Blogs

360 Customer View: DataOps Drives Marketing Personalization Success

Avatar photo Cody Rich June 10th, 2020

Improve Marketing Effectiveness with DataOps

Many companies today want to improve their understanding of individual customers to provide personalized marketing messaging and offers. To do this, they need a holistic or 360-degree view of each customer but face challenges due to their complex data ecosystems. Data is siloed across lines of business and systems, sensitive customer data must be secured, and correctly matching duplicate records makes the task of building 360 customer views daunting. 

In this blog, we’ll review what Customer 360 is and define some of the common terms we hear in association with this process.  We’ll then discuss how you can implement Customer 360 through Arena, Zaloni’s award-winning DataOps platform. Finally, we’ll briefly talk about what value our customers have been able to gain from using this exciting technology.

360 Customer View

At a high-level, Customer 360 is the idea that you can create a single comprehensive picture of your Customers based on taking the data from all the different input streams you have.  You can then use this single point of reference to more accurately serve your Customers and provide personalized messaging. Now let’s define some of the common terms you often hear referenced with a 360 Customer View.

Terminology

  • Entity – A single data set; Often a table from a database
  • Matching – This is the process of grouping records that belong to the same customer. This process is made up of several sub-components
    • Predicates – These are predefined algorithms that are used to identify “matching” records. Predicates will be specific to different types of fields so that they can accurately identify matches (e.g. Address Type, Name Type)
    • Blocking – This is the initial step in matching.  The data is initially broken into blocks based on single predicates.
    • Pairing – Once data is broken into blocks, the individual blocks are compared to each other and an ML model identifies duplicates in the blocks.
  • Clustering – Once records have been identified as matching, we group them into a single “cluster” such that there is one cluster per customer.
  • Survivorship – Survivorship is the process of creating a single record from each cluster and pulling the most relevant values
  • Golden Record – The golden record is the output of survivorship and is the final step in Customer 360.  If done correctly, you should have a single golden record for each customer and it should have the most complete and accurate information pulled from all of the various entities.

Approach

360 Customer View Process
Figure 1 – 360 Customer View Process

Now that we have given some basic definitions, let’s walk through the process of generating a Customer 360 using Arena. After you have hydrated your data lake with the various entities, we can begin prepping the data for the transformations. Oftentimes you may want to filter your data to ensure that it meets certain data quality rules such as “Email must be valid” or “Phone Number can’t be null”. You may also want to apply security measures such as tokenization and masking. Arena has the ability to do all of this for you, however, when creating a Customer 360 it’s better to apply these processes afterward so that you allow the matching processing to identify as many duplicates as possible with as many fields as possible. This will result in a more complete and accurate record. So to begin we will take all our entities of interest and merge them together to create a single unified entity. (See Figure 1 – A)

Starting with the unified entity, you can begin the matching process by creating your predictive model to find all duplicate records (Figure 1 – B).  Arena allows users to easily train their models using a visual interface. Simply select the columns that you would like to use for comparison (e.g address, name, email, ssn, etc.) and then assign the appropriate predicates. 

Matching Model Configuration
Figure 2 – Matching Model Configuration

 Once you set the training criteria the model automatically begins the process of blocking and pairing to identify matches.  The data is broken into blocks as comparing hundreds of thousands or millions of individual records would be far too computationally intensive. Instead the selected columns and matching predicates are used to break the data up into blocks based on the individual predicates.  This is done for each of the columns so the same record may be in multiple blocks. The model then predictively matches a sampling size of your choosing and allows the users to provide feedback and identify if the pairings are correct.  You iterate through this feedback loop to help train the model until it has met your desired accuracy.  Once you have a trained model Arena can apply it to the entire dataset and generate the clustered entity (See Figure 1 – C).  It’s worth noting that the clustered data can provide a lot of value in itself prior to the survivorship as this entity creates the most complete view of a single customer across all of your sources.  

The final step in the golden record generation is the survivorship process (Figure 1 – D).  Using Arena’s survivorship function, you can configure different survivorship rules without having to write any code.  You instead use Arena’s interface to associate different rules for determining which value should be chosen for duplicate fields.  Some examples of this may be for Address: Choose the most recent value or Name: Choose the longest value.  These rules will vary based upon the specific fields present in your entity.  

Survivorship Rules
Figure 3 – Survivorship Rules

Once the survivorship rules are in place, Arena will orchestrate the processing of the data and create a new final entity that contains the golden records. (Figure 1 – E) Voila! You now have a golden record entity that can be used to identify all your customers and show the most accurate information available.

Customer Uses

There are numerous ways that you can use Arena’s Customer 360 that leads to increased revenue and a shorter time to delivery.  One example is a leading luxury brand that has a broad portfolio fashion lines.  Using Arena, they were able to create a centralized data hub that contained customer golden records from all of their different brands and retail stores. With this data hub their marketing team was then able to create customized marketing campaigns and promotions that increased sales and improved customer acquisitions.

Closing

So as our world continues to evolve and companies create new ways to interact with their customers, it’s the organizations that provide personalized messaging that will lead the way.  If you’re interested in implementing Customer 360 or would like to learn more about how Zaloni has helped its customers in their digital journey fill out our contact form and request a personalized demo today. 

360 customer view

about the author

Cody Rich is a Solutions Engineer at Zaloni, specializing in enterprise software and solutions sales within Data and Analytics. You may recognize him from previous leadership roles at MetiStream and QGenda.