w/verification

Advertisers should be prepared to no longer have access to the following information under Apple’s SKadNetwork including:

    • Personally Identifiable Information (PII)
    • User-level campaign and channel attribution data
    • Most Post-install Metrics
    • Lookback Windows
    • Only 100 Campaign IDs Per Media Source Allowed
    • No Frequency Capping

It is recommended that app owners configure Apple’s SKAdNetwork. Advertisers should ensure their mobile measurement partner (MMP) is enabled with Apple’s SkAdNetwork API. They should also update software development kits (SDKs) to support the SkAdNetwork option.

Media Partners & Platforms Responses and Recommendations

What has changed Recommendations from
In response to iOS 14.5, Facebook rolled out several changes related to tracking and measurement. Facebook also provided key actions for Web vs App advertisers. Some of the major changes included:

  • Default attribution windows changed to 7 day
    click (1 day view still available but with limited data from iOS14.5+ users)
  • New Aggregated Event Manager Tool to use to assign 8 priority pixel events
  • Advertisers were asked to verify domains
  • Breakdown reporting for conversion events has been deprecated
  • some conversion data will be delayed in
  • Conversion events will now be applied to the
    conversion day/time vs last click/view.
  • Facebook was transparent that audience sizes
    will decrease.
    1. Leverage broader targeting
      where possible
    2. Leverage look-a-like
    3. Do not break out
      campaigns by device or OS

Actions to Take for Ads Ecosystem Changes

iOS 14 Vertical Guidance and Recommendations

LinkedIn has not asked advertisers to make any
significant changes as a result of iOS enforcement. They outline which advertising features are impacted which include:

  • Retargeting audience sizes becoming smaller
  • Member interests/traits may not be used with a contact list (whether uploaded directly
    or through a marketing platform) or a website retargeting audience when selecting targeting facets for your campaign
Preparing for Apple iOS 14
App Tracking Transparency
Changes on LinkedIn Ad
Campaigns | Marketing
Solutions Help
Pinterest continues to use their on-platform signals to make continual improvements to their advertising solutions. These first-party signals are unaffected by Apple’s iOS 14.5 changes. For  audience solutions that rely on off-platform signals, Pinterest expects there to be a reduction in audience size from Pinners who opt-out. Solutions impacted include: Customer
Lists, Dynamic RT and Visitor RT.
Snapchat made changes to reporting:

  • Conversion reporting dimensions deprecated: Region, DMA, Device Make, Snap Lifestyle Category, and Product ID
  • Lookback windows: 28-day view deprecated
  • Opt-out users will automatically be excluded from certain targeting types, which will result in smaller audience pools across pixel custom audiences, Snap Audience Match and Retargeting pools for Dynamic Ads.
  1. Build look-a-likes of Pixel Custom Audiences and Snap Audience Match audiences to regain scale
  2. Prioritize prospecting
    spend in Dynamic Ads
  3. Consider uploading your
    own first party Custom
    Audience lists using
    hashed email and/or
    hashed phone if allowed by your company’s ATT
  4. Create Snap Engagement
    Audiences from
    past/current campaigns
  5. Configure hashed email
    and/or hashed phone
    number in your Snap Pixel
    for performance
    improvements Web:
    Preparing for iOS 14:
    Mobile Web Advertising

App: Preparing for iOS 14: Mobile App Advertising

Twitter will not attribute off-platform web
conversions for people on iOS 14.5+ who have
opted out of tracking via the App Tracking
Transparency prompt. This decrease in attributable web conversions may impact campaign performance. As attributable results decline, it will become more difficult for ad systems to learn and optimize ad delivery. With that, they are movingquickly to explore privacy- preserving measurement solutions that would help report on audiences who have opted out of tracking.
  • Leverage look-a-like audiences of your custom audiences
  • Leverage the “expand your
    audience” option & set
    expansion to “Broad”
  • Switch to Auto Bid as
    desired. Auto Bid offers
    the best chance for scale
    at the best price.
  • Switch to Auto Bid as
    desired. Auto Bid offers
    the best chance for scale
    at the best price.
  • Be cautious of layering on
    Demographic & Device
    targeting as this will
    reduce scale
  • Add Targeting Features
    (Keywords, Interests, etc.)
    to expand the audience
  • Review App Activity
    Combination Audience
    lookback windows
  • Where appropriate,
    consider extending lookback windows to maintain audience sizes

Twitter and iOS 14 Resource

How Twitter is preparing for
Apple’s iOS 14 IDFA updates

For advertisers tracking app installs, TikTok will
support measurement and optimization based on SKAdNetwork (SKAN) API, Apple’s attribution solution. TikTok is currently engaged with all of our Mobile Measurement Partners (MMPs) who have either released or will release an updated version of their SDK which supports SKAdNetwork API., TikTok will support SKAdNetwork reporting for dedicated
iOS 14 campaigns in both Business Manager and
Events Manager.
Recommendations vary based
on App vs Mobile. Please refer
to the resource link below for
more details. Dedicated
campaigns are required to
reach iOS 14.5+ users.
Publisher Resource:Business Help Center
The Trade Desk is invested in a mix of short and long
term solutions that will help drive targeting and
measurement scale in iOS. Examples of these
solutions include:
Koa for ID-less environment: This feature will use our Koa AI engine to model an impression profile and target similar users in environments where there are no identifiers.
Probabilistic frequency capping to offer some
frequency control where there are no user IDs
Display ads are auction-based and depend on user data, personal interests, age, and gender. Now, some of these behaviors and characteristics are less visible, and targeting is restricted to the current page the user is viewing

DV360 Targeting Solutions: Google audiences canhelp you find relevant users where no cookie or ID’s exist by combining user and contextual signals with machine learning. Targeting expansion is now available across affinity, in-market, and installed appaudiences in addition to look-a-like audiences. Targeting expansion is automatically enabled when targeting eligible audiences.

If running in-app ads or app
install campaigns, take a
campaign consolidation
approach vs. segmenting
campaigns. Group audiences
and line items together and set
a limit to 8 campaigns per app.
A new URL parameter has been introduced to help comply with Apple’s policies and help measure the results of ads on iOS. This new parameter will help attribute conversions back to ad campaigns and works with conversion modeling to give a more accurate measurement on iOS.

To support this new parameter, ensure that you have the global site tag (gTag.js and/or Google Tag Manager) properly implemented on your site.

Remarketing Lists for Search Ads (RLSA) will
experience impacts because traffic coming from
iOS14 users will not be identified as before. Instead, they will be grouped by interests, making tracking less accurate.

Campaign Manager expanded the conversion modeling framework to model both click and
view-through conversions that are lost due to iOS changes for users in browser. Google analyzes unattributed conversions and uses observable data to train data and create attribution modeling attribute conversions to clicks and impressions that were not
eligible for attribution due to iOS changes.Google now has a tag parameter (“ltd=”) that provides a way for third parties to specify user consent on clicks. When “ltd=1” is present, user IDs and click IDs are disabled for the click event.

This makes it possible to report on how many clicks were consented and unconsented using the tagparameter and a new metric, “Cookie Unconsented Clicks”. To get the volume of consented clicks, you can build a report with “Clicks” and “Cookie Unconsented Clicks” and subtract unconsented clicks from total clicks (“Clicks” – “Cookie Unconsented Clicks” = consented clicks).

Supporting partners through
the rollout of Apple’s new ATT policies – Google Ads Help
oogle-ads/answer/10417364FAQs about iOS 14 changes –
Google Ads Help