A tale of two start-ups

A friend of mine recently told me about i-Rates, an analytics start-up that is primarily focused on revenue optimization for the hotel industry. My interest was piqued, for I had once briefly worked at a revenue management consulting firm. But that’s pretty much where the similarities end.

In particular, my old firm was a professional services group whose senior leaders all had mathematical sciences backgrounds, while i-Rates is a software vendor whose management team is straight out of a John le Carre novel. It all starts with a Russian scientist and entrepreneur named Vadim Asadov, the president of NeurOK, a mathematics and software development group based in Atlanta and primarily staffed by Russians. At a beach volleyball game in San Diego, he meets Ira Vouk, also Russian, who happens to be an award-winning hotel revenue manager. The hotel revenue optimization problem intrigues Vadim and his colleagues, and soon thereafter, he and Ira have formed a company. A few months later, an American named David Callander is brought on board; prior to i-Rates, he had spent more than 20 years selling technologies ranging from long distance telephone service to interactive television to the hotel industry.

Callander was actually my first point of contact at i-Rates. Given his sales background, I expected a product pitch. Instead, I got an education in the complexity of the hotel business – who owns what, who pays for what, who provides what to whom, who sets the rates, how those rates are communicated – and the bewildering array of information systems that support all of this activity. Though it seemed quite confusing at first, Callander was kind enough to follow up with valuable resources – diagrams, white papers, web links – to help paint the picture for me.

I dug in, and fairly soon the situation began to make more sense. In fact, the more I looked at i-Rates, the more familiar the company’s story seemed. Surprisingly, this familiarity was because of my experience at Blue Pumpkin [1], a workforce management software company focused on the call center industry. Despite the very different domains, the parallels were striking.

Competitive Landscape

“The upper tiers in the industry have had revenue management solutions in place for the past 15 to 20 years,” Callander told me. “But mid-tier operators just can’t afford the license fees and implementation costs for the high-end systems that are out there – and this mid-tier is where the vast majority of the market is.”

When Blue Pumpkin arrived on the scene, there were two dominant vendors who had expensive, well-established solutions for large call centers that were willing and able to pay for them. Meanwhile, more than 90 percent of the market had no workforce scheduling software solution at all, despite the fact that agents typically constituted 60 percent to 80 percent of operating costs.

How/why does this happen? It usually goes something like this

1. The early adopters of data-driven software solutions are typically large companies. Some of them will be willing to invest the resources for such innovations, for they have the most to gain by improving on the margins.
   
   
2. Following the Willie Sutton rule [2], entrepreneurs make the seemingly rational choice to develop their solutions to target these large customers.
   
   
3. Because these large customers typically have sophisticated IT groups and specialized needs, software developed for them is often too complex, and too expensive, for smaller customers.
   
   
4. Once deployed with these initial customers, however, the vendors must organize themselves to support them, including training, feature enhancements, bug fixes, documentation and conferences. In the face of all of this, developing a cheaper and simpler solution is neither straightforward nor high priority, especially back when software development costs were far more expensive than they are now.

Betting on User-Centric Design

As a new entrant, i-Rates has worked hard to make its revenue management solution accessible to the much broader middle tier of the hotel market. This is partially due to the natural advantages of later entrants (better and cheaper development tools, no legacy code, the chance to learn from others’ mistakes). In addition, Vouk herself had previously spent nearly a decade as a revenue manager and therefore was able to a first-hand perspective on the needs of prospective customers to the design process. This in turn led to software that is less intimidating and easier to learn, lowering the cost of software adoption for prospective customers and significantly increasing the likelihood that the software is used effectively.

From its inception, Blue Pumpkin also had a strong focus on user-centric design and ease of use, and this had been key to helping the company get smaller call centers to purchase and adopt the initial product.

Shortcomings of Traditional Methodology Exposed

The traditional approach to applied optimization is to first build a forecast of demand based on historical data, and then make optimal decisions based on that forecast and associated constraints. However, the rapidly changing dynamics of the hotel industry – most notably the emergence of on-line booking channels — means that traditional forecasts based on years of history are even less accurate than usual. Thus, a key part of i-Rates’ value proposition is its ability to dynamically update future booking predictions and pricing decisions based on actual bookings to date, while utilizing machine learning techniques to recognize and respond to shifts in demand patterns.

Similarly, Blue Pumpkin’s older and larger competitors had developed optimal scheduling algorithms that were explicitly based on FIFO queueing assumptions. However, a phenomenon called “skill-based routing” had rendered the traditional steady-state FIFO models insufficient for many call centers, exposing a significant hole in the existing vendors’ solutions. This provided an opening for Blue Pumpkin to deploy a machine learning-based algorithm [3] that explicitly accounted for this advanced routing logic, which in turn served as a valuable point of product differentiation.

Moving Beyond the Initial Target Market

While the mid-tier of the hotel market is i-Rates’ current focus, its management team believes that the core algorithms that they have developed are quite general and could be adapted to apply elsewhere. While not currently planning to go after entrenched competitors at the high end of the hotel market, where switching costs are higher and decision-making processes are slower, the leadership team is already keeping an eye out for similar problems in other domains.

Blue Pumpkin had similar dreams. Over time, it acquired a consulting firm that focused on employee scheduling, predominantly in non-call center environments, and did a variety of pilot projects to test its software’s capability in other domains. Ultimately, the company successfully developed a feature-rich version of its product and made a successful assault on the high end of the call center market, while maintaining its original product line for small centers, which still make up a huge proportion of that market.

After many twist and turns and ups and downs, Blue Pumpkin was eventually acquired by a vendor with complementary software products targeted at the same market. From my perspective, it was a (somewhat) long road with a (reasonably) happy ending.

i-Rates, meanwhile, is still just a story in the making. But it’s definitely one I’m going to be keeping my eyes on.

Notes & References

1. In the interest of full disclosure, I worked closely with Blue Pumpkin from 1998 to 2003, first as an external consultant and then in-house as vice president for the Solutions Group.
2. Sutton is known, albeit apocryphally, for the urban legend that, when asked why he robbed banks, he answered, “Because that’s where the money is.” For more on his life of crime, see http://en.wikipedia.org/wiki/Willie_Sutton.
3. For more on the Blue Pumpkin scheduling algorithm, see www.aaai.org/ojs/index.php/aimagazine/article/view/1667/1565.