Business Intelligence and “Tribal Knowledge” – Part 2

In Part 1 of this series, we were talking about a firm that had identified that the thing that was keeping them from increasing Throughput – from making more money tomorrow than they were making today – was understanding their customers and the other participants in the decision-making process better. I also pointed out that this firm already had some data available to them that could be used to begin the process of understanding their customers better. They had, of course, their historical sales data. But this firm also had available to them an independent database that contained some additional demographic data about their customers and prospects that could be correlated with their own sales data.

I closed Part 1 by suggesting that they could employ these available data to begin exploring relationships such as:

• Sales by salesperson
  
• Sales by salesperson by geography (e.g., city, state, region)
  
• Sales by salesperson by demography (e.g., size of school or school district)
  
• Sales by product line by geography
  
• Sales by product line by demography
  
• Sales by salesperson by product line
  
• Sales by salesperson by product line by geography
  
• Sales by salesperson by product line by demography
  

There are other data elements (dimensions) available to virtually every firm that we are not including here. For example, if you introduce the additional "time" dimension, it may be easy to spot trends over time – e.g., salespersons, regions, or demographic groups where sales are growing or decreasing over time.

This is an example of what business intelligence practitioners call "cubing the data." The data is summarized by various "dimension." In the example above, the sales data is being summarized and the "dimensions" are:

1. Salesperson
   
2. Geography
   1. City
      
   2. State
      
   3. Region
      
3. Demography
   1. Size of school (number of students)
      
   2. Size of school district (number of students)
      
   3. Teacher/student ratio
      

As I said, all of this can be done using low-cost tools available to almost every small-to-mid-sized business and already on the desktop of almost every computer. Microsoft Excel, especially Office 2007 and later versions, is capable of digesting a large set of data within its own operating context. However, if you or your firm has a Standard Query Language (SQL) server and these data reside in a relational database (such as Microsoft SQL Server, especially SQL Server 2005 and later), you have even more relatively low-cost tools to manipulate and digest even larger data sets. SQL Server 2005 and later is even capable of calculating and summarizing data cubes on the fly. These pre-digested data may then be presented to Excel as a presentation tool and user-interface.

Introducing "tribal knowledge"

So what is keeping companies from leveraging the data that they already have in order to use the insights discovered through such analyses? Generally, in small-to-mid-sized businesses I find the following factors are holding them back:

1. Uncertainties regarding the value – I have to put this one at the very top of the list for one simple reason: If executives and managers in the firms were convinced that discovering new factors about their marketplace – market segmentation – would help them make more money tomorrow than they are making today, they would find a way to get it done.
   
2. Uncertainties regarding the costs – Sadly, the business intelligence community itself has much to do with making small businesses wary of the costs moving into the realm of business intelligence. Many who make their money by selling and implementing business intelligence tools want you to believe that is not possible to make real gains and reap significant business benefits without investing in expensive business intelligence software and spending lots of time, energy and money to build expensive data warehouses and, perhaps, hundreds or even thousands of "cubes." This is simply not the case, but it is frequently the belief.
   
3. Uncertainties about how to get started – Again, in part to the pseudo-mystique surrounding the world of "business intelligence," many executives and managers do not feel that they "have what it takes" to get started benefiting from understanding their customers and marketplace better by leveraging the data they have been collecting in their ERP systems for years. There are simple ways to get started and one can always make the leap to more sophisticated business intelligence applications when conditions warrant.
   

But, wait!

So far in our discussions I have intentionally left a tacit implication on the table. That implication is the one that drives far too many executives and managers in companies of all sizes, and it is this: What is valuable and can be leveraged in "business intelligence" is found in our data systems and the data stored or collected.

This is very far from true!

Some of the most important contributions to making computer-based "business intelligence" valuable do not come from the data, nor from the software. These valuable contributions come from the people that have worked in your enterprise year after year. Your people know things about your customers, your prospects, your products, your industry and your marketplace. I call this kind of knowledge held within a business enterprise "tribal knowledge."

Now, tribal knowledge in every organization extends well beyond the examples I will suggest in this series, but I think you will begin to see just how adding tribal knowledge into the blend with the data you have available to you extends the power of business intelligence and may lead to truly valuable breakthrough thinking.

Suppose that in analyzing sales data currently available, they looked at the data summarized in a certain way and the graph looked like the following figure:

The questions that ought to be asked when looking at such a data summarization should be along these lines:

• Why are sales in category 'A' five times better than sales in category 'E'?
  
• What can we learn from what we do to get the results in category 'A' in order to apply it to the other categories?
  

Now, let me bring this down to more practical examples:

• Categories are product lines: What factors make Product Line A perform so well? Do we sell it differently than Product Line E? Do we promote it differently? Do we sell it to different kinds of customers? If so, what are the differences between the kinds of customers? How can we apply what we know about how we sell Product Line A to improve results for Product Line E?
  
• Categories are salespersons: What does 'A' do to get results that 'E' does not? Are these results simply differences by sales territory? Are there demographic differences in 'A's customer list from the customer lists of the other salespeople?
  

Naturally, this of questioning can go on and on, limited only by the management team's ability to think of the "right" questions to ask. Some of the questions can be answered using the data and re-summarizing it in a different way. For example, to answer the question, "Are these differences [between salesperson results] simply differences by sales territory?" it may be necessary to re-summarize the data by sales territory. However, if salespersons and sales territories are synchronous and exclusive, then one might need to compare similar but broader territorial results to see if a pattern exists. (For example, if the salesperson assigned to Washington State is Category A, then one might compare results for other West Coast states to see if they are similarly high even though different salespersons are assigned to these territories.)

The basic point, however, is that the people involved in your organization are carrying about with them "tribal knowledge" that can help you and your management team discover new ways to segment your market and increase Throughput.

[To be continued]

©2010 Richard D. Cushing