NB: This is an article by Kelly McGuire, PhD.

As we approach a year that will very likely be marked with more uncertainty, increased competition from all directions and even more pressure to perform, many hospitality executives are making a New Year’s resolution to take better advantage of their organization’s data. With all of the buzz and hype around data and data science, it’s easy to resolve to “get more of it”. It’s harder to actually do it.

So, instead of letting your resolution to become more data-driven go the way of eating healthy and getting more exercise, let me kick off the New Year by providing some practical advice for a data-driven 2017.

Keep in mind that not every organization is ready to bring in hard core data science . Organizations first build an organizational commitment to fact-based decision making. To do this, they must get in the habit of asking data-driven questions. They need to grow their analytically-minded talent, and they should invest in some basic data gathering and analysis technology (beyond excel). Before bringing on data science, they need to get used to asking for evidence to back up recommendations and conclusions. Only in this type of environment does the investment in data science really pay off.

Whether you have already brought data science in, you are ready to invest, or you are just beginning your journey, here are six tips to set yourself up for a data-driven 2017:

#1: Start small:
Every organization’s data is a mess. Trying to bring order to the entire data sphere could take years. But good data is the core of good data analytics and good data science. Pick a smaller sub-set of data to work with from an area of the business that you either already know well, or have good contacts in who can help you clean and interpret the data. Smaller projects based on smaller data sets can have the advantage of faster time to completion, and therefore to value. So, divide your big vision into phases, with smaller data requirements. In our group, we always phase the delivery of our analytics solutions so the business gets some advantage quickly, and we keep momentum going. Sometimes the first phase is as simple as data visualization or automation. For example, we have been working on an analytical application to support owner recruiting efforts. In order to attract new listings, we need to demonstrate the potential performance we can deliver if the prospective owner were to allow us to manage their vacation rental property. We learned that existing process of pulling together performance data on comparable properties was highly manual and very time consuming. We needed that data for our modeling efforts anyway, so the first step in the project was to build an efficient database of current properties’ performance, and make that data accessible to the end users through a visualization application. Automating the current process was extremely valuable to the recruiters, and relatively easy for us to deliver. The recruiters were excited to work with the new tool, and were much more receptive to the analytically driven performance predictions that came next. Never underestimate the impact of data accessibility! However, try to make your data accessibility project as self-service as possible, you don’t want to be distracted by reporting requests during the modeling and application development process.

#2 Ask the right questions:
When it comes to data science, the questions are way more important than the answers. It’s very tempting to just dig in and see what the data says, but undirected analyses can end up exploring rabbit holes instead of having business impact. Make sure the questions you are asking are appropriate to the data set, and business-impacting. Try to get to the root cause rather than a symptom. Think about proactive, forward-looking questions like “what segments are most likely to respond to this upcoming campaign” or “what is the expected booking pattern from this feeder market” as opposed to reactive questions like “what was revenue last month” or “how much group business came from this channel”.Last year, I had a chance to hear the founder of Data Kind, Jake Porway, speak. Data Kind is an organization that provides data science services to non-profits by matching data scientists that want to give back with problems to solve*. In one recent project, they helped the Red Cross to reduce fire-related deaths. Jake said that when the Red Cross initially came to them they were looking for a cost-effective way to distribute smoke detectors. Data Kind could easily have solved this problem, but instead they explored the root issue Red Cross was trying to solve, which was to reduce the number of fire-related deaths. They ended up building a predictive analysis to identify areas at highest risk for fire deaths, and target those communities for the smoke detector program. This analysis addressed the overall goal, rather than a task within that goal. The founder made the point that organizations that are inexperienced in data science often aren’t aware of what’s even possible, and therefore, don’t always know the right questions to ask. This is really hard, but really important. Challenge your team to ask questions that get at the root cause, rather than identifying or analyzing the symptoms.

