In his role as chief scientist at Atlantabased consulting firm Revenue Analytics, Jon Higbie helps clients make sound pricing decisions for everything from hotel rooms, to movie theater popcorn, to that carton of OJ in the fridge. And in the evergrowing field of data science where startups dominate much of the conversation, the 7 yearold company has a longevity that few others can claim just yet. They’ve been around the block a few times, and count behemoth companies like CocaCola and IHG among their clients. We spoke recently about how revenue and pricing strategies have changed in recent years in response to the greater transparency of the internet, and the complex data algorithms that go into creating a simple glass of orange juice.
You and your business partner at Revenue Analytics, Bob Cross, are veterans in the area of pricing and revenue management. What types of changes have you seen in data science over the years?
Our niche evolved from the deregulation of the airline industry in the 1970’s, when airlines had to decide how to price tickets. Their approach was to estimate demand and then allocate seats by fare class. That idea then spread to other travel companies like hotels, and they did the same thing. In the 90’s, other industries that our principles worked with, UPS comes to mind, wanted to apply these principles but they did not have the perishable asset component to the problem, and so we took a fresh look. That’s when the field started changing more from allocating perishable inventory to recommending prices for products. These days, that’s where most of our business is — modeling consumer response to promotional offers and price, and then recommending to companies what prices they should charge. Now, all of those ideas around price elasticity are filtering back to the travel and hospitality industries, and that change is largely driven by the transparency of pricing that’s available through the internet. Online, people can readily shop different rates for hotels, or shop prices for a camera at Amazon versus Best Buy, and companies have to be very scientific about how they set their prices.
A lot has already been written about Revenue Analytics’ work with CocaCola, helping them to optimize the supplychain process for manufacturing orange juice. It’s more complicated than people realize.
Jon Higbie: That was my baby in terms of the optimization model. And let me tell you that was a beast of an optimization problem. We helped them with their notfromconcentrate juice drinks.
I hadn’t given much thought to the complexity of data involved in producing orange juice.
Jon Higbie: It seems like such a simple thing, doesn’t it? I’ll tell you about the project process because it also gives you some insight into how our company approaches a new problem with a client. CocaCola engaged us because they had this grand vision I called “grove to glass,” which was going to optimize the supply chain and the pricing and promotions of all of their juice products. To accomplish that, they needed to integrate all of these disparate data sources into a common platform. So, we started with a revenue diagnostic. We looked at all of the different revenue opportunities: promotions optimization, price optimization , and supply chain optimization — and quantified the revenue uplift from each. We collectively decided to tackle the supply chain part of the problem, because quality control is incredibly important for CocaCola. A key partner was Coke’s R & D group which does all of the sensory profiling, so it was not just about minimizing the cost of the juice, but making sure that we acquire the right juice, send it to the right locations, and blend it with the right other juices to exactly optimize the taste profile for the consumer. And that’s what makes it such a difficult optimization problem. We built a strategic blend plan model to help them with their procurement process. Doug Bippert [Coke’s vice president of business acceleration] invented the name Black Book for the algorithm.
What are some of the common and persistent obstacles that your clients face when trying to collect and make sense of their data?
Jon Higbie: That’s a big question. Certainly for many of them a common challenge is that they may not have the capabilities internally to synthesize data, so they’re just storing it. One of our specialties is processing raw transaction data that’s often stored in an antiquated sales transaction system, a mainframe system, and data is not accessible for analytics. So we have to help pull that data out, and put it in a more modern platform which we can then integrate with tools like Tableau to create the insights. From that, we can also apply the algorithms in R to develop price response models. Also, some of the clients we’re working with are very siloed. Product development has their own databases and data sources, sales and marketing has their own — and the challenge is to pull data from these different groups into a common platform. The Black Book algorithm for CocaCola was a great example of that. We’re really good at knitting different data sources together to create a common platform for decisionmaking across all the functional groups.
You mentioned Tableau and R. What other tools do you use?
Jon Higbie: As a consultancy, we are tool agnostic. We try to fit into whatever stack the client has. Often they don’t have anything in the statistics realm so we recommend that they use R. Left to our own devices, we use Tableau, R, and Oracle Hadoop. We are adept at using anything the client wants — we use SQL server, Netezza, and for BI we use Oracle OBIEE. Those are just some of the tools that we use.
What’s missing from the tools that are available to you, that you would have in a perfect world?
Jon Higbie: Our philosophy is to make the very best decisions with the tools we have available, and also to time-box the solutions into projects that take 6 months or less to create these capabilities. We come in, and we are like a quick-hitting strike force. Focus on a specific problem, and build it really fast. So, we don’t fret too much over waiting until we get the ideal situation. We charge forward. In business, pricing decisions are going to be made whether you have analytics behind it or not.