Why Starting With Cookies Doesn’t Make Sense For Cross-Device Tracking

Are some marketers approaching cross-device backward?

Tom Laband, CEO and co-founder of mobile data marketplace adsquare, thinks so.

“We see many advertisers start with cookie-derived data and bridge that to mobile IDs using cross-device vendors – and to me, that doesn’t make sense,” Laband said Thursday at a company event in New York City.

Device IDs, like IDFA on iOS and Android’s advertising ID, are persistent and often used as the connective tissue to link customer activity and identifiers across devices and channels.

“These are way more persistent than a cookie ID,” Laband said.

But scale is always an issue, and in order to get there, onboarders and cross-device providers are increasingly turning to probabilistic methods to pump up their reach.

“What’s happening now is that [marketers are] taking a desktop cookie and probabilistically connecting it to a mobile user,” said Kathleen Comer, GM of client services at The Trade Desk.

Even if they’re using a cross-device vendor like Tapad, Drawbridge or Oracle’s Crosswise to make the match between a cookie and device ID, the audience that’s being created is essentially a lookalike audience, rather than a one-to-one match to a device ID.

“I think we need more marketers to start using mobile data for mobile campaigns, as opposed to working with the big cookie-focused DMPs out there with fancy cross-device technology that allows the bridging of cookies to mobile devices,” Laband told AdExchanger.

It also makes sense to start with mobile IDs, he said, because the lifetime value of an app identifier is greater than a cookie – even a persistent cookie, which doesn’t expire when users close their browsers, but is at risk of being cleared at any time.

In other words, Laband argued, a more solid strategy would be to coalesce cookie data around a persistent anonymous identifier tied to a device, rather than creating cookie syncs between tracking data across different digital ad platforms – which can lead to data leakage – and then connecting those matched cookies to a mobile ID.

“When you start with a mobile device, there is no cookie syncing or there is less, and then you can use cross-device technology to bridge to cookies,” Laband said. “A profile can be more granular and be more accurate using a mobile ID, and mobile also has data sets that are not available in the cookie world, like location data and telco data.”

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Hmmm, I agree with the sentiment, but probabilistically tying humans to a device or cookie ID kind of got us into this mess in the first place. Throw in the soon to be enacted General Data Protection REgulations here in Europe and this becomes a very shady. Utimately the publisher and the advertiser will have to own the ID of rthe subscriber and customer and should form direct relationship deterministically. That way, 3rd party ad tec doesn’t get in the way, data leakage doesn’t take place and ad perfomance improves. Publishers then get more of the ad spend for their hard work and advertisers get better ROI. It’s why we built Adapptive.

  2. Can Mr. Laband explain what impact there is on keying to device ID when the person gets a new device? Phone replacements (upgrades) have been common, about every two years or so, the way US based carriers have promoted the devices, though that cycle may be changing now. In any case, what is the impact on data reliability when the same user gets a new phone (or tablet) for the old? Same user, same phone number, new device ID.

    • Hi Henry. I was actually at the event yesterday with Tom. To answer your question, it all comes down to HOW you’re gathering this information. Our company (OneAudience) presented yesterday at the AdSquare show and we explained that we gather this device ID information through our SDK (Software Development Kit). Based on the permissions we have to gather data like this, we are always pinging the device to get the most up-to-date information. So the use case you present where ‘what happens when the user gets a new device?’ — as most people back up the apps on their phone/tablet, if they restore that back up on their new device, we automatically receive the updated Device ID because our SDK is transferred through the backup. Does that make sense?

      • Henry Blaufox

        Yes Ari, it does. I suppose for purposes of industry trust, it is just a matter of showing how many people carry over the apps and other information to new devices. That would offer a view of coverage reliability over time. Also, how far back in time is overkill when profiling user attributes?

    • tyler pietz

      @henry the impact is virtually the same as when a cookie expires or is cleared from the browser cache; either you successfully drop a new cookie / capture a new id or you don’t. however, device ids are far less fractional and persist for much longer (on average) than browser cookies, meaning ids are significantly more stable

  3. Interesting article…certainly agree with the thought that targeting for mobile devices needs to use mobile data and not based on some desktop cookies.

    Question – so which are the companies which are currently offering this mobile data for mobile device campaigns?

  4. Wasn’t Facebook’s Atlas supposed to ‘crack’ this, by using a Facebook ID, cross-device, to match/track a user vs. ‘standard’ ad cookies? I heard (very few) good Case Studies around this – a shame it didn’t take off as well as it did, likely given that it had a very limited UI/platform when I had used to it…

  5. @Henry: Just as Tyler says, the Mobile AD ID is more persistent, meaning users do not delete it as frequently as a Cookie ID. When it is deleted the associated data is gone. BUT: Average life time of a Cookie ID = 4 weeks. Average life time of a MAID = 9 months

  6. @Anup: There are few platforms aggregating mobile data from different sources, adsquare is one of them. We differentiate between 1st vs. 3rd party, mobile vs. offline, declared vs. inferred, descriptive vs. predictive data. Ideally as a data buyer, you know where the data is from, how it is built and derived, what it describes and how recent it is…

  7. Jason Cutter

    Tom’s comment on starting with a mobile device and then using cross-device technology to bridge to cookies is definitely the best methodology. We map devices to HHs and then are able to use IP address for laptops, desktops & connected tvs. One thing that isn’t called out in this article that should be is the fact that marketers want transparency on cross-device methodology in order to believe the hype. 4INFO published/patented for anyone to see. It’s the reason I left a comfy client-side role evaluating technology vendors to join. The other thing is that matching to HHs with high accuracy opens up the oppty to close the loop on offline transactions which is what I was primarily trying to do at my client-side retail gig.

    • Also note that high accuracy does not mean, cannot mean, one hundred percent. Overall, eighty percent or so is attainable, and should be sufficient. It means the marketer is reaching the intended target eight times out of ten. Why would one not be satisfied with those results, especially if the incremental cost to get beyond the eighty percent threshold is considerably greater than getting to eighty? Such diminishing returns are rather common with this software category.

  8. It’s important to separate the 2 issues: 1) tracking users (devices or browsers) that reject cookies and 2) connecting them to a common user (person) through a device graph.
    As the the title states, trying to bridge devices (#2) without persistent cookie-free tracking is a fruitless exercise. At Flashtalking, we overcame this challenge through the acquisition of D9 (renamed fTrack) which provides highly accurate cookie-free tracking based on 50+ non-PII signals and advanced algorithms to identify and track mobile and desktop users (web and in-app). Using fTrack IDs (vs. cookie IDs) we can deliver much more accurate cross-device attribution by: 1) using fTrack to re-aggregate impressions, clicks, visits and conversions for each browser or device, and 2) use fTrack IDs (vs. cookie IDs) to match and bridge users to 3rd party device graphs.
    Given the growth in mobile engagement and desktop browsers that block 3rd party cookies, device graph match rates and bridges rates are significantly understated when relying on cookies. The lift from cookie-free tracking is so great, we would not waste time and budget trying to do cross-device analytics without it. For more see a case study on Adexchanger at https://adexchanger.com/mobile/uk-airline-monarch-tries-cookie-less-attribution-cover-data-gap/