The Future of Business Analytics
For those handicapped by very short attention spans – business analytics will not consist of people creating charts, graphs and other pretty data using the latest visual analytics platforms. This is to some extent a temporary aberration (for reasons to be discussed later). Business analytics will increasingly become automated as the task becomes too complex for manual analysis. This means machine learning and artificial intelligence will come to dominate business analytics.
There are many reasons why visual analytics have captured the imagination of millions of business users. The most obvious is that they feel empowered by the ability to easily assemble attractive graphics displaying some aspect of business performance. It was only five or ten years ago that most business users had to rely on IT folk – something they truly resented. And so autonomy has proven to be extremely attractive. The suppliers of pretty data platforms know this and boldly assert that IT support is not needed from moody, unresponsive IT people. In actual fact we are already past the zenith of the pretty data fad, as people start to realize that slicing and dicing data and formatting it with designer color schemes can become as tedious as everything else – unless of course a visual is being put together that totally discredits a political rival.
There are some very obvious, unambiguous trends in the way businesses use data that spell the demise of the visual analytics era. These can be summarized as:
- Diversity of data sources and types means data is becoming too complex for us to process manually. Companies such as Paxata, ClearStory, Platfora and others supply automated data preparation platforms which employ machine learning methods to prepare data for use. Dragging boxes around a screen and connecting them with lines will soon become an ineffective way of dealing with thousands of concurrent data sources.
- The speed of data is set to make a quantum leap. The Internet of Things means data will increasingly become real-time, and people fiddling around with graphs and icons just will not cut it. Data traveling at these speeds needs automated analysis, where everything other than exceptions remain invisible to the human eye.
- The patterns within data are becoming more complex. Humans are very poor at handling complexity – we can juggle around five things at a time – maybe seven if you are Einstein. Machine learning algorithms can handle hundreds or thousands, and can discover relationships that might never have been suspected.
- Volumes of data are much less important than the ‘big data’ era might suggest. This was just a very good marketing exercise – the word ‘big’ being almost as powerful as the word ‘new’ in the marketing person’s armory. Data size is important in some industries (retail for example), but for many of us it is almost irrelevant. If we only had the problem of dealing with a billion rows instead of ten thousand we would just be dealing with a scaling issue – not complex, and not worth making a fuss about.
But data is only half the story – the other more significant half is what we do with it. Artificial intelligence is a bewilderingly broad collection of methods and techniques which can add intelligence to business processes and analysis. Machine learning, optimization, probabilistic methods, search, logic, knowledge bases, language processing, video analysis and a large number of other methods are already making their way into the applications we use. These methods will be slowly adopted by businesses (although much more aggressively adopted by consumers) because of a certain level of fear. But as always, the laggards will either go out of business or get swallowed up. The notion that people creating bubble charts and scatter plots will be able to monitor the health of a business in just five years from now is ridiculous.
All of this does not mean that business people have to become machine learning experts. Quite the opposite. They will be increasingly presented with online services that suck in data and spew out meaningful analysis. This leaves business people to get on with the job of turning that analysis into action (although this will also be increasingly automated).
Get ready for a fast and furious ride as AI and machine learning make their way into your business analytics, just as they are in operational aspects of business. Resistance will be futile.