The Now or Never of Big Data Analytics
For those in business, it’s essential to understand the differences between different types of data analytics. Why? Because you need to apply the right model to the actual business problems facing your enterprise. In this post, I talk about strategic, tactical, and streaming analytics and their applications in business.
Analytics has been around for a while in the shape of Business Intelligence systems that are generally perceived as a tool for outputting managerial and financial reports. Starting from tax forms for end-of-year submission to customer satisfaction level reports, all constitute parts of the broader efficiency improvement programs that enterprises tend to run. Mostly, these reports cover what we call “cold data”, with data running up to the previous day. It’s usual that these reports can take hours to complete, summing all the figures, grouping data, and so on. There are some delays, of course. For example, the report may just show data up to yesterday, but for high-level management purposes that’s more than enough – enterprises can change their course in minutes. These reports look at a general direction towards a longer horizon. So let’s call them strategic.
The Right Tactics
There’s another type of query that we can refer to as tactical. Imagine a call center where operators are on support for a bank service. Every time a customer calls, the system receives the incoming phone number and checks everything related to this caller – things like previous calls, feedback given, services, transactions, and anything else that can be in the system. Here’s where the real-time quality of queries comes in handy. Almost immediately, the operator has all the information necessary to serve the user. We don’t want to have to re-ask all the details of the customer, wasting their time on information they’ve already given on previous occasions (and that they’d have to give next time).
Tactical queries are fast, but they handle the same amount of data as strategic queries. They happen again and again, so the system can remember the most common questions and be prepared in the future to achieve data interrogation at lightning speed. It’s important to mention, however, that tactical queries are also performed on historical data.
So far, we’ve mentioned two types of data analytics: strategic and tactical. Most traditional data analytics scenarios cover those two types.
The latest advances in data processing allow us now to work with data on the fly as it comes in. This is called streaming analytics. Think of sensor data from factory equipment or posts that people leave in social media. The latter has a broad range of applications starting from disaster detection to improving the brand image of a company.
Now, when we have these three pieces of Big Data – strategic, tactical, and streaming analytics – the principal difference between them is the time required to produce results. In modern data management systems, the most promising solutions need no latency.
Keeping it Real
All data coming in the system should be analyzed in real-time. However, when one needs greater scope, the system can get the latest data and go back to a given point in time to see the tactical movement of a trend. Also, when the maximum horizon of vision is required, the system will receive real-time data and go way back to some historical date to generate the results. We all can learn from mistakes, but it’s much better not to make them by having the right solution that spots and takes down risks. The real value of big data is now when it commands action, and it’s vital for the industry to understand that analytics will be so fast that it will always be a step ahead of human reaction. And its impact will only grow.
Read more about our our big data analytics solution FusionInsight for carriers and enterprises.
Disclaimer: Any views and/or opinions expressed in this post by individual authors or contributors are their personal views and/or opinions and do not necessarily reflect the views and/or opinions of Huawei Technologies.