Business Intelligence: Are you overlooking Lost Gold in your data?

How much value could you extract from your existing data?

How much would it cost to improve on/increase the amount of your data?

One of the interesting observations when dealing with distributors is the vast amount of data collected on a daily basis, yet not really put to the best use.

We have had countless conversations on all levels in this industry about data, data quality, price of collection, value of statistics etc. Whenever we ask about this, the typical reaction can of course be one of many, but a few do stand out:

• “Oh no, we have neither the time nor the resources to collect a lot of extra data”
• “We trust our employees, and it is our policy to give them the flexibility to do what needs to be done”
• “We use our common sense; our business is too complex to put on formulae”

These quite common reactions reveal, at least to us, to some degree a healthy sense of proportion and reality. They also, however, reveal a lack of knowledge about how to collect data, how to enrich them, and how to present and use them; and perhaps also a lack of understanding of how much value actually could be extracted. It is our observation that tradition plays a strong role here.

Many companies in the distribution industry have simply never been accustomed to dealing with data in the light of Business Intelligence. This, by the way, is not exclusive for the distribution business by any stretch of the imagination. Some focus a lot on Business Intelligence. A lot, focus somewhat and a few hardly focus on it at all. A frequent assumption is “It costs a fortune, and the return is questionable”.

Pressed for an opinion on the matter, however, nobody really wants to be at the wrong end of that spectrum. So… what can we do with what we have already? Can we do better without heavy investments?

I will write more about the actual data in my next article on Intelligence and Decision Making – which is already in the pipeline. For now, I just wish to convey some basics of Data collection and Business Intelligence. There is plenty of literature on this subject, I will refrain from listing them here, just list a few points of general consensus worth noting:

• Consistency: Your data must be consistent. If you have too many voids in your data flow, then you should work on THAT before you try and derive too much intelligence. You should work on THAT before you proceed too far with higher ambitions. You should not try to conclude anything on too sparse a data sample, for two main reasons. Firstly, you may discover, that the “holes” in the data stream actually conceal structural problems in themselves. Secondly, statistics just work better in general the more of it you have.

• Credibility: It feels almost superfluous to mention it but spending time on analysing the validity of your data once in a while is often time well spent. Do you measure/register correctly?

• Comparability: Another fundamental issue is that of comparability. We all know the meaning of the phrase about comparing apples with pears; yet often enough we see violations of this basic rule for statistical analyses. What often happens is that people collect what they CAN rather than what they SHOULD. The notion that “something is better than nothing” is often proved counterproductive. One could polemically say: “It is better to be blind, than to be systematically led in the wrong direction.”

• Natural actions: This is crucial for any Business Intelligence set-up. Your data collection should not rely on “somebody registering something, somewhere” on a regular basis. It has been proven repeatedly, that these systems always break down sooner or later… usually sooner. Always collect your data in connection with an activity that must be done any way – or – collect it in a completely automatised manner. “Extra registrations beyond that? – forget it!”

• Enrichment: Ever so often you will see data collected from which you cannot really derive anything directly. In order to have a useful analysis, you will need to enrich the data by subjecting them to various calculations and algorithms. If, for example, you wish to derive the cost per address visited on a distribution route from the total cost of a route, you will need to apply quite complicated calculations to your gps- and activity-registrations. I will compose an article about this subject as well, soon.

• Correlation: When you correlate data from diverse sources, you are moving in the direction of what we call BALANCED SCORECARD: This is about elevating Business Intelligence to a level that enables you to make predictions. It is a discipline of its own within the more general expression Business Intelligence. In short, by correlating the right data, you should be able to see patterns in your business that are otherwise impossible to spot – before it is too late that is. If you can pinpoint the right observations to correlate, you can find tremendous value; but this can also be a bit of a jungle.

• Presentation: The last point on this short list as introduction to BI is that of presentation. It is not only a matter of looking at columns vs. pie charts. Unless you have hired an army of autists to go through endless lists of data, you are well advised to think hard and well about how you present your data.

It is beyond any shadow of doubt that companies who master the above are much more competitive than those who do not. Our advice is that you pay attention your data gold digging.