In this stage technology helps automate existing reports or ones that were too time consuming to manually assemble. Reporting tools can create and distribute scheduled reports to analyst and users. Business Intelligence tools can get data from multiple sources and tables within sources, eliminating the manual gathering and assembly of data so analysts can begin to analyze immediately.
Success: Productivity improves dramatically. Data that took hours per day to gather and assemble are now ready for analysis first thing in the morning. Data that are important for decisions but were previously too time consuming to assemble can now contribute to the company's strategy. The ROI for making the data visible and accessible is very high.
Challenges: Reports tend to be canned versions that have been automated. Many go unused or provide little value. No one is tasked to determine which reports are being used and provide value. Reports and data weren't designed to answer business questions or solve business problems. When value is provided the next steps to increase value is not taken. The ROI for gaining insight to make good decisions may not be that high.
Many organizations get stuck in this area because technology is the easiest change. Organizational changes and process changes to become a true data-driven organization is much more difficult.
This stage is where analytics begins. It's easy to confuse canned reporting and business intelligence with analytics. They are not the same. The book, Behind Every Good Decision: How Anyone Can Use Business Analytics to Turn Data Into Profitable Insight defines analytics as the science of applying a structured method to solve a business problem using data and analysis to drive impact.
Intuition + Data = Actionable insights → Good decision
At first glance it may seem odd to define data driven decision making (DDDM) as the combination of data and intuition. Isn't the point of DDDM to remove intuition from making decisions? It turns out you still need the intuition, or rather a more scientific version of intuition, to help turn data into actionable insights. Intuition that can take a problem, chew it up and turn it into ideas and hypothesis.
This doesn't mean just using the data from stage 1 or stage 2 and trying to gain insights. You have to go backwards. Start with a question, usually a why question, formulate a guess as to why something is happening. "Why is " revenue for a product falling, there an increase in quality issues on a production cell this month, website traffic decreasing this quarter. Then determine the data and the analysis needed to prove or disprove the hypothesis. This is where intuition and data come together.
It's not that organizations are not asking why and formulating hypothesis. This happens all the time in all companies, even in stage 1 above. However, it's usually implemented without using data. Experience, hunches and gut turn the intuition into decisions, but not always good ones.
Success: Problems, answers, and the data to support both, all come together rather than in isolation. This leads to better decisions and results that should align with company goals. Data and reports are actually used rather than just generated. The ROI for from turning data into insight and then actions is very high.
Challenges: Becoming a data driven organization is part of the process to enter and advance in stage 3. Leadership, people and process are the key drivers. Leadership has to believe in, support and make data driven decisions. People with analytical and project management skills must understand they need to drive the process from "why" to data supporting the hypothesis which then can turn into action. And repeatable processes must be in place which everyone uses to go from why to action. These are all hurdles which take a lot of organizational momentum to go over.
The final two stages represent the future. Many companies are doing both today, but the vast majority are still making their way up to stage 3. Predictive, as the name suggests, is using historical data (including current which immediately become historical) to predict the future. Zillow is a good example. If you look up your house on Zillow you will see an estimated value. Someone has figured out the many variables that can impact your house value; location, schools, transportation, square footage, rooms, amenities, neighborhood selling prices and their success rates, just to name a few. Using statistical and machine learning models on past data, Zillow, can predict your house selling value.
Prescriptive then takes this a step further and determines a course of action. For example, GE uses IoT data and its Predix platform to capture large amounts of data from customer assets such as turbines and engines. From here, Predictive analytics can determine a potential future failure point of these assets, and using the same data, Prescriptive analytics can model a recommended fix.
The growth of IoT now and in the future will make Predictive and Prescriptive analytics much more common. The use of AI to solve many problems such as driving, making investment decisions, finding and extracting fossil fuels and providing healthcare would all be classified as Prescriptive analytics.
Data and analytics may be the hottest area in IT at the moment, but is expected to continue into the future. Part of the reason it's not a fad, is that the definition of analytics I used above; intuition plus data turned into insight to drive actions, describes what most living intelligent beings do from birth until death. Now we have created the technology and methodologies to go much further than our biological capabilities can take us.
Reporting tools, and even Business Intelligence, are very common and now easily available to the smallest of organizations. This availability is making data more accessible while minimizing the labor needed to assemble it. That's the first step. Then the harder step is to begin to do the detective work - hypothesizing solutions and answers, then determining the data needed to prove or disprove the hypothesis. This takes both domain expertise, statistical and management skills. The ROI for moving up the roadmap are impressive as shown in the graphic below from the Ironside Group.
This doesn't mean you have to hire Machine Language experts, because just like Mr. Pareto found - 80% of problems can probably be analyzed with simple basic math. However, you do need certain people skills to move away from canned reports and default dashboards, people who can think through problems and come up with their own analysis and plan. That's the big step for 80% of organizations which are just beginning the journey.