• by Ray Schultz , Columnist, December 5, 2016

Marketers have accepted the idea that they can no longer measure attribution one device at a time. But what should they do? Drawbridge fields models and device graphs to accomplish this task. It works with partners such as Foursquare, Placed and PlaceIQ to provide in-store attribution. One challenge is to link devices. Most linking schemes are deterministic — Drawbridge believes in a probabilistic approach. For perspective on these subjects, we spoke with Rahul Bafna, VP product management and partnerships for Drawbridge.  

What kinds of mistakes do you see companies making in cross-device attribution?
Rather than mistakes, we’ve seen a gradual shift in behavior. A few years ago, digital identity resided almost entirely in a browser cookie like Chrome, Firefox or Safari. But the digital identity of any consumer now resides across multiple devices. Companies understand that just because consumers see an ad on one device doesn’t mean they convert on that device.

Where are people converting?
You have to look at conversions as a whole. Mobile ads are influencing consumers, but people are converting on desktop. In 2016, mobile is 50% to 55% of the digital spend, but two thirds of conversions are happening on desktops — it’s a larger screen, it’s easier. It’s now fairly commonplace to look at targeting and conversion measurement at the user level, not on the device level. Understanding ROI is a little more complicated. Consumers are not necessarily just seeing one ad and converting — they’re exposed to multiple ads on multiple devices. Then there are offline conversions. 

Offline?
In-store. Digital purchases are maybe 10% or 12% of all purchase activity; physical store purchases are maybe 85% or 90%. Companies have popped up that use mobile data to attribute conversions. They look at the location of the user’s phone. This works well for big-box stores with a larger footprint. In our case, we’ve seen everything from banks to restaurants to organic supermarket chains drive traffic with multiple devices.

Do demographics influence where people will convert?
Smartphone coverage is now above 80%. So there’s no strong bias toward age for in-store purchases: It largely mirrors the general population. There may be some bias in digital purchases. Ad categories tend to be gender-specific: If there’s a campaign for women’s shoes, it’s weighted toward women.  

Any other new devices?
Linear TV — you’re likely to see more granularity. And it’s at the household level, not just the user level.

Which product segments are succeeding at attribution?
It happens when there’s a longer cycle, when the product is expensive and involves a lot of research, as in travel and e-commerce. And there’s increasing volume coming from Tesla vehicles and German cars with higher Internet connectivity.

Is there any link between these newer devices and social media?
From an advertising platform perspective, the types of data aren’t very useful there. Facebook and Google are now being called ‘walled gardens.’ It means that not only can you not take data from these platforms into your own, but you also can’t bring in your own data or targeting methodologies.

What about wearables and the Internet of things?
There’s no dedicated network connection, and limited or nonexistent ability for advertising or measurement. 

Drawbridge also offers ad-bidding services. What changes have you seen in that space?
There are a couple of big triggers. One is adaption of cross-device targeting technology by the major DSPs. The second is the launch of cross-device measurement by major ad services.

So what would you advise marketers?
It depends on the vertical. But in general:

• Understand what users are doing on the devices they own. Regardless of the vertical, the audience is using multiple devices, not only for buying products but also for consuming media.
• Don’t try to compare ROI on the device level — i.e., smartphones vs. desktops. You’ll get a lopsided view of what’s actually happening. Look at how the ad spend is done on the user level.
• Avoid using a last-click or last-touch model — it’s too narrow-minded. Instead, look at the full funnel of consumer exposure, and measure the true effectiveness of advertising.

What’s next for Drawbridge?
We’re announcing another partnership, providing access to product-level transactions tied to a user. Clients will be able to attribute that back to the ads.