This is the subhead for the blog post

The Impact of Cross-Device Conversion Tracking

Digital media affords advertisers the ability to track users from initial exposure to eventual conversion. This measurability has driven digital media’s explosive growth; the more effectively marketers collect and analyze data on their campaigns, the more effectively they can optimize. But conclusions drawn from data are only as good as the intel they’re based on, right?

As discussed in VP of Decision Sciences Feliks Malts’s previous post about DMPs, DSPs, and identity graphs, it’s not as simple as dropping a cookie on a user’s lone device and tracking them indefinitely. According to eMarketer, an average US household accesses 9 to 12 smart devices per day; each of those devices is likely bombarded with ads, third-party trackers, and cookies from a slew of vendors and platforms.

One conversion, multiple touchpoints

Pat sees an ad for a new exercise bike on their phone. Later, Pat sees that same ad on their work computer. When they’re done for the day, they go to the bike’s website on a personal computer and convert. Which ad gets the credit? Which device gets the attributable conversion? More than likely none of them (it will just show up as a “direct” or “organic” desktop conversion un-influenced by ads).

How many times have you made a purchase on your phone? Maybe occasionally depending on the platform (Amazon’s mobile UI is great for eCommerce). And yet we know, both as consumers and advertisers, that whatever the device breakdown of ultimate purchases, the majority of digital media spend is on mobile now:

Bar graph of US mobile ad spending 2020-2024

What does that tell us? Well, unless the advertiser is using an attribution model that takes into account cross-device tracking, the majority of that spend isn’t getting appropriately credited. Setting phones aside, there are other digital mediums – CTV, digital audio, etc. – that are even less likely (or impossible!) to attribute direct conversions. Are you going to buy a new set of shoes straight from your smart TV? Probably not. And yet we know that these ads are effective (through media mix modeling, match market testing, etc.) even when we can’t directly connect a conversion to them.

Why getting the full picture matters: a cross-device conversion tracking case study

For example, we deployed a digital media campaign for one of our eCommerce clients. Each ad platform (such as Facebook and The Trade Desk) had its own attribution and tracking, and 3Q used Google Campaign Manager to help with deduplication. Pretty standard stuff, right?

3Q’s preferred programmatic DSP (demand-side platform) is The Trade Desk (TTD). TTD has multiple cross-device attribution models available from industry leaders such as Tapad and LiveRamp. The campaign wasn’t segmented or siloed to a single device; instead, we took an audience-centric approach through a combination of first- and third-party data (thanks to Lotame, our preferred DMP). Everything was set up to a 60-day lookback window to help identify latency.

So how’d it perform in the first six weeks? Well, depends on who you asked.

Shocking, right? Fortunately 3Q had seen this type of performance gap before, and we had an idea of what was behind it. So why the colossal difference? Cross-device conversion attribution.

TTD uses a mix of deterministic and probabilistic third-party device graphing by industry leaders while Campaign Manager was only attributing device-to-device (saw an ad on a phone, bought on the same phone). Well, luckily we can turn on Google “cross-environment” tracking and pull a retroactive report, right? Sure! Here’s what that showed us:

Not much better. This is because Campaign Manager leverages Google’s antiquated (and limited) cookie-based device graph. It’s largely probabilistic and tenuous at best. But maybe TTD is hyper-inflating their numbers to look good for themselves, right?! Well, we pulled TTD conversions without any cross-device attribution and got this:

Much, much closer, with the delta being explained by cross-media duplication or similar measurement vagaries.

As advertisers, our job is to spend our client’s money as responsibly and effectively as possible. We care more about the business outcome (which for most 3Q clients is growth) than the measured results. However, we need accurate data and measurement to make sure we’re doing our job.

If we had relied solely on Campaign Manager’s system (cookie-based, limited or no cross-device tracking), we might have made the erroneous assumption that TTD’s CPA was 10x higher than it was, and an irresponsible use of money. Instead we found that it was highly effective; it just took users a handful of impressions on their phone before they went and bought on their desktop. To corroborate this, we used a combination of media mix and Pearson r-value
correlation modeling, which showed a very strong relationship between our TTD spend and our overall business conversions.

This measurement blindness is only going to become more important as cookies phase out and probabilistic models become more common. Partnering with vendors and using platforms that are able to cut through the clutter and appropriately track your customers is essential to measuring the success of your media dollars.

Learn more about our digital marketing attribution services!