Kochava launches deterministic attribution for mobile web
The attribution firm is now matching its mobile web cookies to its mobile device IDs, to definitively track the impact of mobile web ads on app installs.
Measurement firm Kochava is out Friday with a method of deterministic attribution for mobile web advertising.
Essentially, GM Grant Cohen said in an interview, Kochava can now link its mobile web cookies with its mobile app device IDs, to definitively track a given user’s action in an ad inside a mobile browser that leads to an app install or other app-based activity.
“Stay in our lanes.” Previously, he said, the company’s attribution “had to stay in our lanes.”
That is, user activity inside a mobile app can be tracked via a mobile device ID — that is, through the device’s Apple IDFA (Identity for Advertisers) or Android GAID (Google Advertising ID). They are not usually available for tracking ads in mobile web, and they are considered “deterministic,” because the ID lets the device be specifically and definitively identified.
So, in that “lane,” Kochava can track in-app ad activity.
On the mobile web, a Kochava-tracked ad can generate a cookie from Kochava’s own domain. Cookies have a limitation, in that they can only be read by the domains that drop them. When a user clicks on a Kochava-tracked ad in a mobile web browser, the ad’s link generates a hidden kochava.com/something webpage through a redirect, often in background.
This Kochava domain drops a first-party cookie into the web browser, which can then be read by Kochava. The clicked link then continues to its ad-defined destination, such as another web page or an app.
Kochava has thus been able to track the user’s interaction with an ad in web browsers, or with an ad inside apps — the separate “lanes.”
Crossing the “lanes.” The problem, Cohen says, is when there’s an ad from a mobile web page to an app, such as to an app store for an app install. That is, when the “lanes” are crossed.
For instance, a mobile web page ad for Hilton Hotels that leads to a trip planner inside a Hilton app, but the user doesn’t yet have the Hilton app.
If the Hilton app has to be downloaded from an app store, Kochava loses sight of the user once they hit either the Apple App Store or the Google Play Store. In the former case, Cohen says, Apple provides virtually no referral link info, and, in the latter, Google only provides the top domain, such as “cnn.com.”
It’s difficult to definitively determine from just a domain if an ad delivered to cnn.com/sports for a given user led to this app install.
To make a probabilistic or likely match, Kochava creates a “digital fingerprint” consisting of several unique parameters that are available from the web ad, including IP address, user agent (which includes device model but not ID), and a time stamp.
But this does not definitively determine that User A saw this ad on the mobile web and then downloaded the Hilton app, because it doesn’t definitively connect the web user with the app user.
Or when it’s the other way: from an ad in an app, to a web page. In that case, Kochava had the mobile device ID for the in-app ad, but no way match the user in the browser with the user in the app — even if a first-party cookie had previously been dropped in the browser and was still active.
Matching cookie with device ID. What is new, Cohen said, is that Kochava is now matching that cookie with the mobile device ID so that, the next time it sees this user’s mobile browser, it can assign a definitive mobile device ID.
In other words, the web activity can now effectively be identified by a mobile device ID.
“Previously,” he said, “there was no way to attribute mobile web ads to app installs, deterministically.”
There are, of course, a few caveats.
This match can only occur if the user has clicked on an ad in a browser and has gone to install an app or later clicked on an ad in an app, so Kochava has dropped a first-party cookie, has the device ID and can connect the two.
The tracking between cookie-and-device-ID only happens the second time Kochava sees this user, since it is busily dropping a cookie and lining up the match the first time this occurs.
And, finally, the first-party mobile cookie is generally good for only 30 days, in most cases, after which it evaporates and the user once again can’t be seen on the mobile web.
Cohen said Kochava is now looking at matching its first-party cookie/mobile device ID with first-party cookies from other providers, which might be “younger” in the 30-day lifespan and thus might allow a match to continue beyond the 30 days.
I asked him why, if Kochava already had the mobile device ID for in-app ads, and already generated a first-party cookie, it took so long to match them.
“Good question,” he said, adding that it required “a feat of engineering to make it work” as well as an attribution provider with “massive scale,” such as Kochava.
“I’m convinced we’ll get copy-catted,” he said.
Why you should care. Tracking users on the mobile web has become a major problem for marketers, given the objections from Apple and Android against third-party cookies.
Matching a first-party cookie to a device ID ties a transient browser tracker to a persistent device identifier and, at least within the 30-day window, can provide a definitive way to match actions against the delivery of specific ads.
As Cohen suggests, expect to see more attribution firms and ad trackers tying transient mobile web cookies to stable mobile device IDs.
Opinions expressed in this article are those of the guest author and not necessarily MarTech Today. Staff authors are listed here.