‘If you don’t understand that artificial intelligence will become a daily factor, then you’re not going to win.’
Aug 5, 2019 9:28:44 AM. By: High Tech Campus
MarketRedesign is a company whose mission is to revolutionise pricing intelligence. They have recently developed a machine learning tool that automatically calculates the best price B2B companies can ask for their products. They’re also located at High Tech Campus, so we thought it high time to get better acquainted.
Ruud Schmeink, CEO of MarketRedesign, likes to collect photo’s of bankrupt shops with a sign on their shop windows saying: ‘lowest price in town!’ The point he wants to make: pricing intelligence is crucial. If the price is too high, you won’t get enough customers. If it’s too low, you will go bankrupt. ‘What on earth can you ask for your product?’ Schmeink asks rhetorically. ‘That’s really a multi-million dollar question.’
The mission of MarketReDesign is all about finding the pricing sweet spot for their clients. Their unique selling point is that they don’t rely on gut feelings or sales experience, but on hard, solid data and the latest developments in artificial intelligence. Their most recent breakthrough is a ma-chine learning tool called PriceCypher that has completely automated pricing intelligence. Companies like DSM, Essent and Innogy are already using the tool with great success. ‘Our clients have been able to improve their margins with five to fifteen percent,’ says Schmeink.
Since about seven years MarketRedesign has moved their main office to the High Tech Campus. We talked to CEO Ruud Schmeink about pricing, artificial intelligence and how the two are related.
First things first: why did you decide to locate to High Tech Campus Eindhoven? ‘Because we have big corporates as clients it was important to us to position ourselves professionally. The High Tech Campus has the international exposure and the flavour of entrepreneurship we were looking for. It has great infrastructure and also is close to the Technical University of Eindhoven, which is a lifeline for us since we invest heavily in AI research.’
Which problems do you solve for companies? ‘Our idea was to make B2B companies more profitable by analyzing their pricing intelligence in a quantitative way. When you’re selling something, the first challenge is coming up with a price. The next challenge is that in B2B you have to deal with a lot of negotiating, bargaining and haggling. If a company buys a lot of your product, they will want a discount. So how much discount are you going to give and where will you stop? If you give too little you can lose the deal, but if you give too much it will eat up your margin. You have to look for the sweet spot. That sweet spot is calculabe.’
How? ‘When customers buy your product, they leave a pretty big footprint in your data system about how they behave in negotiations. We give our clients a graph with all the different price points for all the different customers buying the same product. Each price point is an actual negotiation between a salesman and a buyer. The graph could tell you for instance that a certain customer received a very cheap price, whereas his peers all pay more. Then you know that was a bad deal and you can give your salesman an alternative target price for the next negotiation. When we sort our client’s customers from large to small, we always see the same thing: 30 to 50 percent of their deals don’t generate any profit. With our graph our clients see the gap between what is possible in terms of pricing and what their salesperson actually asked.’
Isn’t it a bit unfair to give different customers a different price for the same product? ‘It’s actually fairer in the sense that the same groups of customers receive the same price. It’s only normal that larger companies who buy more or negotiate harder get more discount. Then there’s also the fact that B2B companies often have to deal with dishonest negotiation behaviour from buyers, like bluffing about cheaper offers from competitors. We’re kind of restoring the balance.’
Let’s talk about PriceCypher, the tool that uses AI to automate the pricing process. How does it differ from other AI applications on the market? ‘Many companies are into what we call ‘level one AI’ where you just have a simple model that looks at one or two variables, but doesn’t cover the entire business. If a salesperson has a question about something the model doesn’t cover, then the model is useless. What we have tried to create with PriceCypher is an ensemble model, that covers all the variables in the business like a patch-work.’
What kinds of data do you use to feed your model? ‘One is the size of the customer. Does he buy more and need more discount? Where is he located? What applications is the product used for? You can sell the same plastic for a car or for a table, but the price will be different. These are the primary variables. Then we link them with behavioural data like pricing aggressivity, loyalty, or how the buyer reacts to underlying price fluctuations of energy and oil.’
What is the role of machine learning in this process? ‘What’s really important is to select the right data variables. We used to do that by hand, which would take months. Now Auto ML [Machine Learning] plays an important role to detect the relevant data from the complete database. We have developed our own Auto ML technology, which outperforms the available versions out on the market. For certain industries the tool has become so good at the selection process, that we are only one step away from completely automating the crea-tion of the model itself.’
You’re talking about an AI that creates another AI. ‘Yes.’
Is that going to take over a lot of jobs? ‘Definitely certain sales jobs are going to become obsolete. But it cannot replace the human factor completely. Especially in big negotiations you still need human brainpower, because there will always be factors an AI model cannot cover. If you know one of your customers has had a bad year due to a fire in one of his factories, that’s something no algorithm in the world can catch in a model. But the human salesperson will know and can decide to change the price a little. So I would use your human sales force on the few big deals and let the AI automate the pricing for the majority of small deals or online channels.’
Are companies ready to embrace this level of AI implementation? ‘It’s going to be challenging and change management is necessary. From the CEO right down to the sales people, everybody needs to understand that this technology is going to change the way pricing works. If you don’t understand that AI will become a daily factor that is going to automate work, then you’re not going to win. Make sure you’re not just admiring the nice graphs, but really learn to exploit the possibilities.’
Ruud Schmeink of MarketRedesign discussing the impact of artificial intelligence with prime minister Mark Rutte during a Dutch business delegation to Boston in July 2019.