This blog is part of a tailor-made content series around the SAS Forum Belux 2015. It is linked with the event track called “Data Management”. Click here to join the event and learn more about the other 3 tracks (Internet of Things, Digital Society and Data Science).
Everyone seems to be talking, these days, about radical innovation and disruption by start-ups like Uber or AirBnB. Most solutions in this field tend to focus on organisational culture, collaboration and leadership. Though I’m utterly convinced that these measures are absolutely crucial, I believe that only truly information-centric organisations will be the ones surviving and thriving in these fast-changing times.
In this post, I will ask you 4 very short – and deceptively easy – questions. If you can offer a positive answer on these 4 points, then your company is future-proof. If you cannot, I regret to say that your information strategy is in serious need of reviewing.
1. Can you get business value out of your information and data insight?
We are living in exponential times when it comes to information. As with many things, this is both an opportunity and a risk. It is almost strange to see, though, how decision-makers are obsessing about these ‘new’ types of data – Big Data as they are usually called – while they tend to neglect the much easier to mine, much safer and often much richer data that lies within their reach.
Do not get me wrong: Facebook posts, sensor data, Fitbit records, RFID data, … - what I call informal events data – certainly offer valuable insights and will definitely drive new business models. But they are more complex to deal with, demand a significant effort and machine power and a lot of companies are not ready to use them in a sustainable way. I call this type of data HOV-data, as it’s `Harder to Obtain Value’ from, given the need for additional processing or algorithms. It’s important to see beyond trends and not to lose sight of the low-hanging fruit: the kind of data analysis that’s more concrete and closer to reality. Just think of how Albert Heyn missed the opportunity to outperform a (then) small newcomer like Bol.com because it was lacking focus, customer-centric information management and failed to leverage its enormous treasure of customer data from loyalty cards – a true differentiator when applied properly.
`Data lakes’ can offer a valuable solution for addressing many insight needs. But they are not the be-all and end-all of analytics. Hadoop is not the only solution on the planet. If you have to analyse terabytes of data, then you do indeed need a massive parallel processing platform that can scale horizontally. And yes, without this kind of technology, your company will not be able to handle very large volumes. But if you are looking for ways to react in real time to click actions on your website in order to enhance conversion, Hadoop might not be the answer. What you do need is a complex event processing solution of a very different type, which can analyse and respond to events in less than a microsecond.
If you want to realize the value and potential of your data, you have to understand that there are very different models available which all excel at addressing the problem they are designed for. Unfortunately, there is ,at this stage, still no single/magical data management solution, as vendors would have us believe.
2. Is your data correct?
The current trend is to see data quality as something that we can work around. And it can be, providing you are operating in a statistical environment and are looking for patterns. A data scientist looks at quality objectives from a statistical point of view: (s)he will often remove outliers and their algorithm will still provide the insight an organisation needs. But data quality does not always have this ‘stretchable’ quality objective. Imagine you are loading a truck with packages. In that case, you need to know the exact measurements of those packages. Statistical relevance of the measurements will not help getting the packages in the truck if they are oversized.
It is important to separate the trends – about cognitive computing and forecasting models – from the operational reality of an organisation. The inherent quality of data still DOES matter in a lot of cases. It is not because the margins of data quality are flexible in some pattern-searching cases that we can just extrapolate this to other situations. So it’s always a good idea to have your data scientists – if you have them – collaborate with your data engineers, to reach a balance in the matter.
3. Do you have the right data points?
The amount of data you collect is not the benchmark here – the level at which you can capture useful context is the main differentiator. “There is an App for that”. People consult banking apps in their phone up to 4 times a day. Can your bank gain valuable insight out of these events? The ultimate synergy comes from the technical ability to capture the data in a non-invasive way and to make sure they serve a business purpose rather than just hoarding the data. The scope of these data points is not only event data, or high volume data, but also the `small data’ that can provide the context that increases your ability to get actionable insight. Fitbit data without any knowledge of the person wearing the device is not likely to yield great results. So choose your data points wisely and make sure you don’t have any blind spots or end up being a data hoarder.
4. Do you have permission to use the data and is it properly managed?
Privacy should be a major concern for companies. Not so much because of the compliance rules which seem to be proliferating in recent years, but because it is what your customers expect. There are two main issues here: transparency and consistency. If you are using your customer’s data, be open about it. "The customer is not a moron. She's your wife," as advertising guru David Ogilvy used to say. And just like your wife (or your husband) is going to find out if you did something shady, so will your customer, if you hide it.
If you use their data, let them know. And stay consistent with the image you are projecting towards the outside world. Apple, for instance, uses privacy as a value proposition, so they should remain faithful to that proposition at all times. Google, on the other hand, goes quite far in leveraging the data of its users, but they are very transparent about it and do not try on another image. So always stay true to your company values in this matter. The data privacy must be at the centre of your data processing and you should have privacy by design rather than as an afterthought.
In 2014, 229 data breach incidents involved the personal records of people in Europe. Globally, all data breach incidents tracked resulted in the loss of some 645 million records, though not all of these breaches exclusively involved people in Europe. Within Europe, we confirmed 200 cases involving people in Europe, and 227 million records lost in Europe-specific breaches.
Just remember the toe-curling example of Target – ‘outing’ a teen pregnancy of a young daughter to her unsuspecting father – purely on the basis of her online search-behaviour – and the subsequent data breach that happened and that took the company from hero to zero in no time: resulting in the resignation of the CIO and CEO and a plummeting stock price. You don’t want that to be the next Target.
I hope, for your sake, that you were able to answer these 4 questions positively. If you didn’t, this might be a great opportunity to re-examine your information strategy.