Complete Information System or Complete Analysis?

The 360* view. How many times you must have come across this promise made by umpteen vendors? Assuming that this is possible, the obvious question is how this can be generated? More important, the question is how this can be generated when there is a proliferation of software within an enterprise or organization not to mention a potentially vast number of relevant external reference points. Let us step back a while to understand what is involved.

Imagine the time when information systems were silos and didn’t ‘talk to one another’. Most organizations had multiple systems each serving some purpose and thus containing some information. And yet, organizations had to create and circulate MIS to all decision makers. (MIS as a function meant something else’; today, as a subject at Masters, it is quite different). I just wonder how this would have been done! Must have been a tough task. I would love to talk to people who worked during such times. Any analysis would have been more the mind at work than any system.

The environment is better today, although interoperability remains a problem. In my opinion, the decisive change came via ‘interfaces’ and ‘interface management’. Later, the advent of XML made a huge difference, when web-enabled applications had begun to gain significant ground. Some built businesses around xml-enabling applications to create the possibility for publishing cohesive, intelligible information that suggested possible courses of action or intervention. While we are better off relative to the initial stages of IT, we have a different problem. We have many different software each catering to specific needs. Each generates data that has the potential to throw light on some part of the activities.

What is important to remember is there is no one complete information system. Often analysis of enterprise data might be indicative of certain behavior but the insight into that will lie elsewhere. We need to understand how data that is captured within an organization may be related to external variables. Whatever the business. Consider someone in the business of investment. Let us say they notice, during a certain period, a noticeable spike in demand for a certain asset class. The user will have to traverse outside the terminal to find why some investors have found an attraction to that asset class. I never tire of saying this – the intelligence of a system could well lie outside it. The internal data only shows an increase (or decrease) in the demand for a certain kind of asset but cannot explain why this has taken place. The answer, if there is one, lies outside, but demands domain knowledge. It is the combination of internal data together with external data that offers a ‘complete’ picture.

In some systems, intelligence could simply be the pattern. Consider the fast moving consumer goods business, where the most crucial data is the ‘average repurchase period’, the time taken by customers, on average, to replenish their stocks. Or what good are bought together or on specific occasions. This can be found out through careful study of the purchasing patterns of all customers with significant variations if any. And understanding sensitivity to price points. Typically, this is found out through specific schemes and their successes. For any volume business, this is a vital piece of intelligence. The entire business’ success rests on understanding these patterns.

For instance, a senior manager of Deutsche Bank while discussing the demand for structured credit derivatives (with The Banker) commented that the spurt in demand was related to the tightening of credit spreads and that once spreads relaxed, more traditional credit products would find their way back into the market. The point is that the information on credit spreads was not in the same database. An insulated analysis of data relating to credit derivatives would have missed the crucial business connection. That’s experience talking! Can we have a system that incorporates context?

A few years ago, a leading telecom company in India admitted that it had no problems collecting information about its subscribers, but using it to improve the customer experience was proving more of a challenge. “If a customer’s call drops at the same part of the network, at the same time every day, we should have the data on that part of the network, we should have the data on that customer. The customer probably calls the customer service line so we should have that data as well, but it’s not that easy”. The problem is that all of this data is stored in separate silos, and the challenge is “how we can be proactive to get the right data to the quality we want”. The aim is to tackle problems before they necessitate a call to the helpline, which is particularly challenging in a market where a lot of customers buy mobile access from agents operating from small kiosks, and therefore rely on the helpline as their primary contact point with the service provider. If a customer does end up calling, the strategy should be to “turn it from dealing with a frustrated customer into an upsell opportunity”.

Ultimately, any organization, business or otherwise, is looking for ways to improve its performance and should evaluate every tool and technique only from that perspective and not get caught in taxonomy. A small company could use Excel to generate the insights it needs to improve considering Excel’s potential to analyze data, based on its own available tools and on specific queries. Pivot tables can reveal so much, if you knew how to use them. An analysis of the accounting information system can itself reveal so much intelligence. Vendor analysis, Procurement lead time, ageing schedule of debtors etc can all be extracted from the accounting system. In fact, much before ‘analytics’ became a fashionable term, intelligent users of accounting systems always evaluated them based on how good their decision support systems was, what techniques did it have to help you extract intelligence from information. I have written about this elsewhere – of a time when analysis was integral to an information system.

Takeaways

A 360* view is not possible from an information system alone

Data that is captured within may be related to external variables

Objective must be to find intelligence not use technology

Look always for the ‘Why’