You might already have some cool customer analytics based on machine learning in place. And you might be even using some semantic data when creating and targeting online audiences. But together these two become the ultimate Doomsday Machine against your competitors when they meet up in a DMP.
At its simplest DMP (Data Management Platform) is a place where data from different sources is gathered and managed to make actionable, smart decisions. A decision can be, for example, where to place ads with the help of Demand-Side Platform, a.k.a. DSP. DMP does this by sending instructions. These instructions are at the core of the whole DMP system – to whom, when, where and what. If you’re not yet familiar about the role of DMP’s in marketing, you can read a more in-depth description of common DMP functionality and use cases from this article.
The data itself inside DMP’s is not very smart although DMPs are called “smart databases”. With advanced audience analytics and semantic intelligence, we can get way beyond and build a DMP ecosystem with scalable outputs second to none.
Let’s go through a few examples of how to make your DMP even smarter and the data more actionable.
With the help of semantic intelligence collected from cookie data, a brand can learn more about audiences who are interested in it and its marketing communication. We can, for example, track users who have viewed a brand video and compare them to the ones who have skipped the same video and then match these users with interest segments created in a user profiling tool the like of Semasio.
This simple and straightforward exercise combined with the built-in exhaustive audience data, cookie matching, machine learning algorithms and statistical models the semantic intelligence utilities use to draw the insights, provides us with knowledge about the interest and intents of our core audiences. Right after this we can target our digital advertising to the very same segments we know are interested in our brand. We can also add this insight to our DMP for future use.
It is about understanding not only what happened, but by whom and why.
We can also use insights from survey based segmentations when creating and modeling smart online audiences. It is possible, for instance, to align our previous strategic customer understanding with current online advertising by creating and modeling lookalike audiences based on these insights. The outputs from this exercise, too, are scalable – thanks to DMP!
If you want to read more about segmentation practices in the age of DMP’s, take a look at here.
Modeling Customer Future Value
A good thing about DMP is that it’s very flexible. We can extract data from it, model the data and then send it back. With a single key (like a cookie ID) we can collect data from various sources and then match it to a single user. After extracting this data, we can model user’s, or in this case, customer’s future value based on various variables like age, transactions, location etc. When this data is sent back to DMP we can target customers with different future values with different, modified and optimized messages. The coolest part might be, however, that these outputs from advanced mathematical models are scalable as we can model lookalike audiences with the help of DMP’s cookie pool.
All in an effort to increase business value in a short- and long-term through more effective targeting of your key messages.
Predictive Conversion Modeling
Advanced analytics can be of great help when predicting web page conversions.
Think about this: with the best in class machine learning techniques you can predict conversion probabilities for each user visiting your web store. With real time connection between DMP and statistical computing software environments like R or Python we can target the advertisements based on the specific interactions as well as the likelihood of conversion.
For instance, we can create a rule that targets people who have over 20% probability of purchase and have visited the promotion page of a specific product. Once we have identified the most beneficial behavioral patterns, we can use the cookie data of the most prospective visitors to build larger target groups out of similar web users. All the groups can then be used in programmatic buying of advertisements.
The coolest thing is that all the information mentioned above can be loaded to DMP on real-time basis. Advertising efficiency guaranteed!
For most of the cases it’s crucial to have a partner with deep knowledge about the benefits of DMP, access to vast amounts of cookie based semantic data and skills to execute demanding machine learning models.
Your own CRM data is very valuable and must be included in the process but it’s not enough. DMP’s value lies in its ability to collect and combine data and to make actionable, real-time decisions.
Once again (I wrote about this in Kauppalehti a way back), C-suite’s willingness and commitment to use data and advanced analytics is vital when turning organization into a truly data-driven actor. C-suite people are more willing to give their support after having seen some actual benefits and results from data and advanced analytics. Scaled audience analytics and smart & effective online advertising would definitely be one these.
Moving into the world of advanced analytics and DMPs is not a small project but nobody wants to be left behind so better to start to plan it sooner than later. Why not already today?