Lifting The Veil On Audience Modeling: It’s Not Just Retargeting
Over the past two years, I’ve been focused on applications of the remarketing toolset. Each time I have a conversation about my work, the response can be summarized by “Oh, you’re doing remarketing.” But I’m not remarketing. I’m building audiences. After the umpteenth conversation about how remarketing is a single form of audience-based advertising, I […]
Over the past two years, I’ve been focused on applications of the remarketing toolset. Each time I have a conversation about my work, the response can be summarized by “Oh, you’re doing remarketing.” But I’m not remarketing. I’m building audiences.
After the umpteenth conversation about how remarketing is a single form of audience-based advertising, I want to open up the hood and provide some clarity on the evolution of remarketing. Allow me to explain how the underlying tools of a simple and effective marketing channel have enabled some of the most innovative and effective marketing methods available today.
List-based marketing is not a new concept. Direct mail, one of the most effective marketing channels ever, has been around for decades; and its modern twin sister, email, is an online marketing workhorse. So when someone had the genius idea to leverage a little pixel/cookie technology to build lists of users for online advertisements, it was destined to be a successful medium.
In its first incarnation, list-based marketing in the display ecosystem was dubbed “Remarketing” (remarketing and retargeting are used interchangeably). And, because marketers demand innovation, that same core technology was spun off to segment users by various on site behaviors, search histories, and eventually model common attributes to identify incremental traffic opportunities rather than just monetize existing traffic.
Remarketing Vs. Prospecting
At the highest level, you can break your campaigns down to remarketing and prospecting. Remarketing is when you’re advertising to someone who has been to your site before, and prospecting is when you’re advertising to someone who has not been to your site before.
Common Types of Remarketing
- Shopping Cart Abandonment
- Product page views
- Engaged visits (multi-pageview or long duration)
- Event (signed up for newsletter)
Common Types of Prospecting
- Category / Channel Targeting
- Search remarketing
- Lookalike modeling
- Affinity Marketing
Think about each time you have been retargeted. What was it you did that led to your becoming a person of interest?
The Art Of Segmentation
At a logical level, audience segmentation is relatively straightforward. Users with a common intent and value can be grouped together. In other words, you can show the same ad with the same bid to people that have a common interest and likelihood to convert.
Interest is defined by what you did — did you put a TV in your shopping cart to see how much it costs?
Value is defined by your predicted conversion rate. Let’s pretend that 1% of users who bounce after adding TVs to their shopping cart end up purchasing a TV within 30 days. A good marketer can calculate the value of a visit from a user who has already added a TV to their cart:
Value of a return visit multiplied by the Ad Cost goal becomes the bid in a CPC remarketing model.
A Leap Of Faith
Prospecting, on the other hand, is entirely based on assumptions. These assumptions are well calculated, but the further you get away from a known behavior and the deeper you dig for relationships, the less accurate our assumptions get.
The backbone of prospecting is built on the monetization of publisher data. A publisher like CNET offers affinity marketing — an advertiser like Roku might be interested in leveraging CNET audiences who reviewing streaming devices or set-top boxes.
Lookalike modeling has quickly become one of the most popular and effective forms of prospecting. Ad networks leverage proprietary tracking to create high value audiences (such as high lifetime value customers or known converters) and identify common characteristics of these users. Then they look for users who are not in the audience which have similar characteristics and add them to a “Lookalike” audience.
Google offers its own unique prospecting via search remarketing. Users having searched specific queries are added to audiences and can be remarketed to with the Google Display Network. So, for example, you can advertise to a user who searched for a key term but didn’t click on your search result.
All of these targeting methods have their strengths and weaknesses. But no matter how you spin it, they are all unknowns and bear risk for any direct response advertiser. If you try out prospecting audiences, start with a small budget and low bid and slowly increase as things can get out of control quickly if you’re too aggressive.
Whether you’re looking to monetize your existing traffic or bring in incremental traffic, the remarketing toolset is incredibly powerful. Down the pipe remarketing companies like AdRoll, Tellapart, and Criteo offer great solutions, but just because you’re leveraging their tools doesn’t mean you are maximizing your remarketing potential.
I strongly encourage you to look into your data for correlations between actions and events. Layering in third-party data or leveraging more sophisticated tracking solutions can enable some fantastic DSPs (Demand Side Platforms) and data aggregators, unlocking all sorts of new ways to monetize your data.