Create Intelligent Dimensions by Observation

To identify intelligent dimensions try to understand how the decision maker uses the information which a BI report gives. Identify what exactly he is looking for in these reports. See how he interacts with the report. These observations give you a clue to an intelligent dimension.

Also, remember that the human mind will keep searching for new meanings in data. Hence this process is always iterative. You will have to keep observing ways in which decision maker actually reads the reports, get your clues and build the intelligent dimensions. After some time the end-user will know that you can build such intelligent dimensions, and then you will get your ideas right away from him.

Let’s take the AR report which we discussed above. Some of the intelligent dimensions which a BI developer can build are :

• Aging Buckets: On the basis of the due date and the reference date, age of each invoice can be determined, and invoice can be grouped in aging buckets. In a BI scenario, you can however think of multiple aging buckets in the same report. For example one aging bucket could be just <180 days overdue and > 180 days overdue. Within this aging bucket the user could see <30, <60, <90, <120, <180 days overdue.>
• Aging Categories: You could also classify the dues as Not Due, Current Dues and Overdues to give a more meaningful analysis. ‘Not Due’ identifies all the invoices whose due date has not yet passed, ‘Current Due’ identify all those invoices whose due date has passed but are within an acceptable number of days and ‘Overdues’ are those which have pass the acceptable number of days limit after passing due date.

The above dimensions have been traditionally used for analyzing AR.

The following dimensions can be added to give an extra edge :

• Impending Overdues: These are all invoices which are at present in the ‘Current Due’ category, but which will move into overdue category if not collected by the end of the month.
• Impending Critical Overdues : If your critical overdue limit is say 180 days, then identify all the invoices which are at present less than 180 days, but which will move into > 180 days by the end of the month is an intelligent dimension.
• Notional Interest : Interest loss due to delay in collection from the due date to the current date at a specified interest percentage (which should be the cost of capital for the business) can greatly help in identifying the seriousness of the problem. The user can easily sort the invoices in descending order of interest loss, and identify such invoices and take action for the same.

These intelligent dimensions make the reports more ‘actionable’.

The receivables are classified into Current Dues and Past Dues. Further the current dues are classified into regular dues and potential past dues. Similarly the past dues are classified into regular past dues, critical dues and potential critical dues. The user now does not have to search for information. He can quickly put a filter and drill down to the customer and the relevant invoices in order to take his action.

Combine Data from Related Functional Areas

Sometimes juxtaposing data from related functional areas adds a considerable amount of intelligence to the area under consideration. The BI developer should try to find out where the data of the current area is getting impacted from.

For example in the above example, AR data is impacted by Sales Data as well as Treasury Data. So if data pertaining to sales and money receipts from customers can be shown along with the customers AR balances, it can bring greater clarity. Similarly if a monthly trend of sales, receipts and AR balances can be plotted, it can help the end user define credit policy for individual customers. See Report 1.2 of how data from multiple functional areas is plotted together.

In the Report 1.2, you can see the data of AR, Sales and Receipts plotted in a monthly trend. Various metrics like DSO (Daily sales outstanding), weighted average credit period etc. also plotted. The user can now select a customer, check his credit pattern over the last 12 months, and decide a suitable credit policy for him.

Combine traditionally different reports into one

In the traditional method of reporting, generally one report format addressed one thought. For example :

• A report would classify Inventory as per ABC Class based on inventory value.
• Another report (based on transactions) would identify inventory as per its movement into FSN-Fast, Slow and Non Moving inventory.
• A third report would identify Inventory Turns.

By combining these three traditional reports into a single BI report with separate dimensions for ABC, FSN and Inventory Turns, you can give the decision make a fantastic view into inventory data. The user can study one within the other. Using this report, with a few clicks he can locate a problem area. The report becomes very actionable.

In Report 1.3 you are able to see a summary of FSN class within ABC class. You can then drill down to the item level to identify a problem area.

Creating actionable reports

The BI reports should be actionable, and for this the BI developer should identify and build intelligent dimensions. This will be the key factor in acceptability of the BI Reporting system. In our practice we spend as much as 50% of the time of a senior consultant in discussing with users their requirements and observing how they use the BI Report in order to get clues and ideas for developing intelligent dimensions.

By Sanjay Shah
Express Computer