Predictive analytics (PA) can save companies money, assist them in identifying upwardly trending buying curves so they can be first in getting products and promotions to market, and anticipate equipment failures on production lines before they happen. PA can also identify potential disruptions to supply chains, predict climate threats, and identify the employees who are most at risk of leaving the company for other opportunities.
SEE: Report: SMB’s unprepared to tackle data privacy (TechRepublic Premium)
Despite this, the uptake of predictive analytics has been slow. In a 2017 Dresner Advisory Service survey, only 23% of companies indicated that they were actively using predictive analytics.
Given the COVID-19 crisis and other analytics projects that are closer to the front of the line, this relatively slow adoption rate hasn’t significantly changed. It led ResearchGate to state, “More companies realize that predictive analytics enables them to reduce risks, make intelligent decisions, and create differentiated customer experiences. … Yet, the adoption rate is slow, and organizations are only beginning to scratch the surface in regards to the potential applications of this technology”
One reason for the trepidation is that developing effective predictive analytics models takes time and resources. If companies lack the ability to hire or fund these resources, they must look for commercial software solutions, but in most cases, best-of-class solutions have not yet been clearly established. Then, there is the most likely user of predictive analytics—corporate finance departments. By nature, finance departments are conservative, and prefer to adopt solutions that have been time-tested.
All of these factors could convince a CIO that it’s best to wait for predictive analytics to mature as an analytics discipline—but is it?
“Predictive analytics is the new competitive advantage that can help businesses leapfrog others to the top of their industry,” said Christopher Warden, CEO and founder of Liquid Lock Media. Warden identified fraud detection, customer relationship management, marketing and sales, price optimization, and operations management as key areas where predictive analytics can make a difference.
Companies can gain competitive advantage by using predictive analytics and will even risk falling behind if they don’t. But at the same time there is still much uncertainty and risk. This is a reality that CIOs and businesses must deal with.
Here are several best practices that can help.
1. Decide how accurate your predictive analytics need to be
We all want 95%-plus accuracy—but do we have to wait until it is achieved?
There are companies that are using PA at relatively low risk in operational areas that might only start out with 65% accuracy. They don’t allow these models to make the end decisions, but they use them as general guides that can point to trends that can help in decision-making.
2. Decide how much risk you can assume
This is a corollary to the first point on analytics accuracy.
There are cases where companies can afford to take on more risk with lower PA accuracy levels. Over time, PA accuracy will improve based on what is learned. This also enables you to start getting PA into the game.
3. Plan for disruption, and continue to refine your predictive analytics models
Business and outside factors continuously change.
For every predictive analytics model that is developed, companies must continuously improve and refine them. This improves the accuracy of the analytics, and also assures that the company moves forward at the pace of business.