Monday 22 October 2012

The Politics of Business Intelligence



Do you have political views? Does your stomach knot when you consider a presidential term of a) Romney or b) Obama (delete as appropriate)? Do you confine your views to Politics with a capital P or do they permeate your work, family and social life? Some say everything we do is political, so let’s consider the politics of business intelligence (BI).

A fundamental spectrum of BI is that of centralisation. At the highly centralised end of this spectrum we have a strong IT team defining and delivering reports to end users, with business users passive recipients of this information. At the decentralised end of the spectrum we have the business community within a given organisation being (almost) completely self-sufficient in defining and discovering the information they need. I say almost because there will always be a place for IT in managing the infrastructure of BI systems. This spectrum parallels political choice between centralism to localism. Whilst localism is currently the more popular of the two, regarded as delivering services relevant to the communities that use them, it is also understood to be harder to manage, relying on the skills and integrity of a far wider number of people and requiring more sophisticated control structures to enable equality whilst also promoting diversity. This directly corresponds to the provision of BI; enabling users to get the diverse range of information they need in a decentralised way requires guidance so that they understand the business meaning of the information they are using, communication so that they effectively share with each to avoid different versions of the truth and duplication of effort, training and support so that they get the best out of the tools that are available – so important tasks for IT. The end result can be an agile organisation cutting costs and increasing profits in different ways and at different levels. The alternative of a centralised team creating corporate reports is usually easier to govern, but delivers more of a lowest common denominator, encouraging the different parts of an organisation to act in a more regimented way and often masking local issues and opportunities. Ironically smaller organisations often find it easier to choose the decentralised approach due to a smaller IT capability relative to the size of the organisation.

A twist on this is the classic and long-standing information dictatorship or information anarchy analogy, with the dictatorship an overly-centralised version of the above sometimes ironically ending in information anarchy, where business users are so hamstring by the lack of relevant information they resort to building their own data silos, typified by a proliferation of MS Access databases and Excel spreadsheets.

Representation and the voice of the people are hot topics, with organisations such as 38Degrees helping take people’s opinions to the government between elections, famously in the case of helping stop the great woodland sell off. Two parallels spring to mind here … 

Firstly, organisations vary in the amount of feedback they want from employees into strategy and decision making. How much does your organisation value its employees’ opinions? One way to gauge is the information that is provided to employees… can they/you easily access information about company performance, projects, services, costs? See this other post for more on information as a tool for democracy. 

The second potential canvasser of opinion with modern BI is that of sentiment analysis. We have the tools now to understand trends from textual information that people have written in blogs, survey, feedback forms and the like. This information can then sit alongside the traditional numeric data that we are used to reporting, sometime called unstructured and structured data.

In our experience there are two primary approaches to managing BI projects, and this is perhaps applicable more widely to other IT projects, that of the free market versus the planned economy approach, to take a  slightly economic tangent. Some of the pinnacles of human achievement, such as the Great British National Health Service, have been achieved through the central planning and delivery of services in a way that can take a longer view than market trends tend to, and serving the interests of people beyond those who are directly influencing market forces. This reflects what is probably the standard in project management for IT implementations, including BI - that of PRINCE2 and similar methodologies. The ‘planned’ methodologies such as PRINCE2 have a strong focus defining the project according to value it is intended to deliver to the organisation. This enables projects to focus beyond the immediate day to day requirements and pressures to what the impact will be. The clear definitions that PRINCE2 promotes make it popular with projects that involve more than one organisation, for instance when a supplier or partner is brought in. This is contrasted with Agile methodologies which enable projects to move in a more fluid way, responding more easily to feedback and requirements that surface along the way. And how many times do we see real requirements surface and potential impacts understood only once the business start to see the first project deliverables. This ‘free market’ project approach can therefore have two main advantages – there is less upfront planning time required before the first impacts are seen and it can respond to changing requirements and organisational changes that happen during the project.

So, Obama or Romney? Centralised or localised? Planned or free? What do you personally believe in, and do you apply this to your work?

Written by Angus Menter,  Senior Business Intelligence Consultant, DSCallards

1 comment: