Tech vendors love touting their proprietary ID solutions.
But the alternative ID landscape is incredibly fragmented, with about 60 proprietary ID solutions on the market that can be added to a publisher’s Prebid User ID Module.
So how are publishers – especially long-tail publishers that tend to be strapped for tech resources – supposed to pick the ID solutions that work best for them?
How alternative IDs work
There are two broad classes of ID solutions: deterministic and probabilistic.
Deterministic IDs rely on signals like email addresses and device IDs that can be directly tied to a specific user to create a match between two different data sets.
Probabilistic identifiers draw on a variety of signals that are more anonymized, like hashed IP addresses, browser type, geolocation and even things like what fonts a user has installed in their browser.
ID solutions also differ in whether they use client-side or server-side data for audience matching, said Patrick McCann, SVP of research at CafeMedia, which services more than 3,000 publisher clients, during a presentation at the AdMonsters Publisher Forum in Miami last week.
Client-side data is what is available on a user’s device or within their browser, like installed fonts or sites they’ve previously visited. Server-side data is derived from the ad server and includes IP addresses, which are typically hashed into an anonymized set of numbers before being shared in the bidstream.
What’s safer?
Some client-side data matching is coming under increased scrutiny from a privacy perspective, because publishers end up without transparency into what is being extracted from a user’s browser.
For example, Mediaocean’s FTrack ID uses only client-side data for its probabilistic ID matching.
“If you install their Prebid module, they run JavaScript that you don’t get to see because it’s secret and it’s obfuscated,” McCann said.
But publishers are supposed to be the gatekeepers of user consent. So, if publishers lack transparency into what client-side data is being extracted for identification purposes, they can’t ensure that advertisers are only using consented data.
ID solutions like Parrable, on the other hand, rely on server-side ID resolution, which uses signals like hashed IP addresses to detect unique devices.
But IDs that rely on IP addresses may not be long for this world.
Apple introduced iCloud Private Relay in 2021 as a means for Safari users to hide their IP addresses for marketing purposes. And last year, Google introduced its IP Protection solution for Chrome, which used to be called Gnatcatcher, as part of its Privacy Sandbox initiative.
The cookie question
IDs that rely on third-party cookies also won’t be viable for long. So, some ID solutions create first-party cookies on the publisher page instead.
Prebid’s SharedID generates a random number to identify a user and stores it as a local storage value within the publisher’s first-party domain (aka as a first-party cookie). Publishers are able to share the ID with ad tech companies that match it against their own ID graphs for retargeting and frequency-capping purposes.
Google AdX buyers can see an immediate lift in Safari CPMs when they use SharedID because it makes it easier to frequency cap across a single domain. (The ID doesn’t enable frequency capping across publisher domains.)
“Research has shown that buyers [mostly] care about frequency capping on the same publisher, not across publishers,” McCann said. “I’m not sure that that’s accurate, but at least it’s better than nothing.”
Buy-side ID modules, such as the Quantcast ID, Teads ID and Criteo ID, use the same JavaScript-based methodology as SharedID to create a first-party cookie within a publisher’s domain.
Because these solutions don’t generate a third-party cookie, they will remain viable even after third-party cookies are deprecated.
In contrast, The Trade Desk’s UID 1.0, which does generate a third-party cookie, will eventually be obsolete. That’s why The Trade Desk introduced UID 2.0, which instead relies on an email address provided to a publisher with a user’s consent. The email addresses are hashed and passed back to the publisher as an anonymized token that can be shared with SSPs and stored on the user’s browser as a first-party cookie.
What’s working?
CafeMedia has seen success with Prebid’s SharedID, McCann said.
“If you set [SharedID] as your publisher-provided ID in GAM,” he said, “then you end up getting about a 7% or 8% lift on AdX buying in Safari, which translated to a 1% lift for us in total revenue.”
LiveRamp’s Authenticated Traffic Solution (ATS), which uses a deterministic identifier called RampID, is also working well for CafeMedia, mainly because RampID is one of the most widely adopted identifiers among advertisers and data holders like Experian, according to McCann.
“The fact that someone’s credit score or purchase history on their credit card is available to the advertiser makes them incredibly valuable,” he said, “so just by having a LiveRamp match at all, we’re getting north of $7 CPMs.”
When CafeMedia authenticates traffic through ATS, it sees 300% to 500% CPM lift for Safari users and 20% to 30% CPM lift on Chrome.
Yet some buyers don’t want to pay for ATS, which, in McCann’s view, accounts for some of UID 2.0’s success.
UID 2.0 “is a play by The Trade Desk” to appeal to buyers that want to avoid the fees that LiveRamp charges for its ID, he said. “Advertisers are going for [UID 2.0] because they don’t want to pay LiveRamp’s fees.”
