Business Intelligence

neural cells

Business Intelligence

Business Intelligence (BI) is on most corporate lips. Having looked at over 240 projects and company deployments of BI and studying hundreds of more cases through media it is difficult to ignore that most organisations fail to make the most of it, and knowing people it is difficult to argue with the majority.

Technologists argue that companies are not investing enough in BI, people using BI, tools, models, statistics, methodologies, information architecture and so on. Technologists have also a very clear agenda. Lets get organisations to spend on the latest trendy tools and toys. Looking at the share price of organisations like Tableau it is difficult to stop that, even if one wanted to. The desire for pretty picture runs through BI like breadcrumbs through Hansel and Gretel’s forest, however getting out of the forest continues to be a challenge.

So maybe it is time to rethink the information exploitation desire by returning to a few base principles. Here a few that might be helpful:

a) Information without context is just data.

b) Statistical methods exist to test hypothesis, not to make up stories.

c) A picture is worth 1000 words, both can mean nothing and be a waste your time.

d) look outwards, not just inwards, into your own organisation. Industries are disrupted, merge, and colaborate much more intensively. Being blind to that when it comes to business information exploitation would be criminal.

d) Insight without action is a bit like a parachute without gravity, a useless waste of space.

e) Running a company on historic information is like driving a car looking through a rear-view mirror.

f) Information Integration is the home of the error multiplier (90% accurate System 1 combined with 90% accurate System 2 combined with 90% Accurate system 3 leads to almost 1 in 3 actions based on inaccurate information)

g) The simplest argument is the most convincing. That means finding simple truth in complex information models, not making things simple.

h) Any BI system finds it difficult to compete with 30 years experience and the most powerful difference engine in the world is the human brain. But it also belongs to a creature of habbit. Do you have the culture to invest in anything other than your past believes?

i) All analytics are either compliance centric, or a step towards automation (I am happy to be challanged on this). Everything else is just a pretty picture that creates a lot of arguments.

We could add hundreds more, but they all drive towards the same thing: Exploiting Business Information can be tricky, and maybe a company needs to check itself before it runs of in one direction or another. So where do we go from here, now that throwing money at an issue isn’t solving it. How should BI work? where is there real payback on Information investments and how do we turn a company around to live successfully in an age where information is everything it seems? Here are some suggestions:

Viewa) Information collection processes need to be streamlined to capture as much context as possible for the information that really makes a difference to a companies ability to execute and change. So what type of information is there? Normally we just rush to naming master data style objects like Customer or Product. Lets for example consider customer: What attributes can we hold with that? location, age, sex, purchase history, browsing data, product log data from customer, social media feeds, support logs, activities, events attended, … but which one of these attributes actually make my value proposition more successful. How about their bank balance? How about their real life priorities? How about their goals and aspirations? Do you care as a customer if someone who is representing 0.0005% of your annual budget has something to say? A customer centric organisation holds over 1,000 characteristics per customer, but only of 3% companies looked at have done a customer process analysis against these attributes to find out which attributes are indicating a readiness to buy, and how these signals can help segment customers and change corporate customer engagement strategy. That means 97% of organisations are playing with data, making no real use of information.

which leads us seamlessly to

helpb) Statistical analysis is there to help us test a hypothesis. companies over 1,000 employees invest 5% of their revenue in IT. Testing this hypothesis to a statistical fact would require only 0.5% of the sample to fall significantly out of that investment average. Believe me when I tell you that this “fact” does not hold up to statistical significance testing. It is therefore an irrelevant point when selling IT to companies. however the actual IT spend of an organisation may tell you something if IT spend correlates for example to profit as well as revenue in combination with customer retention/repeat spend. You can think of it in terms of Apples and Oranges. Apples and oranges are Fruit, But eating Fruit doesn’t equal a healthy diet. On average we should eat 5-7 Fruit and Veg a day. If I combine this with 10 Mars Bars per day I am likely to have issues, irrespective of the fruit. If  only eat 7 fruit a day, I might not have a social life (gas?). In short using statistics is critical to investigate you data, find significance and then test a hypothesis.

A word of caution: Using statistics in isolation may help in optimising a process, but it may end up negatively effecting everything else.

Viewc) Master data is still the king. nowadays you have to add business context into it as well (what store stock return figure am I looking at (ready for resell, ready for destruction, ready for repair,…)) You have hard master data (customer, product,…) and you have soft master data (hirarchies, Taxonomies,…) nd you have configuration data. All describe slow moving dimensions but all have a clear single meaning. Manage that tightly, and well over time, and score it in terms of quality, add it to every report where it is used and you get your funding quickly to address the qualitative dimension of you core data.

There are more things worth concidering. Use of intelligence and so on. What about AI, what about virtual integration. These are all great for future acticles. But these are the most important once in my view. Get them right, combine them with strong leadership and this will fly.

Thoughts?