It is possible for almost every hospitality company, no matter the size or sector, to bring some data science techniques or technology into the organization at some level. As an industry, we can make great strides in our ability to balance excellent guest service against revenue and profit goals with the right use of data science.  

 

We are a little bit concerned, however. With all of this excitement comes a lot of noise. In such a complex area, this creates confusion. The buzz around data science is generating two different reactions in most companies. Either it’s so confusing that executives are reluctant to invest because the value proposition isn’t clear, or executives are so excited about what they have heard, that they are demanding that their teams “go out and get some data science” right away.  

 

The second reaction is probably most common. And it’s this reaction that is making hospitality organizations feel a bit like the title of this article – desperate. Everyone is running out looking for something, but they aren’t exactly sure what that something is, or what to do with it once they get it.  

 

It’s very important for organizations to make the distinction between “getting some data science” and enabling more fact-based, data-driven decisions across their organizations at all levels. We must believe that this is what organizations are really seeking. So, inspired by the classic ‘80s movie starring Madonna and Rosanna Arquette filled with misdirection, misadventures, mistaken identities and newspaper classifieds, we will help you define and position your data-related challenges, so that you can identify when you should be Desperately Seeking a Data Scientist, and when something else will satisfy your data-driven desires.

 

What is a data scientist, exactly?

 

First, some definitions. With all the hype, the main thing that has become muddled is exactly what a data scientist is. In fact, data scientists have a very specific skill set that differentiates them from others that leverage data or technology to gain operational insights.

 

A data scientist sits at the intersection 

of three specific capabilities: 

 

Math, statistics or operations research expertise: Data scientists are experts at the application of advanced analytic techniques. They understand probability theory really well. They are, generally speaking, very well educated, definitely master’s if not Ph.D.-level degree holders in these fields.

 

Hacker mentality. Notwithstanding the negative connotation, data scientists can creatively and quickly build code to gather, manipulate, analyze and present large data sets. They are comfortable in open source environments like R and Python, as well as with more traditional hard core statistical applications like SAS. They also build the technology environment around the algorithms, including the data access and the visualization of the outcomes. These folks can live in Excel, but they’ve moved way beyond it. They are coders. They can build production quality applications.

 

Business experience. Data scientists have substantive business expertise that helps them translate the business problem to an algorithm, and present the solution in business-relevant terms. It’s not just about knowing the math and technology techniques; it’s even more so about how they define and construct the problem. A data scientist is very skilled at this.

 

You should already take two things from this definition of data scientist. First, this is a difficult combination of skills to find in one person. The skill set is rare and those that have it are in extremely high demand. Second, you may be thinking that while no one person may have all three, you have people or technology in your organization with some of these skills. You are right on both points.

 

Many organizations can go a long way in the direction of data-driven decision-making before needing to hire a data scientist. You can use a combination of your existing analysts and technology solutions that either facilitates analysis or contains pre-packaged data science algorithms to gain business insight from your data. Let’s take a look at how that could work.

 

 

Analytics Matchmaking

 

In the movie, Jimmy and Susan used personal ads to find each other, but sometimes unexpected people showed up. (Kids: Before the internet, we used personal ads to meet up. They are like a mashup of a match.com profile and a tweet – short descriptions of who you are and who you are looking for.) So, let’s place some ads for common hospitality analysis goals, and see who will answer our data-driven dreams.

 

RGD (revenue generation director) seeks intelligence about my hotel’s current 

performance against my pricing strategy. You can show me current pace versus 

last year. I like understanding my market position, beating the competition and 

making my bonus. Ability to help me understand if I am ahead or behind is a plus.

 

Revenue management does use advanced analytic techniques like forecasting and optimization.  However, this RGD is looking for a report that pulls together a couple of data sources and allows her to visualize key metrics together. It is transactional data, so it’s probably not huge and it is probably well understood. The RGD wants to compare a set of metrics, not predict an outcome, so there’s no hard core advanced analytics required. It could easily be delivered in Excel or a visualization solution. So, our trusty revenue analyst would be gearing up to meet our RGD in Battery Park.  

