Case studies

Why insight is essential and software can’t provide it

Essential: vitally important, fundamental
Insight: penetrating and often sudden understanding, as of a complex situation or problem

I once asked the inventor of a clever search engine to build a business analysis engine - a software machine into which we might pour gigabytes of business data and then ask a series of broad questions. It would report with insightful answers, or at least observations.
My colleague declined, suggesting that until artificial intelligence had been perfected computers weren’t really that clever. They couldn’t shed much light on an organism as complex as a business.
I wondered why. Surely we know that Business Intelligence is the answer to our problems. And I just wanted to enhance it a little? What are the weaknesses in a computer-driven ‘solution’?
On reflection, in principle and in practice, there seem to be several.
A computer has to rely on a model that simulates the way the business operates and reacts to stimuli, with built-in priorities, structures and behaviours. Unfortunately, such models are rigid and crude. As businesses change, flexing and evolving, the models can’t keep up.
So, what happens? Well, either there is no underlying model - diverse measurements are taken and reported, but little attempt is made to connect them - or the model reports a Red when it finds two Ambers and so on, and the only insight that emerges is that the model is wrong.
Ultimately the result is Catch-22 - in order to gain insight from the system you have to already possess the insight and tell the system. You don’t know in advance what you don’t know, so you never know it. Simple requests – please graph Germany’s sales data – however, it can do.
There are also weaknesses under the heading of lazy implementation. ‘Solutions’ start with data warehouses and software, not users and needs. Little attempt is made to understand what is really important to the organisation, and only data that is available and of a certain type is reported. Communication (the ‘reader interface’) actually displays the designer’s fondness for gimmicks, and ignorance of the principles of graphical communication.
The typical outcome is expensive failure. Long timeframes, high expenditure approved by senior executives, benefits rarely seen. So, what is the right approach?
Start at the start. With what the organisation wants to achieve, how it will do this and how it will know it is succeeding. Design concise reports that communicate this clearly and inform decision-making. Automate where appropriate – for fast, efficient, accurate data production – but let people interpret and communicate the insight.
Until, that is, we really do have artificial intelligence.

“…solving a problem simply means representing it so as to make the solution transparent” Simon (1996)
Which great decisions have your leadership team made recently on the basis of their monthly management pack?

© 2008 Biologize.