ID Spoofing Dents Campaign Performance

The clandestine swapping out of IDs in cookieless environments without consent hampers the campaign performance, data shows. Go figure.

These practices, dubbed ID spoofing, have been under increased scrutiny since February and are the subject of conversations within the IAB Tech Lab. It’s tricky to detect when ID spoofing is happening, and this will only get harder with cookie deprecation.

Data from demand-side platform Basis Technologies found that when publishers and supply-side platforms use techniques disguised as regular cookie-based targeting, conversion rates decline.

“For the campaigns that we’re buying that had very high level of ID integrity … the conversion rate is the highest,” said Ian Trider, vice president of product, DSP at Basis Technologies.

There are several adtech reasons why ID spoofing is bubbling up. Auctions with IDs monetize higher than those without, so there’s some incentive for SSPs to maintain signal. Buy-side tech firms believe they’re bidding on users via a cookie, but the sell side is using other methods to identify users. Some on the sell side say they are being transparent.

ID bridging across browsers occurs when a publisher uses a signal like an email or IP address to link a user on Safari with who they are on Chrome. The DSP will treat these impressions like any other group of Chrome users, but this user cannot be tracked throughout the web.

Basis explored the impact of this practice by looking at the rates of ID consistency for 3,126 campaigns on the web during the first week of February. Ideally, the cookie in the bid request should be consistent with the cookie on the actual ad impression close to 100% of the time.

If not, ID bridging—or even outright fraud—could be at play.

An anonymous source from a second DSP also uses this methodology to understand the scope of ID spoofing.

Sure enough, Basis found that for campaigns where the average consistency rate between bid and impression was near 100%, 6% of ad impressions resulted in conversions. But when the consistency rate was below 50%, the conversion rate was mostly below 1%. In this case, a conversion is any event an advertiser would want to optimize for, from viewing an ad to going to a brand’s website to buying a product.

Manipulating bid requests in unconventional ways

Basis’ findings make intuitive sense: Brands can only measure conversions if they can track a user from ad to outcome—a process currently mediated by third-party cookies. If the sell side uses techniques to make a cookie only visible in the bid request but not in the rendered ad, the advertising tracking mechanisms break down.

“An inconsistency indicates spoofing of the cookie in the bid request,” Trider said. “We do not approve of any instance of manipulating the contents of a bid request in an unconventional way. It is certainly something that we need to discuss before it happens.”

Basis’ data finds a general positive trend between ID consistency and conversions: When consistency is higher, so are conversions. However, the correlation is not 100%, which Trider partly attributes to the data being aggregated at the campaign level, not by impression, which would be more precise. There are more campaigns with high levels of ID consistency than low levels of ID consistency, and the sometimes smaller sample sizes add noise to the data, Trider said.

Contractual agreements do exist

Of course, it’s possible that a brand is misled about the identity of the user it’s bidding on, and can’t measure outcomes associated with that ad, but that the ad is still driving conversions the buyer doesn’t know about.

“If there were zero improvements with bridging, no one would do it,” said Mike O’Sullivan, co-founder of metadata firm Sincera. “It doesn’t mean that all approaches to this are performing.”

There is an opportunity for adtech firms to offer solutions for brands to configure their websites to register conversions from ID bridging (as long as brands consent to buy ads via ID bridging), said O’Sullivan.

But for now, only what can be measured matters for campaign planning.

“We don’t bid on traffic where we don’t think we can measure conversions,” Trider said, adding, “What can be measured is the only thing that effectively matters to people.”

More awareness to fix the problem

Within Basis’ sample, 7% of the approximately 408 million impressions studied had an inconsistency where the cookie in the bid request did not match what was in the impression level data.

“It’s a minority of campaigns that see a huge impact,” Trider said, adding that Basis is working with relevant exchanges to stop the behavior.

There is also activity from the sell side. SSP Magnite published a blog post last week arguing for more transparency in how ID bridging practices are declared to buyers.  

“If buyer and seller know what’s happening with any given ID creation, they can make their own determination,” Garrett McGrath, senior vp of product management at Magnite, told ADWEEK. “Everybody agrees that there must be awareness of it.”