29 December 2011

Getting started in Business Intelligence (BI) on a budget

This is a simple demonstration as to how you and your firm can get started turning the data that you already have into the information you desperately need using tools you already own. The task of turning data into information for decision-making is the essence of business intelligence (BI).

So, here we go.

Everybody has data

Everybody has data. Many companies are wallowing in data. What they are lacking is “information.”

Read my posts here and here for more about the differences between data, information and knowledge.

Quick! Take five or ten minutes to peruse the following table of data and write down everything that you see in these data to help make decisions about the firm’s future.

Data_Sale_20111227

I will give you one hint: the column identified as ‘ARPAC’ is “Average Revenues per Active Customer.”


Okay. Times up.

Hold on to your list.

Turning data into information—simply, easily, cheaply

In order to produce what follows, I used only Microsoft® Excel™ and its native ability to access databases to fetch and refresh data.

Here’s the first graph I produced:

GRAPH_SalesByMonth_20111227

This is nothing more than a simple bar graph of column “SOSales” (Sales Order Sales, as opposed to Invoiced Sales, for example) shown in the data above. I used Microsoft’s native capabilities to add a “trend line.”

By looking at this simple graph, several questions might come to mind that would bear further investigation:

  1. Why have our monthly sales dropped from just over $8 million a month to an average of about $6 million per month over these 29 months?
  2. Why or how were able to produce about $11 million in sales in July of 2008? What did we do differently? How can we build on what we learned in that experience?
  3. Is my drop in sales related to lost customers?

The next graph that I produced looked like this:

GRAPH_ActiveCustomersByMonth_20111227

This graph answered my question number three above—at least partially. Month-to-month our firm has stayed pretty steady in terms of the number of active customers served. The firm is hovering right in the 250-customers-per-month range.

On the one hand, that is good. It means the firm is steady in this regard, but it does provoke other questions that would need to be answered through further digging:

  1. We are serving about 250 customer per month, but is the same 250 customers, or do I have high turnover rates for customers?
  2. Are we constantly having to spend precious marketing resources to capture new customers, or do we have a high volume of repeat business?

But wait! If we are not loosing customers (at least in numbers), but our sales are falling off (in aggregate), what is that telling us?

GRAPH_AvgSalesPerActiveCust_20111227

The third graph I produced was “Average Sales per Active Customer” (month-to-month). This graph clearly shows that between January 2008 and May 2010, the firm’s average sale per active customer fell from about $32,000 per customer to under $25,000 per customer.

Here again, this graph immediately provides clues worthy of further, more detailed, investigation:

  1. Are these different customers buying less product? Or, are we serving pretty much the same customers, but they are just buying less from us?
  2. Either way, we should figure out why: Are they buying similar quantities, but our prices (and, perhaps, margins) have shrunk over this period? Or, are they buying smaller quantities of merchandise or services from us?
  3. Either way, we should find out why: If they are buying smaller quantities, is some of that business going to our competitors?


Next steps

As you can see, turning the data into information allows our mind to quickly digest it and move toward decision-making. In some cases—perhaps many cases, when you first start—the process will lead to further information gathering.

On the other hand, you will sometimes discover that tribal knowledge already present in your organization will help you take immediate steps to begin making more money tomorrow than you are making today. Frequently, those steps involve no investment at all. Sometimes all it take is understanding better what is happening. Other times, a simple policy change permits significant increases in Throughput and profits.

After all, isn’t that really what you want to do—not spending six-figures on a new business intelligence “solution”?


Read more here about unlocking “tribal knowledge.”


How I did it step-by-step

  1. Identify the data
  2. Build a SQL Server view or query
  3. Connect Microsoft Excel to the data
  4. Build the graphs

Total time: about 2 to 2.5 hours

2 comments:

Anonymous said...

2 to 2.5 hours to do that... I'm glad you're not consulting for us.

The insight you bring is basic. You bamboozle with your acronyms and terminology - but the applied knowledge is a thin veil of knowledge over a vast pot of waffle.

That exercise - from raw data - should take no more than 10 minutes...

Take note if this man is being paid on the clock...

RDCushing said...

Thank you for your comments, Anonymous. Actually, I meant from concept, to design, to implementation, to analysis.

But you're welcome to criticize in any way you deem fit.