Apple’s two latest updates to Safari’s anti-tracking feature, Intelligent Tracking Prevention, have created a headache for anyone looking to attribute an ad clicked on one site to a product purchase or other conversion that occurs on the advertiser’s site. Now Apple is offering an olive branch of sorts: an ad attribution tool dubbed Privacy Preserving Ad Click Attribution.

WTF is Privacy Preserving Ad Click Attribution?
A terrible name for Apple’s ad attribution tool, so let’s just call it Apple’s ad attribution tool. Apple recognizes that companies need to be able to tie ads to business results, like someone purchasing a product on an advertiser’s site after clicking on an ad on a publisher’s site. But it’s not wild about companies using ad attribution methods to track people around the web. So it has come up with its own ad attribution tool that tries to strike a balance between the two by recording when someone clicks on an ad and then converts on the advertiser’s site but without letting companies trace that conversion to a particular person.

Why did Apple need to come up with an ad attribution tool?
Because Apple has made it hard for existing ad attribution tools to work in Safari. Since it introduced ITP in 2017, Apple has said that it doesn’t want to mess with ad attribution. But a first-party cookie workaround that companies like Facebook and Google have implemented to attribute ads despite Safari’s third-party cookie limits could be used for cross-site tracking. So earlier this year, Apple updated ITP to delete the first-party cookies that companies use to measure attribution after seven days of a person clicking on an ad. Then last month Apple cut that window to only 24 hours.

How does Apple’s ad attribution tool work?
Let’s say a person clicks an ad on Pretend Publisher’s site that directs them to Imaginary Advertiser’s site. When the person clicks on the ad, the browser will make a note of that event by recording information including the domain of the publisher’s site where the ad click occurred, the domain of the advertiser’s site that the click directed them to and a numerical identifier for the campaign corresponding to the ad. However, instead of sharing that information with Pretend Publisher, Imaginary Advertiser or any ad tech intermediaries, that ad click information will be stored in the browser and locked to a person’s computer, phone or tablet.

If companies can’t record the ad click information, how would they be able to measure if someone actually purchased a product from the advertiser’s site?
They have to rely on the browser to make the connection. Sticking with the example above, when that person buys a product from Imaginary Advertiser’s site, a tracking pixel running on the advertiser’s site will send a message addressed to a specific destination on Pretend Publisher’s site that is only supposed to be called when a conversion has occurred. The browser will see that the message is sent to that specific destination and take it as a sign that a conversion may have occurred. The browser will then check the ad click information that it has stored on the browser. If it has an entry that includes Pretend Publisher and Imaginary Advertiser, then it will make a note that a conversion has occurred and report the conversion to the companies…eventually.

Apple’s ad attribution tool won’t report conversions when they happen?
No. Apple will delay when it reports the conversion to sometime between 24 and 48 hours after the conversion happened; the reporting delay will be longer if a device is not connected to the internet when the conversion is scheduled to be reported. Apple chose to do this because, if it reported conversions in real time, companies might be able to infer who the person was that converted. But there is a silver lining to the reporting delay.

What could possibly be good about having to wait for one to two days for attribution measurement?
Well, Apple will have the browser delete the ad click information once the conversion report is sent. That would be problematic if multiple conversions are being tracked, such as one for when someone adds a product to their shopping cart and another for when they purchase the product. Without the 24- to 48-hour delay, the “add to shopping cart” conversion will be measured, but then there would be no way to measure if the person actually checked out and purchased the product. With the 24- to 48-hour delay, if a person adds a product to a shopping cart, steps away from their computer and buys the product a couple of hours later, then the purchase will still be measured.

Will multiple conversion events be recorded for a given ad click?
No, only one conversion will be reported. Companies can set multiple types of conversions to be measured, but they will need to set a priority for each conversion type. Then, when the conversion report is scheduled to be sent, the browser will report whichever conversion event had the highest priority.

Who will receive the conversion reports?
The publisher that carried the ad that the person initially clicked on. That’s it, which could be an issue for advertisers who want a third party to verify the attribution measurement. “Opaque third-parties should not receive ad click attribution reports,” wrote John Wilander, the Apple engineer behind ITP, in a blog post announcing the ad attribution tool.

Will this reinforce last-click attribution?
Not necessarily. Sites will be credited for driving a conversion even if a person didn’t convert immediately after clicking on an ad. Apple will store ad clicks in the browser for seven days. That way, if someone clicks an ad on Monday, they’ll have through the following Sunday to return to the advertiser’s site and convert and have the ad that they clicked on be credited for the conversion.

What about attributing conversions when someone only saw an ad but didn’t click on it?
Apple’s ad attribution tool doesn’t take view-through attribution into account. On a message board for developers and others to raise issues with Apple’s ad attribution tool, Wilander wrote: “We are interested in supporting ad view attribution in a privacy-friendly way too. But this spec is for ad clicks and we are not ready to tackle ad views yet.”

Will this affect ad targeting?
It looks that way, at least when it comes to targeting ads to people considered more likely to convert. Because the ad click information will be stored in the browser and not shared with companies, platforms won’t know if a particular person actually converted after they clicked on an ad, only that they clicked on the ad. As a result, they won’t be able to make a note of whether a particular person is prone to click on ads and purchase the advertised products.

Is this going to make retargeting more annoying?
You mean, if someone clicks an ad, are they probably going to be retargeted with more follow-up ads because ad sellers will be pressured to get someone to convert as quickly as possible in order to get credit for the conversion within Apple’s attribution window? Yeah, probably.

When is this all happening?
Unclear. Companies can currently test Apple’s ad attribution tool through the developer-centric version of Safari, Safari Technology Preview, and Apple is in the process of proposing it as an official web standard, which would make it easier for other browsers to adopt it. But if and when browsers, including Safari, would officially implement it is up in the air.

  • LinkedIn Icon