Keyword Targeting: A Primer
We just released Keyword Targeting with lead customer Goldbach Group. According to Goldbach, it’s a game-changer for them when it comes to creating unique audiences. Here’s a brief primer on how it works.
Introduction
If you have been following us on Twitter or the German industry press, you will no doubt have heard about our release of the Keyword Targeting functionality for building audiences. Our lead customer for this release was the Goldbach Group, who was heavily involved in helping us fine-tune the concept and turn it into a full-blown product. Their ad and marketing teams at Goldbach have enthusiastically adopted this feature, as it goes so much further than standard demographic or contextual targeting. We figured we would take the opportunity to walk you through why is they are so excited about it.
A Keyword Wishlist
Let’s say a brand like BMW approaches a publisher to run a campaign to increase and reinforce brand awareness. In a classical setup, the publisher would ask for the desired target demographics, like age and gender to create a broad target audience. The publisher might also have the ability to target users who have demonstrated a general interest in automobiles and hone the user set further, but what if BMW wanted to specifically target users who had interacted with its brand or even one of more of their particular car models? That’s where Keyword Targeting comes in.
In essence, Keyword Targeting is a new way to build highly custom audiences. Think of it like a wishlist that consists of keywords, where you list a set of keywords and our platform returns the users who have interacted with relevant content.For example, if you listed a brand name (BMW) and a product name (X5), you would see the users who read articles about BMW or the X5 car
In order to further specify targeting, you can also define keywords you do not want the audience to have interacted with or “negative” keywords. For example, you might say that you want to exclude users who have looked at car accident or safety recall articles.
This combination of both positive and negative keywords is what we called a topic. With topics, you can create very specific audiences, going far beyond what traditional segmentation can achieve. Some fascinating applications include:
- Creating audiences that have interacted with a particular brand to reinforce brand awareness
- Creating audiences when entering new marketing segments , to create brand awareness
- Creating audiences that have interacted with competitor brands
Standard segmentation attributes, like age or gender, can be powerful tools when audience characteristics are well-defined. However, when the audience signals are more subtle, targeting by keyword interactions allows advertising teams to think of audience targeting in a fundamentally new way.
Keyword Targeting in Practice
What happens behind the scenes is fairly complicated (our system processes 500 billion user / interaction / word permutations daily to enable this feature), but our simple interface makes it fun to create these topics. Defining a topic in our UI really is as simple as typing the relevant keywords, making some minor tweaks if you like, and hitting save.
To give you full insights in how the process works, we provide a full list of URLs where the keywords you provided were found. This is another aspect where the high granularity of Keyword Targeting really shines: you get full transparency on which pages are used to model the audience, and you get full control on whether to in- or exclude them.
How it works
So how did we manage to build such functionality? The secret lies in the machines that keep our platform running and of course with the amazing engineering team that manage them. Our algorithms programatically fetches the pages in our clients’ properties and extract their content. This content then gets processed for its semantic meaning, and indexed accordingly (“Keyword ‘Pokémon’ appears on page X”). Most days, we process nearly 5 Billion of such keywords!
Conclusion
We have glossed over a number of additional features, included in Keyword Targeting, such as like recency, frequency filters or Artificial Intelligence-based expansion. With this post, we just wanted to introduce readers to the world of Keyword Targeting.
We are a predictive DMP, as we go beyond what the classical DMPs out there in the market can do. It’s not just about ‘ingest and display’ like the large DMPs do, it’s about using subtle patterns to make predictions and enabling our users to easily build unique audiences with assistance from algorithms. Keyword Targeting is a perfect example of discovering these subtle patterns at scale and exposing this functionality in a simple UI. It’s one of the ways we help data science and advertising teams globally be more efficient and spend time where it really matters.
Are you interested in experiencing how simple it is to use keywords to build audiences? Get in touch, we’d love to run you through it!
