Everywhere we look these days, we hear about Big Data. I was watching Monday Night Football and saw a commercial for IBM’s Watson. Watson is an Artificial Intelligence (A.I.) that was conceived to play the game show Jeopardy. In 2011, it appeared on its first show and beat its two human competitors by winning $35,734. The two opponents amassed $10,400 and $4,800 respectively. Now IBM is taking Watson’s A.I. and applying the technology to other fields.
In the commercial I saw, titled “A World with IBM’s Watson” IBM touts all the different ways Watson can, and is, changing and helping the world. “415 million people with diabetes could soon predict attacks,” “2.2 billion weather sensors are helping predict disasters,” “10 million tons of CO2 emissions are being reduced,” “Music producers are using data to write their next hit song,” and “3.7 shoppers can find the perfect gift” were some of the statements made in the ad.
“Using computers to analyze Big Data was born out necessity. We had too much information to analyze ourselves and there were critical insights in the information if we had a way to find them.”
This marketing campaign by IBM is pushing the concept of “Cognitive Business,” where businesses use Big Data and Artificial Intelligence to solve problems, make better decisions, and increase profits. IBM is not alone, Big Data and A.I. research and development is booming and new applications are being found daily. As one digs deeper into the details, one might read about “machine learning,” “deep learning,” or “neural networks.” However, these terms can be hard for most people to understand without previous knowledge.
So what is IBM really talking about?
Companies like IBM have spent decades working on A.I. It’s not the A.I. you see in the movie Blade Runner or the series Westworld – there’s no robot that’s conscious and trying to figure out it’s place in the world. The A.I. research IBM and others are doing is centered on data analysis and prediction. Over the past 15 years, technology has advanced to the point where we can store a seemingly infinite amount of information. And during these 15 years, that’s what we’ve done.
When the Watson commercial mentions “2.2 billion weather sensors,” that’s only possible because we can actually record everything those 2.2 billion sensors are capturing. However, soon after we started collecting all this information, we realized no human could conceivably sift through it all in any reasonable amount of time. There was too much information and too many dimensions to it. To address this problem, researchers turned to decades-old concepts from the statistics world.
Going back as far as the 1950s, statisticians and computer scientists have developed techniques for using computers to analyze information and find patterns. By the 1990s, there were many papers published on the topic and plenty of successes. However, the techniques were often impractical because computers at that time were too slow to analyze and learn from large sets of data. When the new milenna came, that all changed though. Computer processing power had caught up and now these learning techniques could be applied to massive amounts of information. The term Big Data arose from the ability to store and analyze enormous amounts of information. Now in 2017, Big Data is being used successfully everywhere.
How can this help a digital-minded dealership?
Using computers to analyze Big Data was born out necessity. We had too much information to analyze ourselves and there were critical insights in the information if we had a way to find them. The question to ask as a dealer is: do I have a plethora of important information that’s hard to analyze? The answer in most cases is going to be internet shopper activity. Everyone has seen the stats: car buyers are visiting less dealerships and spending 90% of their shopping time on the internet. Car shoppers spend hours doing research before they step onto a showroom floor. Meaning, everyone is visiting your website before they decide to go, or not go, to your dealership. For example, in the last dealership I worked with, they were receiving 60 website visitors for every up. Now the important thing to realize is that each website visitor your dealership receives is leaving tons of useful information that can be analyzed. There are countless things to measure. How many photos did they click on? What order did they click on them? Did they click on the same photo multiple times? Did they start filling out a lead submission and stop? Did they hover their mouse over your phone number? What order did they view your inventory? The amount of things that can be tracked is immense. However, as we’ve seen, analyzing large amounts of information like this is commonplace in the world of IBM Watson and others. The high dimensional nature of the data is not a problem.
At a high level, if you wanted an A.I. system like Watson to sift through all this internet customer activity, you would have to give Watson a goal. For example, you could show Watson your internet customer activity and your sales data and ask it figure out what internet customer activities precede a car sale. Then Watson could start telling you which car on your lot will sell by looking at the Internet activity on your website. This technique of predicting sales, a type of predictive analytics, is not new and is becoming more popular as people continue to stay in their home and to do their shopping from their couches.
Expect to see more ads for IBM’s Watson and similar services in the future, as we find new ways to use computers to analyze monumental amounts of information to find valuable insights.
In your business, think about how this technology could help you and what questions you’d like answered. Think about all the actions people can take on your website and what that means about their buying interests and disinterests. There are a wealth of questions to be asked, and methods to answer them now, thanks to Big Data.
Attend Noah John’s session “Predict Your Inventory Car Sales TODAY With Big Data” at the 22nd Digital Dealer Conference & Exposition to understand the basics of predicting car sales to identify which cars in your inventory have a high likelihood to sell and which do not, and how predictions have helped dealers raise and lower vehicle prices. You’ll learn how to collect 20 to 40 times more valuable customer behavior information on your VDPs using Google Analytics.
Author: Noah John
Noah John is the Co-founder of Autoscores, a predictive analytics company that helps car managers predict their inventory car sales. Before Autoscores, Noah was a software engineer at EA Sports. He has BAs in computer science and mathematics from Rollins College and a MS in computer science from Colorado State University. His graduate work included machine learning, neural networks, and computer science education.