Menu
-
INDUSTRY EXPERT: CAPABILITIES NEEDED TO BUILD UNIFIED CUSTOMER PROFILES
Author: David M. Raab is Founder and CEO of the Customer Data Platform Institute. He has many years’ experience as a marketing technology consultant and industry analyst. Visit www.cdpinstitute.org for more information.
Complete accurate customer profiles are the lifeblood of your MarTech system. Many companies purchase Customer Data Platform software to assemble these profiles, while other companies build them with tools designed for other purposes or with systems they’ve created on their own.
Regardless of which path you choose, creating and deploying profiles, requires the same set of tasks. These are links in a chain: if any link is missing, the chain cannot serve its intended purpose. So, it’s essential to have them all.
-
Use the following list as a reference to ensure that you have developed a complete customer profile capability. You’ll see that each item relates to a specific type of software.
Capture:
Data must be captured from websites and other systems where customer activities take place but are not otherwise recorded. Tag management software such as Google Analytics, Adobe Experience Manager, and Tealium provide this function.
Ingest:
Data must be loaded from systems that originally create it, such as order processing, CRM, and customer service. This requires extracting the data from the original system and pushing it into the customer profile system. Pipeline software such as Fivetran, Spark, and Airflow provide this function.
Clean:
Data must be checked for accuracy, completeness, consistency, and other quality considerations, and corrected as necessary. Data quality software such as Informatica, Tibco and Talend offers this.
Store:
Data must be stored for further processing and use. The storage system must accommodate all data types: structured (such as transactions), semi-structured (such as web logs), and unstructured (such as text). Common platforms to store customer profiles include Snowflake, Databricks, and Google BigQuery.
Link:
All data related to the same customer must be given a common identifier, even though the original identifiers might have been different for different sources: an email address, a browser cookie, a mailing address, etc. This is called identity resolution, or matching. Vendors
include Infutor, Merkle, and Tapad. Note: These particular services may not be available in the Indian market.
Aggregate:
Data for each customer is assembled into the customer profile. This includes calculated values such as aggregate lifetime purchases, time since last website visit, and predictive model scores. This aggregation can be managed by master data management systems such as Resulticks, Aqfer, Reltio, and Tamr.
Share:
The enriched profiles are sent to other systems, either in a batch process or on demand. This is the function of reverse ETL software including Hightouch, Census, and Rudderstack.
Integrate:
Selected data about an individual customer is used in processes that run across multiple systems. This is done by orchestration systems or integration platforms. Examples include Mulesoft, Zapier, and SnapLogic.
Segment:
Sets of customers who meet specific criteria are extracted from the profile system. Users define the selection criteria using segmentation software, or broader analytics software with segmentation capabilities. These include SAS, Tableau, Power BI, Qlik, Sisense.
-
Every item on this list is a complicated process by itself. You must look carefully at your own business requirements to understand exactly what capabilities you will need. This is a substantial task, but it’s unavoidable if you are to build a suitable solution. Requirements are needed regardless of whether you buy a single CDP that offers all these functions, assemble a collection of separate tools, or build your own system.
If you do choose to connect separate systems, you face the additional task of analyzing their integration features to ensure they can work with each other. This goes beyond understanding how they connect to ensuring that they can share data as quickly and in the volumes
you require. You must also plan to allocate adequate resources to set up and maintain these connections. Continuous maintenance is essential, as vendors often make changes that improve their products but can change the way they connect.
The best approach to customer profile creation will depend on your requirements, your resources, and how quickly you need to put a solution in place. Making the right choice is not easy, but fixing a wrong choice is even harder. So take your time, choose carefully, and then share the full benefit of complete customer profiles throughout your MarTech system.
