Have you ever noticed when some ad is chasing you after visiting a particular site? Just show interest in certain goods and use the Internet to learn more about it, for example, on the manufacturer’s website. And here you go—advertising with this product entices you to go to the online store and buy it.
How does this happen?
The era of Digital Marketing
We live in days of active technology development. That’s a fact that everyone is aware of.
Since the 1990s brands and companies have changed the way they do their business and now they widely use technologies for marketing and selling of their products and services. So to learn and use different strategies and trends on how to promote their business with technologies, a Digital marketing term was invented.
According to Wikipedia, Digital marketing is the marketing of products or services using digital technologies, mainly on the Internet, but also including mobile phones, display advertising, and any other digital medium.
Digital platforms are increasingly incorporated into marketing plans and everyday life. More and more people use digital devices instead of visiting physical shops. In result, digital advertising campaigns become more prevalent and efficient.
The question then arises as to how we should process data about users and actions they do, in order to convert them into profits using advertising campaigns?
What is a DMP
Here DMP comes to the rescue. DMP (Data Management Platform) is one of the key stacks of a software platform (often, but not always, an advertising platform) that collects, organizes, stores and analyzes data from any source. Therefore, it acts as a tool for determining the target audience.
Simply put, DMP is a large database that stores processed data about your customers, as well as a set of services that work with this database.
The processed data don’t just come out of thin air. They are obtained by analyzing raw data and storing it in a database. For example, those might be structured events and users. Data will be stored in different, but related database tables.
Raw data comes in constantly. It can be information about the website visits and user’s actions, for example, subscribing for a mailing list and downloading the promotional material, filling out the registration form or buying a product.
Data is not limited to web pages only. Just as easily, we can receive information from mobile devices, like smartphones or smartwatches, each time a user interacts with any applications.
That data might also come from CRM systems or social networks. Even from your TV set—which videos were viewed, in full or in part, your likes and comments. This raw data differ from the prepared ones. We can get any input data and in any quantity. There is no web DMP or mobile DMP. Modern DMP should be cross-platform and be able to ingest all the data.
DMP vs Data Warehouse
But it’s not just storage. When we talk about Data Warehouse, the information gets there without any preliminary processing. In DMP it undergoes an additional transformation. Based on the incoming data the necessary models are built, data analysis and segmentation is done, etc.
How DMP collects such data
In fact, everything is simple. For websites, when a javascript tracker is triggered, it collects the necessary information and sends it to the server (collector).
For mobile applications, TV sets, watches, etc. these are special software development kits (SDKs) that are used during application development. As a result, they perform the same actions, send the necessary information to the server.
For the 3rd-party CRMs—it can be API (Application programming interface) which interacts with the system, receives the necessary data and transfers it further to the collector.
Note: Below, we will be explaining business logic on the example of a tracker for websites, but it should be understood that similar algorithms occur in the case of mobile devices, CRM, etc.
Once DMP got the data
So, once DMP got the data it should decide what to do with it.
The first step is to organize the data. There is a lot of data and we need to bring them into a structured form to simplify the work with them in the future, and also to save space in the database.
Further, this data is enriched. For example, having an IP we can get the user geolocation—country, region, city, zip code. With a user agent (browser data), we can get information about the device that the user is using: it’s brand and model, all knowledge about the operating system, the size of the screen, etc.
Creating a User Profile
ОК. We have a lot of events, but where do users come from? Again everything is simple. The first time you visit a website, you are assigned a unique ID, which is saved in the browser using cookies. In the future, the same cookie is used for other visits. If you later log into the CRM system, the DMP, having that unique ID, can already get more accurate data about you. But even without logging, often DMP has already enough information to build your profile and figure out who you are: а man or а woman, how old you are, what you are interested in and what you intend to buy at this time.
Where that info comes from? Maybe Machine learning?—Of course!
So, the DMP receives an event stream. Each event has its unique ID which is used to create the original user profile. So far there is not much information there. Although if we’re talking about the page view events, the body of the event has the address of this page.
Such events are sent to another stack, a classification. The page’s content is parsed. Different machine learning algorithms are used which based on the content of the page, make an assumption about you. And the more pages you visit, the more accurate information is gathered about you.
How to place a tracker on the website
Well. You may have another question now. It’s clear now that to get a user profile we need to set up a tracker somewhere. But what if you own a few websites? What are the odds of someone visiting them so that we can create this user’s profile?
In such cases advertising platforms do help. An advertising campaign is created and along with the banner display, the tracker is triggered. You surf the Internet, advertising materials are shown to you and, guess what—at the very same time data about you are sent to the collector server.
How does it help us in business?
So, a DMP is big data. And we want to get some benefits from it.
Analytics
The first thing you can get from this data is analytics. Various reports, such as visit statistics, the daily number of sessions, gender and age statistics, and much more can be obtained through processed data.
Segments
The second one, but no less important, it is possible to build segments on this data basis. For example, you can choose women 25-35 years old from the US out of all profiles who are interested in cars. Why do we need this? To help other systems to study an audience.
For example, it could be a DSP (A demand-side platform), which using DMP data, helps advertisers find their target audience. So, in the future, you will be able to start a new advertising campaign, but banner advertising will be shown not to all visitors, but only for those who are interested in it.
What are the big DMP players
According to G2 Crowd, there are lots of big DMPs and the 10 most popular leaders are:
- V12 Data
- Lotame
- Adobe Audience Manager
- MediaMath TerminalOne Marketing OS™
- Oracle BlueKai
- Salesforce DMP
- Signal
- SAS/ACCESS® 9.4 Interface to PC Files
- Google Audience Center
- Nielsen DMP
Of course, each of them has its own pros and cons and it’s up to you whom to choose.
Are there any open source DMPs
We don’t think there are open source DMPs. However, there is a certain great analytics software. Some of them are:
- Matomo (formerly Piwik)
- Snowplow
Also check out in our next article how to build a simple DMP, its architecture, which programming languages and technologies are relevant in this domain.
Conclusion
Digital marketing is an actively developing industry that requires a multi-channel approach for communication and interaction with customers. Therefore, to communicate with your audience most effectively, you must uncover the full potential of the data that you have already possessed or are able to possess. Data Management Platform, in turn, makes this possible.
