Today, a quick write on Master Data. What is it, and why is understanding it important to any business.
Every business performs some type of transaction. That transaction is done by customers, vendors and performed with respect to other objects, let’s say widgets. So, each time a widget is sold, or moves from one location to another in the example of production location to distribution location, each of the actors is actually part of the Master Data, while the transaction is not.
In that example, the Master Data is the actors, or objects which transact the business. The transaction itself becomes a Measure in the BI solution, while the actors become Dimensions.
Why is controlling the data that describes the actors critical? It is this data that is the truly actionable data. Let’s take an example and maybe it’ll become more clear. Some years ago Target stores embarked on a mission to understand their customers. They began with capturing customer data as they could, using Credit Card data of their own and frequent shopper information. They analyzed the purchases, or transactions, the actors were making. What did they buy? Essentially, a portion of the project was designed to “target” prospective mothers. That portion of the project focused on the purchasing habits of prospective new moms as having been registered for baby showers. In that data, they were able to identify the unique combinations of products purchased by expectant moms. The customer, or new mom, is actually Master Data in the equation. What they were buying, the transactions, gave clues as to the current state of the new mom, her potential due date, etc. With careful analysis of the transactions, they were able to infer certain things about the actor, and thereby drive sales with promotions directed towards those customers. In this case, coupons for diapers and formula. All businesses that sell things are doing this by way of frequent shopper cards, used to enact a discount, retailers are able to infer many things about their customers. Age, health, weight through eating habits, preference towards certain ethnic foods, all contribute to a profile of the customers that patronize a business.
Rules around Master Data (MD) are as critical as the data itself. Corrupt or incorrect MD will lead to incorrect understanding of customers, suppliers, widgets, and shippers. Careful steps must be taken to create and enforce policies that preserve the integrity of such data. The concept of Master Data Governance was born out of this need.
Master Data Governance is a decision made, prior to ingesting any data, that is used to determine the UPDATE, INSERT, and DELETE rules for each entity of data. Concepts like FILO, FIFO, and conflict resolutions have to be addressed in advance of data entering the system. These policies can become complex, in that each system can read the MD but only certain channels can update that data, and then only certain portions of the data. Let’s take an example.
In the previous example, a retailer was profiling customers based on purchasing habits. A customer in their system has particular data, name and address for example. If the customer moves, the address is updated with maybe a store CC record. That channel of a store credit card, and the credit card system which issues billings to the customer should be accurate and up to date. The MD can be updated based on that system, because it’s assumed to be accurate and current. Now, the retailer knows the customer has moved and can put that customer into their “recently moved” customers campaign which sends coupons for cleaning supplies, shelving units, and wide screen TV’s. But, if for some reason the address for a customer is updated in error, maybe from National Change of Address and it is inaccurate, then the retailer has sent coupons and offers to possibly an non-customer or off into space. That can skew or nullify a specific marketing campaign, wasting effort and time / money.
Anytime a Business Intelligence solution is being conceived, a Master Data solution should be considered prior. The identification of the business actors, the governance of the data used to describe those entities becomes paramount to effectively running a business, and providing value to customers.