The evolution (and big secret) of mobile ad targeting

Mahi de Silva is the CEO of Opera Mediaworks.

What is the deal with mobile ad targeting?

To start, we have to acknowledge the giant differences between desktop and mobile.

1. A lack of third-party cookies support. Safari, the default browser for iOS devices, doesn’t support 3rd party cookies and it accounts for nearly 50% of all mobile browser traffic in the US.
2. Though we have device identifiers, such as IDFA on iOS and Android Advertiser ID on Google, there are many restrictions around using these device identifiers for targeting and data co-relation.
3. Browsers aside, the majority of internet traffic now comes from mobile apps. Here, cookies are not persistent from app session to app session, and there are significant limitations around sharing information between apps. This makes it much more difficult to use behavioral and contextual information for targeting advertising inside mobile apps.
4. Advertising technologists, of course, have scrambled to come up with solutions to these big obstacles — and it seems like every week someone is coming up with some new workaround.

But the real secret to mobile targeting is not about finding the one magical technology that will allow you to track a user across multiple devices, or hack a system that was set up to rightfully protect user privacy.

The secret to mobile targeting is layering

(Yes, the same kind of layering that keeps hikers warm on a high-altitude trek, and conversely lets you avoid an embarrassing sweat situation after going from a cold morning walk commute into an overly heated office.)

You start with the foundation of identifying a user as best you can with deterministic data elements that you do have access to, given the restrictions noted above. Most ad tech companies use some form of basic “fingerprinting”, based on data elements gleaned from the network communication between the mobile device and an ad server. There are some tricks, of course, like accessing an anonymous subscriber ID from mobile operators to help you improve your algorithm, but at this level you’re still dealing with rudimentary data sets.

Then it gets fun.

1. First-party data

Many publishers have registration information and login access to their content that can be used to track a user as they cruise through the web and across multiple devices. Google, Yahoo, Facebook are the obvious ones, but there’s many many more.

2. A cooperative data management platform (DMP)

Publishers, governed by their privacy policy, provide non-PII information to their mobile ad platform, which then allows for the creation of a proprietary user identifier. (At Opera Mediaworks, it’s called the “Opera ID.”) That information goes into a collective sea of data that other publishers are also contributing to.

So, we might have age and gender from one, location data like zip-plus-four from another,or content affinity from still others. There’s strong controls to ensure no leakage of data, and only those that participate can benefit from it as they leverage that information for real-time targeting.

3. Zero-party data

The DMP serves as more than just a publisher data dump. It’s also a living, breathing data collector of running campaigns. A user who clicks on multiple car ads might then be indexed as an “auto enthusiast.” When we see lots of referrals coming in from social media apps, we’re clearly seeing a high-volume social media user. Because the data is being generated from essentially “no party” but rather learnings from within the ad platform, we call it zero-party data.

4. Third-party data providers

There is certainly value to be derived from partners like BlueKai and Datalogix. We often get requests from both publishers and advertisers to integrate third-party data. On the front-end of the process, it helps identifying audience segment for a particular campaign. On the back-end of the process, correlating campaign data with third-party data providers, helps deliver post-click or post-campaign analysis that helps advertisers see the ROI for their mobile ad spend..

5. App ownership and use

When we have permission to correlate data from a publisher that has the Opera SDK embedded in their product, one of the things we can do is detect the nature and use of those apps. Just as the cards you carry in your wallet are telltale signs of your personality and behavior, the apps on your phone can indicate a lot about a user. If Bleacher Report is the most-used app, for instance, it’s fairly safe to say that person is a giant sports fan.

But, these layers are only the beginning

In the past year, many big steps were taken to build rich consumer profiles, leveraging consumer behavior to provide audience segments that were tailored to an advertiser’s specific needs. But 2014 so far has been far more about how to up the ante on the correlation of data in order to get more granular about each consumer. This is also to leverage their online behavior to effectively “follow-up” on all of the outreach and messaging that has been done.

By this time next year, the hope is not that people will be asking “what’s the deal with mobile targeting” but rather questioning “how can I deepen my intelligence for more accurate mobile re-targeting.” That’s a much more evolved question — and that is the path that mobile is inevitably heading.

As CEO of Opera Mediaworks, Mahi de Silva runs the world’s largest mobile advertising platform that powers the global mobile economy. Opera Mediaworks, a fully-owned subsidiary of Opera Software, serves more than 14,000 sites and applications, with more than 65 billion ad impressions per month reaching 400 million global consumers.