“Hurry, hurry, hurry! We need insights from all this data!”
That’s the mindset of many companies today as they are under increased pressure to generate previously unseen information from the data they are sitting on. They now have to turn to more advanced and even predictive analytics, and they are have to do so on mountains of untouched data.
Where Do You Start?
The advanced analytics and big data spaces can be intimidating. There are so many buzzwords like Hadoop, machine learning, data science, Spark, cloud data platform, and Internet of Things. There are also many companies like Hortonworks, Microsoft, Amazon, Cloudera, SAS, and IBM. Navigating through all of the hype and keeping up with the rate at which technology is changing can be a fulltime gig.
Instead of trying to dive headfirst into uncharted advanced analytics and big data waters, you might try easing into it. Any advanced analytics and/or big data program is likely to take up to 3 years to fully mature, so why try and solve 3 years of challenges in a couple of knee-jerk decisions? Below you will find some tips for easing into advanced analytics with respect to people, data, and tools.
Build a team with 3 or 4 resources you already have in the organization. That’s right, it probably doesn’t make sense to go out and find an expensive data scientist right away.
You likely already have business analysts, developers, and others who collectively have the skillsets necessary to excel in advanced analytics. These people already know a lot about the data and business processes in the organization. They typically already have a good relationship with business and IT teams, and they also share a real passion for solving problems and providing insights with data.
Cross training among the team provides an efficient way for each member to round out his or her skillset. Team members should also be encouraged to continue learning new and relevant skills and technologies.
This does not mean you should completely abandon the idea of hiring a data scientist at some point — just put it off 6-12 months while you establish the rest of your team. Bring the data scientist in as someone who will lead the team and help drive the program. This way, he or she will already have people he or she can rely on that possess shared knowledge, passion, and a solid collective skillset. He or she will be able to hit the ground running and get familiar with the data, the business, and the problems that advanced analytics can solve. He or she will be able to get involved in more complex data science and advanced analytics work he or she is trained for while relying on others for some of the tasks such as data prep, project management, visualization, etc. This will lead to a more productive and less frustrated worker.
Break down the barriers of data access for your analytics team. Traditionally, users have only had access to canned views of data in reports, dashboards, and analytical cubes. They’ve also typically been limited to viewing data relevant to a specific business area in the organization. This limits the ability for data exploration and mashup. A productive analytics team will need access to more raw forms of the data. This includes data in data warehouse staging environments, extracts from vendors, and other miscellaneous sources of data around the entire organization.
Be on the lookout for other, unrealized sources of data. Organizations generate a lot of data they often don’t even realize they’re generating. Examples might include log files from applications or unstructured data in notes and comments fields. There’s also a tremendous amount of free external data sources that can accessed and mashed up up with your company’s data. Analytics teams should be encouraged to tap into these previously unrealized data sources, as they can lead to new and better insights with minimal impact to cost.
As with people, you should tap into the tools you already have or are free and easy to acquire and work with. Most companies already use tools like Excel and Tableau for reporting and dashboards. Take the next step in learning more advanced techniques for data exploration and data mashup with these tools. There are also free tools like Microsoft’s Power BI Designer that make it easy for you to connect to a wide variety of data sources and visually explore and present your data.
Take the attitude of bolting tools and technologies to your existing data architecture. You should not have to throw away and replace what you already have. Big data solutions like Hortonworks, cloud data platforms like Microsoft Azure, and advanced modeling tools like RStudio should all be treated as extensions to your existing data platform. Each of these solutions serve very specific functions and are designed to integrate with other solutions as a part of a larger architecture. That said, extend your tools and skillsets to satisfy the requirements of specific projects as opposed to trying to take a big bang approach.
The Long Term
Having a sense of urgency about advanced analytics and big data is a good thing, but panicking and making reflex decisions can hurt your long term prospects. At the end of the day, you want establish a successful analytics program, architecture, and strategy that will provide insight and value for many years to come. Easing into it using the tips above can help to insure that happens.
Have questions about easing your company into advanced analytics and big data? Contact us today for a consultation.