If you are about to invest big in an analytics tool, stop! Stop and ask yourself if you are purchasing a solution looking for a problem.
If you have not invested time on the below, then there is a high chance you are wasting a lot of your money:
Waiting for massive, multi-year data warehousing and data cleaning programs to complete before starting analytics are also unnecessary. Both these exercises are essential, but there is entirely no pre-requisite to get everything perfect according to a data quality framework before you start getting value from analytics.
Most competent data teams will be able to clean data as they work, often in a matter of days rather than months, and start delivering value right away. More importantly, it is through this exercise that practical data quality issues tend to surface, rather than an abstract ‘requirements gathering’ process.
A much better way of ramping up an analytics team is to invest in people. You need the right people to advise, demonstrate and trial a range of techniques. Build up both the team as well as the operating environment by offering access to data, setting the right expectations, and aligning incentives to analytics decisions instead of analytics delivery.
Open source tools, and heck, even Excel and SQL can bring your organization far. And once it is clear where analytical efforts are valuable and what specific use cases and solutions will look like, it is the time to invest in tools.
One final point about creating an environment where analytics will take root and thrive: you need to stop framing the goal as building analytics deliverables and think instead about exploring data sets, experimentation with modeling techniques, and ultimately making better decisions where it matters.
To summarize, analytics is not a linear process and getting value from analytics is not about software and infrastructure. Analytics is exploratory, and its value comes from people and decisions. Starting with infrastructure and technology when the buyer does not understand it puts undue risk, cost, barriers, and pressure on the nascent function. Starting with good people, aligning incentives, deploying open source tools and building a culture of exploration and fast iteration is better.
This is a contributed article from Jason Tamara Widjaja is Associate Director, Data Science (AI and Data Products) in Merck Sharp & Dohme. The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends.