One of the biggest use cases for a Data Management platform is to enable brands to collect audience data from multiple sources in order to create a single view of their customer.
As part of the data collection process, the DMP merges data that it collects from multiple sources. This process is called matching. Matching happens between multiple online sources (such as a marketing website, blog and Web store) as well as online & offline sources. Through data matching and normalization, the DMP can link overlapping records together and enrich audience records as data flows in.
If you take the example of a retail business that has web and physical stores, as in our example brand fitness.at.work, we will have web analytics data collected from the online Webstore. We will also have offline from the physical store point of sale systems.
In order to get a unified view, the DMP will need to match this offline data to devices and web browsers of members. One of the ways this done is through the cookie that the DMP sets on a user’s browser.
When a user logs in online, the user’s member ID can be (masked and) sent to the DMP. The DMP will then tie this ID to the ID in the cookie that it set in the user’s browser. Therefore, the next time the member comes to the site, the DMP is able to recognize the member, regardless of whether the member is logged in or not. The member ID is hashed and stored in the DMP to safeguard against a PII breach. More on hashing later in this post.
Now let us say the fitness.at.work’s member saw some fitness gear online. The info about products she viewed are tracked the DMP through Tag Management.
Now this member goes to the physical gym store and purchases some of the fitness products she saw online. The POS captures these purchases against the member ID. And the next time offline data sync happens with the DMP, it ties the purchase to the cookie via the member ID.
This is how online-offline information is connected within the DMP. It works for cookied interactions. But how will this work in the mobile world where cookies don’t work well? We will look at that in a future post.
Hashing
Building a single view of a customer often requires bringing PII and non PII data together.
In order to respect a consumer’s personal information and also minimize exposure to risk for both brands and consumers, PII information must be hashed (MD5, SHA-256) before storing in a DMP.
In the above example, the member ID uniquely identifies a member of fitness.at.work gyms. This member Id is hashed before sending it to the DMP. And this hashed ID is assigned a unique identifier within the DMP. Now, all subsequent data (for a member) sent by the data sources are sent using this DMP generated unique ID so that the DMP can stitch it together with existing data.
One Comment on “How does a DMP match online and offline data?”
well, what can I say about the way you have put this article so detailed and simple it’s really good and it did this topic, thanks for sharing. So I think the customer data platform is the way to go in this year and I think it wouldn’t hurt u if you have it in your business even if you don’t think its worst, for me it’s worth having and I will be implementing it this coming year. you can check my post on customer data platform for more.