Knowledge is power. Everyone “knows” that. But how does one attain it? There are no easy answers to that one. Analytical prowess will be the battleground for service providers in 2013 and beyond. Are you ready to take the statistical fight to your competitors or will you be on the back foot when the time to run the numbers comes?
Ascending the knowledge pyramid from noise and misinformation to achieve wisdom is not easy.
All too often, data analysis projects are handed off to the support representative that knows how to use Microsoft Excel the best. Whilst this may work occasionally, it is seldom the best option. Statistical analysis is a distinct skill. You may have people working for you that possess this skill. You may not. If you do not, you WILL have to get some. Whether that be on a sub-contract basis or through the hiring of someone with the pre-requisite background, experience and talent. But data mining and modelling skills in themselves are not enough. Statisticians need to be grounded with a healthy dose of reality. You must augment their talents with the knowledge and experience of the field. Your brightest and best support representatives are an amazing asset. Use them. You “just” need to pair them with someone who can ask the right questions and draw the embedded insight from within them in the form of a statistical model or algorithm.
The big data deluge represents a real and present danger.
The interconnected nature of things means that every data point is relevant. And yet if one goes too far down this route then analysis paralysis will become terminal. The first iterations of your models should be used to identify which data streams have the biggest impact on various failure modes. Remembering that different factors are likely to have different weightings for different scenarios. No-one expects you to be 100% accurate 100% of the time. Being mostly right some of the time is a good start. As with all things, version 1.0 should be functional and demonstrate the potential of the effort. Understanding the statistical reliability of your predictive models is essential if you are to base business rules and automated processes upon the outputs of the analytical black box.
Failure to plan is tantamount to planning to fail. Go big or go home!
Last time I deliberately didn’t talk about analytics in the context of the “Dear Santa” wishlist as I wanted to demonstrate that it is not a whimsical fantasy. It is a reality that any support provider willing to invest the time and resources can achieve. I’m not saying that it is easy or cheap. It is not a trivial task. But it IS a necessary one that every support provider will have to go through at some point or another. Why not take the lead, grasp the nettle and use it for your advantage? Assuming you do want to give it a whirl then the following action plan is not a bad place to begin:
- Conduct an internal skills audit (Recognise that you may have to draft in some talent from outside)
- Form a cross discipline team to develop an initial set of analytical models
- Identify data sources and their relative trust level (Based on data quality, completeness, staleness, number of transitions / translations, distance from trusted source etc.)
- Create a data acquisition wish list (What do you need? How can you get it? When will you have it? How can you work around it for the time being?)
- Develop models for most common failure modes (Avoid looking for the uber solution, there probably isn’t one!)
- Test and refine the models until they prove themselves
- Determine business policies for how the outputs of those models will be used. (If they’re not used there is no point in having them after all!)
Que sera, sera.
Whatever will be, will be.
The future’s not ours to see.
Que sera, sera.
What will be, will be.
Although it pains me to say it. Doris was wrong. The future IS ours to see. If we build the underlying analytical capability we need and look hard enough.
I wish you well with all of your statistical endeavors in 2013…
Comments or opinions expressed on this blog are those of the individual contributors only, and do not necessarily represent the views of Gartner, Inc. or its management. Readers may copy and redistribute blog postings on other blogs, or otherwise for private, non-commercial or journalistic purposes, with attribution to Gartner. This content may not be used for any other purposes in any other formats or media. The content on this blog is provided on an "as-is" basis. Gartner shall not be liable for any damages whatsoever arising out of the content or use of this blog.