ANABYTICS

Our little CIOecret. The A-Blog.

The A-Blog's Latest Article:

04.11.15 A-Article 1.2: 'Big Data: Customer Analytics - recommendations for marketers'

bIG dATA Customer Analytics for Marketers

In the same week I heard three consultancy salesmen talk about how supermarkets and large retailers had so much data they didn't know what to do with it. These people went on to tell me how their teams could make it all better by simply developing a Hadoop platform and loading everything they could find in to it. The nature of the comments were such that I think many are new to general commercial data principles and perhaps don't understand that historically retailers are some of the most switched-on of organisations with regard to marketing and data use. Just look at the Supermarket's big loyalty schemes, the fact they have often owned data analytics businesses and continue to invest heavily in this area. The whole point of these strategies was/is to pull data together, analyse the customer better, develop sales strategy and extract profit. Bingo! They aren't so stupid then are they!?

Bigger and better collective data programmes are not a new target. The techniques and technologies available to us are now more powerful but to be successful in our strategic endeavours for these new technologies to be useful we must get back to basics. We must learn from data projects public and still present - considering both their successes (their BIG and innovative thinking) and their frustrations (why they now feel disconnected from their data sets and feel unable to rely on them). In line with these ask yourself the six questions below before you embark on a customer focused big data programme:

  1. What are the holy grail questions with in the business and industry you work in? Do a full survey amongst C and Board level staff. Call it 'mission critical'. Ask them to be thorough and wide ranging, collate the results and extract a set of questions it would be great to have the answers to. Then ask industry consultants to add a few. Make sure these are customer centric.
  2. Do I understand the complete customer journey? Jot this down in a process flow diagram. Be inclusive of everything related to the customer and the business activities behind interaction or desired interaction. Get every part of the business to contribute to confirming and expanding it.
  3. Do I have enough data of the right kind? Just because you have millions of rows of data flowing through a data centre somewhere it doesn't mean you can extract holy grail analysis from it. Go and find out what you've got before you decide to pull it all in to one place. Most businesses don't know what their full data asset looks like. Some expect there is less, some expect more. The results of investigation are often peculiar. Once you have done so, go back to your complete customer journey process flow diagram and identify what data you have for each phase.
  4. What solutions can I put together where data is lacking? You may be months or years away from being able to answer that holy grail business question if you do not have the correct data required. If you can't find a data collection method with in the business it's time to think about developing technologies or purchasing external data. Get buy in from above- don't hide from the truth.
  5. Has my source data been validated? I do not mean 'tested'. Any seasoned Data Scientist or rigorous Auditor will tell you that if you are looking for qualified, professional outcomes from your data analysis, the data sets you are using big or small must be validated to a high degree before analysis takes place. Anabytics has found that many big names, actively selling data sets, modelling technologies and the analytics outcomes drawn from them, do not validate their data in any way. The same old rules apply. If you put sh*t in, you get sh*t out. Eliminate that risk.
  6. Do I need to go big? Or do I just need to be clever? You may not need to go big. No one at Anabytics will force you to. You may be able to pick and choose what you need in order to create a regular database set and draw similar or perhaps more concise results form your analytics and analysis. Think - what is this data, how is it collected, what is it representing, where does it go next, how much history do I need if any, when do I need it in order to make a difference? The list goes on...

I could create a decision matrix for this analysis right now but I'll leave that to you. By considering these aspects and more you will be one large step closer to a sound data development product. If you don't have the time, Anabytics will be happy to help you carry out this work.

Written by David W.M. Russell.

Coming soon A-Article 2.0: 'Getting Data Visualisation Right'