For many years companies have been working to increase their use of predictive analytics and to execute analytic models faster on increasingly granular and growing volumes of data. Recently, there has been a great focus on "faster" from a  technology standpoint, as modelers seek to iterate quickly and fail fast on all the data using a wide variety of sometimes computationally intensive analytic algorithms.

This focus on speed is especially seen in the world of the emerging data scientist. But quick answers are equally important to traditional modelers and for models deployed into production that need to respond faster than ever before.

To meet this need, analytics vendors have responded with technological innovations such as high-performance analytics and in-memory solutions for Hadoop, which have been developed to deliver breakneck analytical processing speeds. The unbounded possibilities for solving problems on larger and larger amounts of data at ever increasing speeds empower data scientists, data modelers and business decision makers.

On one hand, the amazing technological leaps are something to be proud of, but frankly speaking, no one is really looking for speed alone, and focusing just on speed might lead you into a trap!

Focus on experience and culture 

One thing I've learned working on analytics projects is the need to develop an analytics culture, and to have a thorough understanding of the business process you are looking at, so you can gain the maximum return from your investments beyond just buying another piece of technology. Technology is often perceived as the answer to our problems, but good technology incorrectly applied, politically blocked or badly supported will never help you garner the results you need.

I'd like to offer ten steps for making an impact with analytics. Pay attention to how many of these are cultural issues instead of technology issues. You might be surprised:

1. Understand the business process you want to improve, from end to end.
2. Understand how you will measure the improvement of that business process.
3. Baseline the process to understand how it is performing today.
4. Identify where the largest bottlenecks are so you can focus initial attention there.
5. Understand the technology obstacles you need to overcome to move forwards.
6. Understand the people and process obstacles you need to overcome to move forwards.
7. Determine if fixing technology first will just result in new technology but no business process improvement because of 6 above. If so fix the people and process problems first!
8. Acquire the technology and put to work measuring impact.
9. Document what has been done and why it improved things.
10. Speed it up as much as you can.

Number ten is important, but it should be the last consideration, in my opinion. The first nine steps need to happen before you worry about speed, unless speed alone is the technology problem. Beyond that, speeding up more than the process can handle brings no benefits. For example, if you can detect an event, such as fraud, every two seconds but cannot do anything else for five minutes, then that two-second detection is overkill until you can handle the downstream action.

Next steps

SAS has spent the last 30 years helping companies work through these ten steps across many processes in many industries. Today we help organizations develop innovative new business processes and approaches leveraging that expertise. For more information read this post from my colleague Aiman Zeid about building an analytics culture.