2016 was a year of changes for our industry. What were the two biggest changes for you in 2016? What are you looking forward to in 2017?
From my perspective there were two developments in 2016 that were more important than the others. First, there was a big rule change from Visa. Second, artificial intelligence (AI) started to make a real impact on our industry. I’m looking forward to AI making an even bigger impact in 2017.
2016 was a breakout year for AI. It left the realm of science fiction and began affecting all of our lives.
Let’s start with risk. The year began with the Visa risk rule change. Visa Inc. bought Visa EU. The idea was to harmonize the rules between different parts of the organization. Would E.U. rules go into effect worldwide or would the U.S. rules become the new law for the E.U.?
The first official statement was unclear. Did it even affect our industry? When would it go into effect? No one knew.
By spring we had basically figured it out. The acceptable risk levels would go from 2 percent to 1 percent worldwide. There would be no grace period for fines. The E.U. rules were gone and the global rules had tightened, they were more restrictive.
The rule change seemed to present a “reverse Goldilocks” problem. It wasn’t bad for the really risky. It wasn’t bad for the barely risky. It was bad for people with “just the right amount of risk.”
If you were big and aggressive the new rules didn’t really affect you. You had already accepted your +7 percent chargeback ratio. It’s a cost of doing business. The new rules simply meant you set up twice as many merchant identification numbers as you had under the 2 percent rule. (We’re looking at you, aggressive dating companies. We’re also looking at your three hidden cross sales … unlike your customers).
If you were small and unaggressive the new rules didn’t bother you either. Maybe you are passionately serving a niche. You produce your own content. You doubt you will ever have 100 chargebacks and even if you did your ratio is 0.4 percent. It’s always been that way. The new rules don’t mean a thing.
Big, small ... no change. In the middle. That’s where the rule change was bad.
Maybe you run a well-known company with brands that are respected in the industry. Maybe you are out buying traffic. You’ve got to watch your ratio because you are definitely over the count of 100 each month. You were comfortably under 2 percent. But 1 percent … WTF!
For you, this was terrible news. You’ve had to decide between revenue and risk. You had to drop your revenue and your risk. It sucked.
The next big change came from AI. It started to affect innovative companies within the industry for the first time. They began to benefit in several ways from artificial intelligence for their business. 2016 was a breakout year for AI. It left the realm of science fiction and began affecting all of our lives.
It’s surprising how quickly new tech becomes an accepted part of our reality. Self-driving cars are not a big deal anymore. Most of us haven’t ridden in one yet but we won’t be shocked when it happens. The people who have say that you get over it in about five minutes. We adapt quickly to change. Advances that are difficult to imagine seem inevitable and not weird after you experience them. They become normal.
We’re now used to talking to computers. An AI understands what we say and responds with its robotic voice. One example, is the Amazon Echo. Google is making major investments in AI as are Facebook and Microsoft. I’m setting up their competitor to the Echo, Google Home, this week. It’s rumored that Apple has been developing a similar AI-driven home product for years.
Of course, the core of these products has been available in our mobile phones for years. Siri started as a joke and is now useful. Google’s voice-based AI interface on Android is following a similar path from irrelevance towards essential. Samsung just bought the makers of Siri for their next generation AI. They are putting it front and center in the new phones they are developing.
For me, AI has played an essential role in two key areas for our industry in 2016.
First, in 2016 our pricing AI chose, displayed and learned from over 100 million prices.
The goal of our pricing AI is to “deliver the right price for each product and each person at each moment.” Sometimes it’s higher than the standard price, sometimes it’s lower. The right price gets the user to spend more money than the wrong price.
We had all kinds of ideas we tested. Here’s one example of an idea, or, hypothesis. Give rich people a higher price and they spend more money. Poor people will spend more with a lower price. How did we test it? We looked at demographic data in the US. It tells you average wealth. In rich places, like near the Apple store on Central Park in Manhattan, the average income is $118,000 per year. At the other end of Central Park the average income is $28,000 per year. Close neighbors, big differences.
With conventional tools like A/B split testing and humans using databases and spreadsheets to analyze numbers it wouldn’t be that difficult to figure out what price is best. It would take some time. You would need a lot of focused energy. But you could do it. The problem is simply that the number of questions you could ask was very limited.
AI is removing that limitation. It opens up the field. The amount of questions we can now ask is basically limitless. AI goes even further than answering our questions. It asks new questions that we had not even imagined. It removes the limitation of human thought and energy. So, yes, we got the answer to the question we were asking. We also got many more answers to many more questions, too.
It turns out that rich people spend more on higher-priced cars, vacations, homes, clothes, wine and most other things. But when our industry gives them a higher price they don’t spend more. They spend less.
Revenue has two components: Conversion and lifetime value. The price has to optimize both to maximize revenue. A high conversion is great as long as the LTV stays strong. The reverse is also true. A high LTV is fantastic as long as conversions stay at an acceptable level.
High prices for rich people hurt revenue. Instead of waiting to find that out by doing a lengthy A/B split test and analyzing the results, the AI figured it out and adjusted automatically. It does this kind of work daily for thousands of questions. It learns. It rewrites its own code constantly. That’s how artificial intelligence works. That’s why it’s better than human intelligence for these types of problems.
Risk is the next place where artificial intelligence began producing great results in 2016. There are over one hundred different variables that are important to look at when deciding whether a transaction is likely to chargeback or not. The best solution is a cyborg. Human risk experts “strap on” artificial intelligence to better identify patterns and learn.
The early results are exciting. AI-assisted risk is able to find more chargebacks in smaller populations than traditional approaches to risk. That means you get more transactions through and refund fewer of them. AI helps people make more money. Complex process, simple result. As cyborgs, people enhancing their work with AI, we are more precise than ever before.
In 2017 AI will affect other areas that our clients care about. We’ve spent the last couple of months talking with industry leaders about their plans for 2017. Many of them want to increase revenue per customer. AI can help them meet their objective. There are a number of ways that it can do this.
In 2017 we’ll be joining other companies creating AI that will benefit the entire industry. We’re stoked to play a part in it.
2016 was full of changes. 2017 will be, too. We’re excited to help make positive changes for the industry in 2017.
Thierry Arrondo is the managing director of Vendo, which develops artificial intelligence systems that allow merchants to dynamically set prices for each unique shopper.