 

IHC (Independent Hotel Company) seeks GTS (Great Technology Solution) to make my personalization dreams come true. I need to increase conversion. 

You can make an offer if my guest is in a certain salary range, likes resorts by the beach, and lives near LHR. I’ll be eternally grateful!

 

This ad is tricky, because of the word “personalization,” which is commonly associated with data science. However, our IHC already knows which guests it is looking for and what to do when they are found. This is a one-time list pull from the guest database. Once again, an analyst could easily parse the data until they have the list, and pass that along to the campaign execution team.  

 

What if the IHC asked for “better segmentation” or “new segments”?  If the segments are not already defined, the solution requires a segmentation or clustering algorithm. Still, this one-time analysis executed on an existing guest database could be performed by an analyst with some statistical knowledge. It does not require stitching together multiple types of big data, and the algorithmic technique is covered in business statistics classes.  

Contrast that ad with this one: 

 

GHC (global hotel company) looking for BDS (big data solution) to personalize my guest experience. I want to increase engagement and conversions. You must be able to predict 

the best product to offer based on my guests’ profiles, their past behavior and the current context of their website search. Even better if you can do this across mobile, 

voice and onsite. Creativity welcome. Direct bookings, a must!  

  

Here we attract our geeky, but charming, data scientist because we are asking for a prediction (read: advanced analytic algorithm) based on a set of complex data (profile data, past behavior and current context) deployed in a real-time environment. It will require coding, and it will certainly require enough business knowledge to interpret the behavioral cues in a way that is accurate without being creepy. This problem is similar to what Amazon and Netflix do with their recommendation engines, and it is one that is being solved today using a variety of advanced analytics techniques, including machine learning algorithms. Machine learning is a subset of data science involving a specialized set of algorithms that teaches a computer to learn to solve complex problems. 

 

Making the Right Resourcing Decisions

 

We hope you are getting a sense of what you can do with your existing resources, perhaps augmented with some strategic technology investments. Many organizations are centralizing analytical resources, providing them with some technical support, and aligning them with business experts to create a data science “mashup” organization. Analysts with deeper math and statistics skills can be deployed across functional areas, and you can use those analysts with more technical, coder skills to create meaningful tools and visualizations to help the organization access and gain insight from data more efficiently. Setting this organizational structure up first will allow you to generate some quick wins on the roadmap to deeper data science capability.

  

Starting the organization on a path to more data-driven decision-making with your existing internal resources will also help you to determine when it makes sense to invest in a data science-driven technology solution, and when you really need to hire data scientists to create one for you. 

 

There are a few areas where the commercially available solutions are mature enough, and solve such complex problems. Here you can take advantage of someone else’s data science so that you can focus your energies on areas that are not covered by such solutions. For hotel companies, revenue management and reputation management are areas where the investment in your own time and resources to build a solution is generally not worth the effort. Recommendation engines are rapidly approaching this category as well. With SaaS delivery, increases in processing power, and a few decades of experience under their belts, analytical solution providers are creating solutions that are accessible and effective even down to very small independent hotels. Think carefully about the commercially available options before you decide to invest in building for yourself. 

 

So, ask yourself a few questions before you run off and “get some data science.” 

 

After all, you are looking for the perfect match to your data-driven dreams. 

 

•Will the right answer require statistics, operations research or probability-driven math (predictive or prescriptive)? Or is it business rule or business intelligence driven (descriptive)? 

•Is the data structured or unstructured? How much data is there?

•How will the results be analyzed and delivered? With Excel, or will something else be required?

•Is there a commercially available solution? Can I leverage someone else’s data scientists?

 

A data scientist typically tackles problems that require advanced analytics, have some degree of unstructured, large or unexplored data, and require some flexible and innovative technology programming. An analyst can handle report building and data pulls and static statistical analyses on transactional data. A technology solution can handle enforcement of business rules, common analytical applications and transactional data management. 

 

Rather than simply jumping on the data science bandwagon with no idea where it is headed, your goal should be to bring more data-driven decisionmaking into your organization. You can go a long way toward that goal with what you have today, and use your successes to justify future investments.