"There were 5 exabytes of information created between the dawn of civilization through 2003, but that much information is now created every 2 days"

                 - Eric Schmidt

“If markets are to be segmented and cultivated, they must meet certain requirements. Segments must be Measurable, Substantial, Accessible, Differentiable, and Actionable.”          

                - Philip Kotler

Digital platforms of today are generating humungous amounts of data. We need reliable platforms which helps process the huge amounts of data and derive insights from them. DMP is one such platform to collect, store, organise and segment the data. The segments created are used for targeting.

In this article, we will cover how to create a DMP, followed by the advantages of a DMP.

How to create a DMP

The following are the steps involved:-

Data collection

Data cleaning and enrichment

Profile merging

Data segmentation and Taxonomy

Audience creation

Audience activation

Data Collection

Data from different sources are required to be collected for creating a DMP. These data sources are categorized into First, second and third-party data sources.

First-party data

This is the data directly collected by digital product owners. This is the most valuable data for any business enterprise.

Second-party data

The data collected by the first-party is provided to a partner for use. For example, you browse book reviews from librarything.com. The data collected by LibraryThing if provided to amazon.in, then it is second-party data for Amazon.

Third-party data

The data obtained from data brokers and other data vendors are known as third-party data.

DMP should be capable to collect data from first, second and third-party data sources.

Data Cleaning and Enrichment

Data collected from multiple sources is never going to be readily usable. There will be junk values, missing data, values in a column will exist in multiple different formats among other things. The schema of the data from different sources will not be the same. 

The cleaning of data needs to be undertaken along with converting the schema of different input data sources to the required target schema. The cleaned data is further enriched with features from the first, second and third-party data sources.  

Profile Merging

As data is collected from multiple sources, the same user will be tagged with a different profile ID by the different data sources. The same users from different data sources need to be identified and their profiles should be merged. The profile needs to be updated with new events and the latest data if it already exists in DMP. If the data captured for a profile is not there in DMP, it needs to be added. Once the profiles are updated or new profiles added, the audiences need to be recreated. If profiles are not merged, we will have multiple entries for a user in the DMP and the efficiency of targeting will be reduced. 

Data Segmentation and Taxonomy

The cleaned and enriched data is segmented and taxonomy created. Segmentation is finding the patterns in data and grouping them together for easier access. For example, website visitors from India. Hierarchies of segments are also created in this step. Taxonomy (naming) needs to be decided and fixed for the user groups and hierarchy. Care needs to be taken to avoid duplication of segments and taxonomy.

Audience Creation

Audience creation is the process of selecting a combination of multiple data segments for targeting purposes. For example, all the users from 'India', in the age bracket '18-25' and who loves to 'travel'. This is the most important step in the DMP

Audience Activation

The audience created in the previous step is sent to Demand-side platforms (DSP) or Supply-side platforms (SSP) for syncing. Once the audience segments are available in DSP, the advertisers can target the specific audience segment with a high probability of a conversion. 

Analytics and reporting

UI is required for advertisers to measure the efficacy of the campaigns. They need to measure the uplift in the conversion due to targeting an audience segment. The UI will also have the capability to do campaign configurations.

Advantages of DMP

A DMP helps advertisers to improve the performance of their marketing campaigns by targeting the right customers. 

Audience Extension - It is a process that allows advertisers to reach out to the publisher's audience outside their websites. Syncing the DMP with DSP and SSP will allow the advertisers to target the publisher's audience outside of their website. This will help the advertisers to reach the right audience and also help their retargeting efforts. The publisher will benefit by optimizing the ad space and if required by also keeping the website free from advertisements.

Content personalization and product recommendations and Look-alike modelling to improve the conversion rates of acquiring customers are some important advantages of DMP among several others.

To summarize, DMP is a platform to collect the data from various sources, clean and enrich the data, merge the profiles of the same user, create segments, chose different segments to create an audience and finally activating the audience for targeting. DMP helps in improving the efficiency of campaigns by targeting the right users, audience extension, content personalization and look-alike modelling.

In my next post, I will cover what is a customer data platform (CDP), why is it required, how is it different from a DMP, advantages of CDP etc. Let me know if you want me to include anything specific which you would love to see in my next article.

Let me know your opinion on this piece or anything which I missed or which you would like to share about DMP from your experience.