Why Retailers Need Pragmatic Artificial Intelligence (AI) Now More Than Ever

Increasingly heard in retail today is machine learning and AI. Pragmatic AI is a branch of artificial intelligence that uses cognitive search and machine learning capabilities to help retailers make more effective business decisions. It works best when powered by prescriptive analytics. Prescriptive analytics identifies an opportunity for improvement and delivers an easy-to-comprehend prescribed plan of action and resolution. This intelligent prescribed task ensures accountability at all levels throughout the retail organization.

A pragmatic approach to AI in conjunction with prescriptive analytics offers retailers a powerful solution to compete in today’s market.

Artificial intelligence (AI) is not a new subject for retailers. It has been a buzzword in the news for a number of years. The question today is, how do retailers get the most business value out of it in the real world?

That’s where “pragmatic AI” comes into play, as a guide for businesses to be more efficient, increase sales and margins, and better understand their customers and operations.
Many vendors use the term “AI” in their marketing, referring to the highly technical human-like learning capabilities of analytics solutions. Pragmatic AI incorporates cognitive search and machine learning to help retailers understand their data. In turn, prescriptive analytics is the part of AI that makes it pragmatic. It uses machine learning to automatically identify opportunities within the business and then couples these decisions with recommended actions to improve business.

This piece discusses the various approaches to analytics. The two most common include:

Descriptive analytics– produces visuals or aggregated data that leave interpretation to the organization. They are thus open to misunderstandings or overcomplicated next steps for action. Descriptive analytics is highly visual, but it does not help answer the question of “why.” Why is this store in the red? Oftentimes it takes several analysts to discover the root cause of the issue.

Predictive analytics – moves a step beyond descriptive. It creates predictive models of outcome based on previous behaviors. Again, the drawback of predictive analytics is that its results are too complex for the average person. It often requires analysts to interpret and recommend next steps, and it doesn’t fully measure behaviors.

Typically, those producing the information do not understand the business very well, and therefore may not produce valuable or meaningful findings. Both approaches involve smart systems, but they also require very smart people (often PhD-level analysts) to fully understand them and determine the next-step action to solve the issue at hand. Prescriptive analytics is not a super-special or unique type of analytics. Rather, it is an umbrella term that refers to any and all types of analytics that can improve the effectiveness of decisions — whether they are made by humans or software applications. Think of the term “prescriptive” as the goal of all of these analytics — to make more effective decisions.

Profitect’s prescriptive analytics solution takes advantage of customers’ data, whether already in a data lake or still in its initial source locations. The solution then analyzes the data in near-real time, and identifies areas of opportunity to optimize the outcome performance based on guided actions. It clearly defines, in plain language, what was found, why it matters, and how to address the opportunity.

Having the correct “prescriptive action” to take and a workflow that can track user behavior means employees are held accountable for assigned tasks. This effectively closes the loop and ensures that any situation that has arisen (positive or negative) has been resolved.

What makes this approach so pragmatic is that it maximizes the use of technology, but simplifies and democratizes the results in a way that engages the entire organization. Pragmatic AI can be used by anyone in the organization without needing a Ph.D. in statistics. The solution is able to replicate all of the knowledge and processes that the analysts often do when evaluating data, and automate the interrogation of that data. This empowers analysts to now focus on new patterns of behavior to look for with the right tools. The system is also smart enough to continue to dynamically adjust via it’s machine learning algorithm.

This is also why Profitect’s prescriptive analytics is the ultimate example of pragmatic AI. AI will become the ultimate form of prescriptive analytics — and Profitect has the top-rated solution on the market.

Prescriptive analytics and pragmatic AI present retailers with the best opportunity for optimizing outcomes and delivering real returns that improve business operations over other analytic technologies. The following is a real-world example of an opportunity that was uncovered and corrected with Profitect’s pragmatic AI and prescriptive analytics capabilities:

Reward Loyalty, Not Fraudulent Activity
A large retailer with a popular loyalty card program used prescriptive analytics to uncover an issue they weren’t even aware was a problem. Leveraging Profitect they were able to discover a trend involving shoppers cheating their reward points for free rewards. Their retailer’s previous system did not monitor points scanning until the end of each day, so if a customer brought their receipt to multiple stores in a single day, they could receive many more reward points than were due. Profitect’s prescriptive analytics solution caught the recurrence and recommended real-time tracking to ensure receipts could only be scanned once for points, saving as much as $100 per shopper, and adding exponentially more to the bottom line.

Guy Yehiav is a recognized thought-leader on retail technology. Visit www.profitect.com to learn more.

This entry was posted in AI, analytics, Articles, artificial intelligence, descriptive analytics, fraud, fraudulent activity, predictive analytics and tagged , , , , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>