Our Thinking
Getting Started with Big Data & Analytics (1/4)
10 August 2015 | Posted by Darragh McConville

I’m often asked on by clients on the best approach to starting on their big data journey. I prefer to describe it as an analytics journey, or adventure, if you will.  I would like to discuss the ten steps I see as being crucial to establishing a foundation for organisation-wide success.

I see these ten steps falling nicely into four phases that I call: Focus, Build, Enable and Evolve.  I’ll cover Focus in today’s post and will cover the remaining three in later posts.

Focus

The Focus phase will see you: identify your senior responsible owner, describe specific problems you need and want to address and understand the characteristics of the data needed to address these challenges.

1. Assign ownership

It is imperative for you to identify a senior owner for the analytics initiative. There will be multiple key stakeholders who will influence, shape decisions and execute agreed actions, but without a senior owner your initiative could be short-lived.

As the initial owner, your CIO will be a good fit. They are usually at the centre of all things information-based, meaning they will have a vested interest in such an initiative. Senior ownership can, and often will, change throughout the programme with some organisations maturing to such an extent that a dedicated Chief Analytics Officer role is created. [1] A C-level owner will provide visibility to, and secure commitment from, members of the top table.

2. Identify problems

You will identify the problems that you need to solve through the questions that need to be answered from your data. You will require business unit heads to engage their teams to provide these challenging business questions and to complement your team with domain expertise. Questions, or use cases, should each be scored and ranked by business value. This will prevent you from tackling a use case that you are unlikely to gain from once it is solved.

Customers will often start introducing improvements to existing analytical processes before focussing on innovation. For example, your marketing team may have a customer churn report, which takes four weeks to run. They have estimated that by receiving this report weekly, churn rates could be decreased by 25 per cent. This is typically an exercise in re-designing the existing report generation process on a new platform to provide the same insight, but faster.

Innovative use cases, or new business questions will soon follow. In the approach to the general election, the marketing team may ponder, “Does political preference in specific regions influence our customer retention?” With an estimated high business value, you now have the challenge of identifying the data needed to answer this.

3. Profile data

Identifying the raw materials to your innovative use cases requires lateral thinking. Your biggest data resource is not in your corporate data centre, it’s the web. [2]

A data inventory of open, social, sensor, geospatial, telemetric and all relevant data sources is something you should start to create, but your immediate focus should be on your internal data assets. Identify the owners of each of your internal data stores and create a profile of each store based on the famed five ‘V’s (Velocity, Variety, Volume, Veracity, Value). The value of the dataset refers to the usefulness to the use cases in question.

[1]  http://ibmdatamag.com/2013/05/the-emerging-role-of-the-chief-analytics-officer/

[2] http://www.firstpost.com/business/gartner-predicts-3-big-data-trends-business-intelligence-2123597.html

Next up is the Build phase.

No comments
Leave a Comment