Updated: February 2nd
Along with the changes implemented in app advertising and tracking, Apple is looking to implement a similar mechanism that will apply to web browsers.
This new standard called Private Click Measurement (PCM) was released by Apple during 2019, as part of the Privacy by default design for Safari.
“Safari now includes the ability to offer Private Click Measurement, an innovative way of doing ad click measurement that prevents cross-site tracking but still enables advertisers to measure the effectiveness of web campaigns.”
Safari Privacy Overview, November 2019
This protocol is independent from Apple’s ATT (AppTrackingTransparency) framework, which applies to apps only. We do however expect both to be somewhat aligned.
The underlying assumption behind this is that web tracking should remain anonymous (i.e. actions can’t be tied back to a specific individual), while ad platforms (e.g. Google Ads) are still able to report on key metrics (e.g. clicks, conversions etc.).
This protocol will be the default setting for Safari users, who will actively have to de-select it. Unlike app tracking, no prompt will show for each website as this is a browser setting.
Down the line, this standard will also apply to any web browser installed on an Apple device, so goes beyond users that use the default Safari browser. Regarding in-app browsers (Web View), e.g. Facebook’s browser, at this point in time, it won’t be supported, which might indicate stricter enforcement will be used (i.e. complete blocking of tracking).
PCM will also support App to Web measurement (introduced in the iOS 14.5 Beta). This means that clicks from the Facebook app (not just the Facebook website) will also be attributed correctly.
Apple has also submitted this to be considered as a standard by the web community (W3C), and its draft is currently in review by the community (as of January 2021).
The PCM standard has been released to Beta in iOS 14.5, and is available for testing. General availability date is still unknown but Apple mentioned “early spring”.
What does PCM mean in practice?
When a user clicks on a certain ad, he is often then sent to the advertiser’s website (destination) to complete some action (e.g. complete a purchase or fill in a lead form).
When the user is on the ad platform, e.g Facebook, he is identified by the platform. Once he clicks out, additional engagement done on the destination site needs to be tracked so that the advertiser can measure the effectiveness of his advertising. In absence of this data, the only trackable metric is Clicks.
Previously, ad platforms have used methods such as 3rd party cookies to identify the users off their ad platforms. This for example enabled them to attribute conversions back to the ad click.
Additionally, these 3rd party cookies could follow users around the web regardless of engaging with ads. For example, they could identify a user visiting website A that sells used cars, and flag him as “in-market” for buying a used car, so that ads for that topic will show up to him.
With 3rd party cookies being limited, ad platforms have moved to a different method called link decoration. This is a common practice also used by tech giants Google and Facebook (i.e. fbclid and gclid)
In this method, any outbound link from the ad platform (source) to the advertiser’s site (destination) will be appended a URL parameter that identifies the user and links their actions to the ad activity.
Apple’s new standard suggests the following:
Each click can be attributed back only to the campaign that drove this click, i.e. no ad level or user level reporting.
To achieve this, PCM implements two key mechanisms: Aggregated and Delayed reporting
For Aggregation, PCM limits the reporting to the campaign level only, i.e. this conversion action is attributed back to the campaign ID that drove the click. The number of campaigns to 256 (Feb 1st update), so that advertisers won’t abuse the campaign data for personal tracking.
Additionally, PCM also limits the conversion actions that can be reported to 16 (Feb 1st update). These conversions can’t carry any data, e.g. the names of the items purchased or their value.
Delayed reporting is simply a random 24-48 reporting window so that the advertising platform can’t reverse engineer the data to identify personally the converted users.
What about cookies?
Browser cookies will continue to live through this change. In alignment with ITP (Apple’s standard for tracking), tracking cookies are capped for 7 days. This means that attribution reporting for PCM will be limited to 7 days only, per the specific device and browser (identical to current Google Analytics limits in Safari).
How will PCM impact advertisers?
A glimpse into what PCM will look like is already seen on Facebook’s changes to their Ad Manager that have rolled out on January 2021. While Facebook is rolling out a different standard, Aggregated Event Measurement (AEM), in hope of making it the industry’s standard, the implications are rather similar.
Conversion tracking
With PCM applied, conversions will report with a significant lag, making it harder to optimize campaigns on the fly. This will be an issue both for human and algorithmic optimization (done by the ad platforms, e.g. Google’s Smart Campaigns).
Additionally, due to the modeled nature of the conversion reporting, attribution windows will drop significantly, with Facebook setting the standard to be 1 day view/7 day click.
What can you do?
This is a major issue. The lack of conversion data will impact both Facebook’s optimization algorithms and the manual optimization done by campaign managers.
The only “viable” metric that will be measured consistently is Link clicks (not even Landing Page Views), which is often a poor metric to optimize for.
In absence of quality data, a buffer metric should be used to gauge the quality of clicks coming from a certain campaign or ad set. This can be achieved using upper funnel actions such as Add to Cart or Fixel High, which have a high correlation with conversions.
Creative optimization
With reporting limited to the campaign level only, and extrapolated to the ad set (i.e. Modeled Conversions), optimizing a specific ad and its creatives will become impossible.
What can you do?
One option we’ve thought about is to use a single creative per ad set (similar to the Google Ads controversial practice of Single Keyword Ad Groups aka SKAGs). However, this can also be an issue with platforms such as Facebook limiting the amount of ad sets in an account.
Another option is optimizing creatives by clicks and engagements (on the ad platform), though this too doesn’t always correlate with actual conversions.
Audiences Breakdowns
With the linkage between a particular user and conversion now removed, ad platforms will not be able to provide granular breakdowns of user attributes that led to conversion. This for example will limit reporting on device types, location, interests and other key attributes used for segmentation and optimization.
What can you do?
In the ad platforms, you can breakdown ad sets by distinct audiences (as you probably should regardless), to a greater granularity than before. Despite some potential negative impact (i.e. reach and overlap), this can help when drilling down for additional insights in a certain campaign.
Make sure you prioritize key attributes, such Device Type, Operating System and Location.
Remarketing & Dynamic Ads
With linkage removed and data limited, the availability of website remarketing audiences is expected to drop significantly. The limit on data collection will also make dynamic ads reach far less users than previously.
What can you do?
This issue is yet to be resolved. We’re still learning the impacts of this too and hope to have some insight soon, so stay tuned.
