Contextual targeting and advertising: how good can it get?

Behavioral targeting is mostly gone on iOS. It’ll soon be significantly challenged on Android as GAID dies a savage death and Privacy Sandbox for Android rises from its ashes. What we’re left with is contextual targeting … is the mobile app advertising sky falling?

We recently had Remerge CEO Pan Katsukis on the Growth Masterminds podcast to chat about contextual targeting in mobile advertising. The question: how good can this get? The unspoken fear: how bad is it really?

Hit play on the video above to kick off the convo, and keep scrolling to get just a few of the highlights.

And, before we get started, a quick definition of terms as they’re typically used in the mobile marketing ecosystem:

Behavioral targeting: targeting ads to people based on what they’ve previously done in other apps:

• Usually requires some kind of tracking mechanism
• Generally has significant privacy concerns
• Essentially impossible now on iOS with App Tracking Transparency
• Can potentially be done in a privacy-safe way (Privacy Sandbox for Android is one attempt)
• Typically can be used to target narrower groups of people with very specific interests

Contextual targeting: targeting ads to people based on their current context:

• Generally requires knowing what app a person is in right now, or at least what kind of app
• Generally regarded as much more privacy-safe than behavioral targeting
• Typically better for broader targeting of larger groups of people with general interests

First: contextual targeting isn’t just simple context

The first thing to keep in mind is that context isn’t just simple. 

Modern contextual targeting in mobile apps often isn’t like contextual targeting on the web. On the web, Google or some other ad network has spidered a web page or video, knows what content lives on it or in it, semantically understands what that content is about, and targets a relevant ad to an audience that’s generally interested in that content. The hope, of course, is that a person who finds Sportfishing in Alaska interesting will also be interested in ads for fishing flies, or rods, or guided trips, cabins near good fishing streams, or fishy wall clocks with Amazon Alexa inside.

(Don’t ask.)

There’s some of that on mobile as well, but much of that kind of context comes from the app itself or the app category. (Note: some adtech platforms like Inmobi can potentially “read” app content and use that as insight for contextual targeting, assuming the right SDKs with the right privileges are in place.) 

But many key aspects of in-app contextual targeting come from less visible and obvious data points.

“In the end it can be any signal or attribute that we know about the current situation of the user … the simplest one is what time is the device right now,” says Katsukis. “What time of the week is it right now? What type of app are they using … how long have they been using it? So anything else which comes from the publisher and the app itself, like the app rating, you can find out: the category the app is … that gives you some information and understanding like what the user is up to right now, what they’re interested in, where they’re engaging.”

In fact contextual advertising can take advantage of up to 100 signals, Katsukis says, including details like:

• WiFi or cellular connection
• Bids/pricing information from the ad network
• Rough/coarse location
• Device information (brand, OS version, device version)

And this can be used in order to design relevant offers and drive good creative to the user, providing insights to whether they’re at home or office or on the go, where they are in the world generally — example: in Germany during Oktoberfest — and potentially demographically-relevant information such as socioeconomic status, based on device type and extrapolated cost.

Add them all up, and you’ve got data that can be incredibly useful, particularly in certain verticals. My stomach, for instance, doesn’t care what app I’m in when 6PM hits and it feels empty. Match it with other privacy-safe data like weather, calendar, and events, and you add relevancy step by step by step, and maybe get a new Uber Eats customer at the same time.

This really works for some verticals, Katsukis says.

But the bad news is that it’s significantly challenging to find niche, tightly-targeted audiences at an affordable price via contextual advertising. Hence the challenges that D2C (direct to consumer) is having right now in targeting potential customers, and the challenges that some — not all — mobile games are also experiencing.

Second, you need some technology

This probably goes without saying, but you’re not going to be great at contextual targeting by just looking at one or two factors. 

The good news is there’s a lot of data available from multiple privacy-safe sources:

• Device data
• App/publisher data
• Adtech platform data
• Environmental data (weather/calendar/events) 

The bad news is that you’re not going to be able to analyze dozens or scores of device-derived, app-derived, adtech platform-derived, and environment data like weather and events manually or in real-time without some serious tech.

“You can’t really get far with contextual if you do it all manually,” Katsukis says. “The case [of, for examples] let’s target everyone at 7:00 PM on a Friday after work … so they order food. That’s an easy example, but if you want to scale that [and] be really good nowadays you can leverage machine learning to look at all the history, everything that has been done, and even for that, use the ID inventory to try to understand … what’s working and try to make connections with the different other signals you see there.”

In other words, get sophisticated. Build some models; do some training. Use IDFAs when you have them (and GAIDs) when you do. Leverage them to measure your contextual advertising effectiveness and get better and better at it. (Or: get your adtech vendors to do all this work.)

One thing is for sure. You won’t be the only one testing:

According to the IAB, 74% of marketers are using contextual data to improve their advertising post-device ID and post third-party cookies on the web (of course, the third-party cookie just got a new lease on life from Google). That’s more than those leveraging first-party data (60%) and more than those who are leaning on (and paying for) publishers to do their targeting for them in the black box of their platforms (45%).

Exploit the 50%-off contextual advertising discount

One benefit: you’re getting a contextual advertising discount. A significant discount.

“Right now the inventory when there is no ID is still — and I looked it up just today — 53% cheaper,” Katsukis says. “That’s obviously an opportunity for an advertiser to tap into and just figure out, okay: can I get as far with contextual only without an ID, while still having the opportunity to bid 53% lower … as there’s not much competition?”

The clear opportunity is that if you can get even half as good results with contextual advertising as you did with behavioral targeting, it’s a break-even proposition. And if you can improve just a little bit on that … it’s all gravy.

That’s a big if, of course.

Niche apps with specific requirements for players — and a strong desire to hunt whales — will find it toughest. Larger apps and brands with more generic or universally appealing offers and value propositions will have an easier time using somewhat mass media marketing tactics in a digital, contextually targeted environment.

Dive into the whole conversation

Watch the whole conversation above, and subscribe to our YouTube channel so you don’t miss new episodes. 

Prefer audio?

No worries: subscribe to the Growth Masterminds podcast on just about any podcasting platform.