For years, while predictive analytics dominated the analytics spotlight, prescriptive analytics labored in the shadow of its higher profile cousin, despite the myriad advantages it offers. But that is beginning to change. Adoption for prescriptive analytics is at an all-time high, and Gartner predicts the market will reach $1.1 billion by 2019.
Profitect CEO Guy Yehiav is among many who consider prescriptive analytics “the next level of predictive analytics.” Data Informed spoke with Yehiav about the growing interest in prescriptive analytics, how it differs from predictive analytics, and the advantages that it offers.
Data Informed: Please discuss the similarities and differences between predictive and prescriptive analytics.
Guy Yehiav: Prescriptive analytics is simply the next level of predictive analytics. More and more companies are now seeing the value of providing prescriptive analytics and are excited to adopt this approach.
The key difference between the two is that prescriptive looks at current pattern sets and provides actionable outcomes. Prescriptive allows a tailored approach for understanding both user behaviors and unlikely patterns that can cause organizations to lose money and waste resources. While both prescriptive and predictive analytics rely on big data for gathering and understanding customer information, what sets the two apart is who is assigned an actionable plan to fix the task.
For example, if a business uses a predictive solution, typically a technical person would be notified if a product was low in stock or depleted. This could come in the form of a customer complaint or an employee noticing low stock on the store floor. However, if prescriptive analytics were implemented, the pattern would be tracked along the way. If an issue with a specific product was uncovered, the right person would be notified in real time to make the appropriate adjustment, saving both time and revenue.
Prescriptive solutions take predictive to the next level by providing a desired outcome and the right person to fix the problem at hand. Instead of relying solely on predictions based on educated guesses and past results, prescriptive analytics provide pattern seeking machine algorithms that promote positive customer experience and provide resolution.
There seems to be much more interest around predictive analytics. To what do you attribute that?
Guy Yehiav: Since predictive analytics have been around and adopted since the early ’90s, more people are familiar with the term and likely to use a predictive system for forecasting, allocation, and supply chain optimization. I wouldn’t say there is more interest around predictive, but I would say that it’s a term that has been used for and a process that has been instilled for many years. It’s rather common and familiar and creates a level of interest based on comfort.
What is something that most organizations don’t know about prescriptive analytics that they should know?
Guy Yehiav: A common misconception is that prescriptive is harder to implement than predictive. But the reality is that it can take only a matter of days to get a prescriptive solution up and running.
What most people don’t know about prescriptive analytics is the amount of time that can be saved by not relying on exception-based reports. Most of the reports being generated within an enterprise are going to waste, simply because they are difficult to understand. Reports that are generated should be able to create actionable outcomes – prescriptive analytics can replace traditional reports and generate return on investment in a matter of days.
How does a prescriptive model arrive at the recommendations it delivers?
Guy Yehiav: To understand prescriptive analytics means understanding the pattern-seeking technology and machine learning algorithms that the solution delivers. Instead of relying on industry averages – the value lies in the little pieces of data and flaws that prescriptive analytics picks up on.
Once an irregular pattern is found, companies can configure the workflow based on ownership and security and, most importantly, determine the right person to fix the problem. Physically, the prescription can be delivered in the form of a document, pdf, or video for full compliance. The system is able to use a specific workflow for every time an irregular pattern or behavior is found. Crowdsourcing also can be used to understand how different industries are using the prescriptions and how the solution can deliver the most value.
What are some concerns organizations have about prescriptive analytics?
Guy Yehiav: Most organizations remain skeptical when it comes to prescriptive analytics because they simply aren’t familiar with this method. Since they aren’t familiar, they would go into an implementation thinking that it would take countless months before they see results – which is not true. The average implementation is a couple of days for larger organizations. Another attribute that holds companies back is that their current process is working. Yet, when they do make the switch and implement prescriptive analytics, a smart machine is able to find patterns and issues that they didn’t even know existed.
Specifically, I have found a couple of things that hold companies back. The first is that they think they know all of the issues they are faced with and have the means to solve them. The other is that they are trying to build a business intelligence platform first and are waiting to cleanse their data before implementing prescriptive analytics.
What other issues are limiting adoption? Are these concerns delaying adoption of prescriptive analytics?
Guy Yehiav: The education simply hasn’t been established yet when it comes to prescriptive analytics. Not enough companies are offering prescriptive solutions and, as a result, individuals aren’t comfortable with their use and don’t understand the outcomes that prescriptive delivers. If organizations become more educated on the seamless implementation and the results and outcomes of prescriptive, there will less hesitation and more adoption.
If an organization is run by people who are able to make decisions based on data and not intuition or gut feel, is prescriptive analytics necessary? How can prescriptive enhance decision-making even in an organization with a strong data culture?
Guy Yehiav: Great questions. We hear this sentiment a lot when we speak with the analytical-oriented groups within an enterprise. For example, if the company has a big analytics group, that means they have the power to find the nuggets of data that are out of place. However, if you ask those same people how they would communicate the issue, change the behavior, and generate value, the answers tend to be the same and rather outdated. It’s not uncommon that I would hear IT leaders say they rely on sticky notes, emails, phone calls, and Excel sheets to analyze data patterns.
Well, it turns out the people who are in the field and are responsible for creating change are not always analytically oriented. This means they may not understand the issue at the core in the first place. There tends to be a communications gap between the analytical people and the people in the field. In order to close the gap, they need a prescriptive analytics solution that will find the outliers in the system.
What are some ways that organizations can apply prescriptive analytics? Can you share a business use case and the results prescriptive analytics delivered?
Guy Yehiav: Grocery stores, for example, experience shrink on a daily basis. Since food can only be sold based on the expiration date, it’s crucial that grocery stores have a solid process in place for tracking value. One national grocer signed on with a new distributor to sell a specific line of poultry. After implementing a prescriptive analytics solution, the grocer was able to save the company $25,000 in the first five days and more than $1.8 million annually – simply due to a packaging issue.
For a more retail-focused example, prescriptive analytics can identify products that aren’t performing as well or receive high damage rates and customer complaints. A national shoe retailer would have saved roughly $192,000 by not selling a product with a one-out-of-three return rate. The most important value I see with this example is that retailers and businesses are able to enhance the customer experience by providing products and services that are valued.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise, plus get instant access to more than 20 eBooks.