It is hard to ignore the exponential growth in the digitisation of data and the advancement in analytical tools/techniques to exploit this growth. Early adopters have shown the possibility of real benefits in adopting these techniques but in parallel there has also been a realisation of the challenges in doing so. Some of these challenges include: the difficulty in attaining the right expertise (data scientists), the legal and ethical implications of processing data (the ability to analyse data does not mean you have the right to), and the balancing of intuition over analytics (do you go with your gut or do you go with the data). However, even before business encounter these challenges I feel they are being blinded by thought of analytics before ever really understanding what it can do for them.
Case in point is big data analytics. Built out of necessity for companies (eg Google, Facebook) that were being restricted with traditional solutions, big data technologies offer fantastic functionality. The nature in which these technologies are pushing boundaries of what is possible in data digitisation and analysis, is akin to what Formula 1 cars are doing the motoring industry. The analogy fits to the extent that the cars do a task at hyper performance (going around a track over 50 times), they need seriously experienced staff, and can be pretty temperamental. What is also interesting is that if I extend the analogy to include 99% of businesses, they really only require better delivery trucks. Even for those companies that have implemented big data solutions and got value, I wonder if their money would have been better spent elsewhere. A well know big data success story published in the NY Times detailed how Target was able to predict pregnancy so accurately, in one instance it knew that a high school girl was pregnant before her farther knew. However, further discussing this example with a US data professional they felt that even taking the story at face value, Target may have been better placed to invest in improving their stock fulfillment processes.
Explained another way, analytics is but one stage of the Data Value Map (see diagram), yet with the ever increasing emphasis on trends such as big data and advanced analytics, businesses tend to lose sight of the end-to-end data life-cycle. For instance, a business that focuses on the analytics/delivery phase at the cost of the acquisition/integration phase will basically build a capability to make bad decisions faster and prettier. What’s more frightening is that the negative impact is not seen and this way of thinking is starting to gain momentum. In the world of big data the paradoxical notion that trading off the quality of data for increased analytical power is becoming more common. The logic behind this is that if I have so much data, what does it matter if some of it is of poor quality…in the end the good will far outweigh the bad…right. WRONG. Taking into account the rule of thumb that completing a task with bad data will cost you ten times more than if it’s done with good data, you will start to see the bigger picture. To give more context to this rule, recently a business did a data quality audit and found that it had a 92% data quality rating. The only problem was that the 8% defective data was costing them €16.3 million. So my question to you is what would you do in this situation……invest in analytics or sort out your data quality? Assuming you picked the latter, you may also want to do the same for your own business as past experience points to the fact that you have a similar problem, except you are not seeing it.