The most direct application of the use of Time Variant data to study cause and effect in the enterprise applies to the four P’s of marketing—Product, Price, Promotion, and Placement.

• Product—Customers may choose to accept a product and to reject another product. Like Goldilocks in the home of the three bears, one product is too hot, another is too cold, but the third product is just right. Time Variant data in the “three bears data warehouse” will show that the “hot product” and “cold product” were rejected, while the “medium product” was accepted. For a single product, the variations over time may include changes in the label, size, package type, or formula. The ability to track these changes in label, size, package type, or formula and customer responses to each of them individually can be achieved by the use of Time Variant data in a data warehouse.

• Price—Customers can reject or accept a product based solely on its price. The search for the right price point is paramount to every enterprise that can vary the price of its products. The ability to track changes in price and customer responses to each price point individually can be achieved by the use of Time Variant data in a data warehouse.

• Promotion—Promotional advertisements influence customer buying patterns...definitely...probably...we hope...we need to be sure. Promotional advertising can include a wide gamut of possibilities from radio, TV, billboards, endorsement, free gifts, price reductions, and a person in a costume waving at cars passing by. Which of these methods works? Which of these fails? Of the four P’s of marketing, Promotion is the one least able to assert a direct connection between a campaign or advertisement and customer response. Maybe a prior campaign or advertisement already influenced a customer who only now has the necessary funds. Regardless, it is worthwhile to connect the Promotion occurring at the time of a transaction. Time Variant data in a data warehouse can achieve that connection between a Promotion and a transaction.

• Placement—The method by which product is placed in front of a customer can be directly associated with that customer’s decision to purchase the product. If the customer intended to purchase the product prior to the transaction, then that customer went to the place where the product was found. If the customer coincidentally happened to be in the place where the product was found and then decided to purchase the product, the product was purchased in that place. Either way, the product, customer, and place are all directly linked by a transaction.

An enterprise can have many other connections between enterprise objects and their properties or attributes and transactions beyond the four P’s of marketing.

• Manufacturing Foreman—A manufacturing foreman may be linked by the products manufactured during his/her shift to a reduction in product returns and rework, while another foreman may be linked to an increase in product returns and rework.

• Restaurant Manager—A restaurant manager may be linked by the meals served during his/her shift to an increase in the sale of desserts, while another restaurant manager may be linked to an increase in the sale of wine.

• Service Technician—A service technician may be linked by work orders to an increase in repeat customers, while another service technician may be linked to an increase in complaints.

The Time Variant connection between enterprise objects and transactions allows a data warehouse user to query the data warehouse looking for all events associated with a specific manufacturing foreman. The Time Variant result set will include all the transactions associated with that manufacturing foreman, regardless of the location, time, or manufacturing process for which that person was the manufacturing foreman. Likewise, all transactions associated with a restaurant manager can be found regardless of which restaurant that manager has managed. The same is true for any enterprise object directly associated with transactions via Time Variant data.

The correlation between events or transactions and enterprise objects and their properties or attributes is based on the moment in time the transaction occurred. This correlation can be similar to the correlations found in Market Basket Analysis. Customer behavior may be completely unchanged by an enterprise object and its properties or attributes. Conversely, customer behavior may vary based on another enterprise object and its properties or attributes. Time Variant data in a data warehouse allows an analyst to study these correlations based on the enterprise objects and their properties or attributes in effect at the moment of each individual transaction.

Causal Analysis is an analytic discipline that has a wider scope than just the enterprise. As such, the cause and effect between enterprise objects and their properties or attributes and the results that occur within the enterprise is not really comprehensive enough to be considered Causal Analysis. The idea behind Causal Analysis is to identify and understand the conditions that caused an outcome. Thus far, this book has not proposed that Market Basket Analysis or Time Variant data can identify what caused an event to occur, but rather those enterprise objects and their properties or attributes that correlate with an event. For that reason, the discipline of Causal Analysis goes beyond the scope and data of Time Variant data.

Causal Analysis can record and include environmental, governmental, competitive, and economic conditions that actually caused an event to occur. A data warehouse may be able to expand its scope to include these data points and may be used to facilitate Causal Analysis. However, Causal Analysis in all its meaning is a discussion for another book. The scope and context of this discussion of Time Variant data is limited to enterprise transactions and enterprise objects. As such, the set of enterprise transactions and enterprise objects does not include the hurricanes, tax breaks, competitor actions, or school activities that can affect the enterprise.

For every rule there is an exception. Time Variant data in a data warehouse is no exception to that rule. Every enterprise has anecdotes wherein a manager benefited from the preceding manager. The preceding manager had wonderful hiring practices, productive work policies, and therefore a very productive workforce. The current manager required two whole years to dismantle that workforce into a dysfunctional and unproductive team. But, for those two years, the current manager looked very good on paper. The reverse has also been true. A new manager of a dysfunctional and unproductive team will look abysmal on paper until the results of the previous ineffectual manager can be reversed. Until then, the current manager will look bad on paper.

That sort of lagging correlation between the enterprise and its outcomes happens frequently and ubiquitously. For that reason, the presence of Time Variant data in a data warehouse is not an occasion for enterprise analysts to run their Time Variant reports and blindly draw conclusions based on the results listed in the reports. Rather, Time Variant data is a tool within a data warehouse. Like all tools, Time Variant data must be used intelligently and with an understanding and intuition of the enterprise. While Time Variant data can be very useful at finding vendors who deliver inferior product, managers who deliver productive teams, and processes that do or don’t create products that will eventually require rework, Time Variant data requires an insightful analyst to intelligently leverage its connections and correlations.

Exception Reporting may be the use of Time Variant data that delivers the highest ROI. A data warehouse can be used to monitor enterprise and customer behaviors across extended periods of time. When those enterprise or customer behaviors change, to the point of being exceptional, a BI Report can be written to notice such exceptions. When these exceptions occur, that same report can be written to find the enterprise objects, properties, or attributes that changed prior to the exceptional behavior. If the enterprise and customer behaviors never changed, such a report would remain empty. However, when exceptional behavior occurs, such a report would notice the exception and the difference in enterprise objects, properties, and attributes that preceded the exceptional behavior.

Such a reporting mechanism is helpful when exceptional behavior is expected as well as when exceptional behavior is not expected. Either way, an exception report that leverages Time Variant data allows the users of a data warehouse to notice changes in the behaviors of the enterprise and its customers and the changes in the enterprise that preceded them. The ROI of Time Variant data is the ability to correlate changes in the behaviors of the enterprise or its customers and preceding changes in the enterprise.