The Five Pillars of Analytics

It’s commonly accepted that there are five common areas where analytics can help to achieve successful Big Data exploitation; these scale upwards in increasing difficulty and complexity.

The Five Pillars of Analytics So you want to exploit your company's Big Data to maximise market opportunity

It’s commonly accepted that there are five common areas where analytics can help to achieve this; these scale upwards in increasing difficulty and complexity. However, with the increased difficulty, comes corresponding increased market opportunity.
We call these five areas: Explore, Reuse, Exploit, Automate and Experiment.
  1. Explore - Can sophisticated views over the data (such as customer segmentation) be created?
  2. Reuse - Are there opportunities for innovation (such as new products and services) to be derived from the data?
  3. Exploit - Can an increased quality of data be made available for human decision making?
  4. Automate - Can complex human decision-making be replaced by more sophisticated automated algorithms?
  5. Experiment - Are there opportunities for simulation and experimentation to improve performance?
For those CSPs who are currently running BI (Business Intelligence) solutions (and as mentioned in my previous post) to provide their analysis, it is worth noting that there are three categories of analytics: Descriptive, Predictive and Prescriptive. The most common of these being Descriptive Analytics. These technologies seek to categorise, explore, join together and summarise wide sources of data.

The central problem

The central problem with both descriptive and predictive analytics is that they rarely move past the first of the Five Pillars – Explore. Analytics solutions that present data, whether that data is backwards looking (descriptive) or forwards looking (predictive) only solve half the problem. Without the additional intelligence of what to do with that data, the response is often: “Yes, that is interesting. But how does it help my business?”
Perhaps unsurprisingly, many CSPs have currently opted for the “low hanging fruit” of Explore, successfully delivered by Descriptive Analytics technologies, as such activities offer limited gains with minimal effort. While this is fine, this misses the superior opportunities available through the other four pillars. The panacea is obviously to achieve the market opportunity gains, without the pain of the increased complexity these levels involve…
Prescriptive analytics solutions can offer this. Firstly, the system utilises the large quantities of subscriber approved data already within a CSPs estate. The crucial Big Data that your CSP business is already in possession of, but is currently not being exploited can be processed to offer future insights and unlock the value of the full five pillars.

Realise the true value of your Big Data

It is this ability to fully realise the true value of your Big Data and the rich insight from such large volumes of precise activity that drives truly in-depth knowledge. Secondly, by loading in external data sources; this data includes valuable publicly available information such as news events, sporting fixtures, demographic and social media data.
Advanced machine learning and analysis techniques can then be used to join together and enrich the data sets, as well as expand and train existing models specifically for the CSP in question. The result of this is a series of fully realised data sets that can be immediately understood for business value.
What is needed by CSPs is a holistic approach to simulation, whereby the whole customer and network landscape is evaluated for side effects. Most prescriptive analytics solutions shy away from this, as it is a difficult problem to solve.
However, a truly effective prescriptive analytics solution will provide the holistic market insight that your business needs to maximise value and minimise the demands from each and every one of the five key pillars.
Richard Jarvis, Head of CSP Innovation 8 May 2015