I encounter repeatedly the idea that companies need to have a great reporting system and wonderful dashboards before they should begin to think about artificial intelligence and machine learning. This is incorrect and maybe even harmful to companies’ ability to adapt.
The worst situation might be the company which declares they must get their dashboard software more widely adopted before they can contemplate other types of analytics. Likely, that company will never get to a place where all the executives agree the dashboards are great, perfect, and satisfy every reporting and data visualization requirement. That level of adoption will never happen so the company cannot move to AI and machine learning.
AI and machine learning certainly need reporting experience in the sense there needs to be an understanding of what data is being stored, where it is located, and how can it be accessed.
Historical transaction data is the foundation for AI and machine learning. Without the input data, there can be no predictions about future events and activities. However, the historical transaction data does not need to be in the form of a dashboard, chart, or easily readable table to be used for machine learning.
Indeed, much of the reason dashboards do not meet initial expectations at companies is the exact reason companies could use AI and machine learning based predictive analytics.
Dashboards often disappoint because they are bought with the expectation they will solve problems. Really, the dashboard just summarizes historical data into easily consumable charts and graphs. The dashboard still requires a person to look at it, interpret what the data means, a project by judgment how the past will affect the future, and then formulate an action or change in operations.
When people need to make projections for tens of thousands of products or processes, thousands of customers or hundreds of salespeople, the process of just looking at the historical transaction data breaks down. Imagine having to look at a dashboard chart for every product a company sells grouped across every region they operate in. No one is going to do that for more than a few handfuls of products at most – executives and managers don’t have the time to explore their data sitting in front of dashboard screens for hours at a time. They have real jobs they must get done today.
Making predictions quickly is what AI and machine learning does.
Bring in the concept of IoT data, social media, website traffic or almost any other recent data source and you can see how simply looking at a chart of what happened in the past is not going to help much to predict the future.