#3 Aim for high, measurable impact:
If you can’t measure your success, how will you brag about it?? In seriousness, your first few forays should be impactful, easy to understand and measurable. Within the start small philosophy from number one, think about a project that could have high business impact. You want to gain attention and build momentum. This matters at every phase of analytical maturity. When you expand capabilities or move into a new area, you need to find a high impact project with a high likelihood of success (and it doesn’t hurt to have a project stakeholder with influence to brag along with you.) In the early days of analytics at RCI, my boss’s team, then part of the Revenue Management group, had developed a reputation for good analytical work and demonstrated impact. They were operating in a highly manual revenue management environment, and saw the need for an automated analytical pricing solution. Unfortunately, there wasn’t a commercially available pricing solution suited to the Vacation Ownership Exchange problem, so the team determined that they’d have to build their own. They were able to justify the initial resource and technology investment based on their previous good work, and knew that this could be a high impact project because it was directly tied to revenue. The forecasting and optimization solution they built generated $11 million in the first year, occupancy rose and deposited weeks did as well. This was a great justification for such a major investment, and the team continues to innovate on this solution, improving accuracy of valuations and member trading power. This goes to show that this advice holds at any stage of analytical maturity. The success of this project generated the momentum the organization needed to continue to invest in in-house analytical solution development. The analytics team went from a group embedded in a line of business to an enterprise-level, shared services organization, which is where we sit today. We’ve built analytical pricing solutions for all of our lines of business and continue to expand into other areas beyond pricing. One of my 2017 resolutions is to find my next start small, deliver high impact project that can bring the team into a new area and justify further investment (it’s getting harder!).

#4 Document:
This is a small one, but it is pretty important for the long term sustainability of your data efforts. When you have a small and scrappy team that is moving fast, has many different responsibilities besides your data analytics program, or is new at the job, it’s easy to leave this step out. Unfortunately, several years later, when the person that built the report or analysis has moved on, it can be difficult to decipher the original builder’s thought process. I’m not suggesting that you build giant users manuals, but do think about some standards or practices, and do have your analysts comment their work. It’s time consuming, but will save a lot of effort later.

#5 Plan for the future:
Despite starting small, build a vision for the future. This vision will definitely evolve over time, but you do want to do the best you can to be sure that any decisions you make today will not limit you in the future. For example, if you are starting with inventory data, but want to eventually add customer data, do you need to ask IT for more storage space now? If you are looking at BI tools, consider not only whether they can handle your current volume of data, but also how well they will do if that volume triples (or more). Build a roadmap for your projects that outlines the phased approach, but also the eventual end state.

#6 Get educated:
There is a lot of hype around data science and data analytics. These terms have become such buzzwords that they are beginning to practically lose all meaning. The plus side of this volume of conversation is that awareness is building and executives are getting interested. The negative side is that they are now asking their teams to “get some data science now!” causing a big scramble. It’s up to all of us to cut through the hype and find practical ways to bring data analytics and data science into our organizations. Read as much as you can. Find a few thought leaders whose opinions you trust, and follow them. Attend conferences and network with your peers. Pressure your vendor partners to speak in business terms and provide practical examples. We’ll do our best to continue to cut through the hype in this blog, and I’ll also point you to the people that I follow, and the events I attend that help keep me educated.

At the Phocuswright conference back in November, Chip Conley, Head of Global Hospitality and Strategy for AirBnB, when asked what hotels should do to stay competitive with AirBnB, said that for hotels, “data science would be the new revenue management”.

This comment really struck home with me. I don’t think he meant that data science would replace revenue management, but rather, the evolution of data science in hospitality would be similar to the evolution of revenue management. Twenty years ago, hotels had barely heard of revenue management, and today it’s become practically a must-have, and a core component of most hotels’ strategies.

The same can happen for data science. The next opportunity to drive meaningful performance will come from applying data science to understand the guest, optimize operations and drive revenue and profits. As with revenue management, those companies that are able to truly understand and embrace data science will have a competitive advantage. Also as with revenue management, while the core data science techniques might be common across hospitality, the “magic” will come in the application. It will come from the strategies that are developed and the decisions that are made based on the results.

We’re here to help you get your data science program up and running – and to help your team understand the applications and techniques based on our experience. We look forward to your comments, feedback and insights.

*I strongly encourage any data scientist to look into volunteering for Data Kind. It’s a tremendous way to use your data science talents and skills for good. And it might help you meet a New Year’s resolution to give back!

Kelly McGuire PhD, is Vice President of Advanced Analytics for Wyndham Destination Network

Kelly leads a team of data scientist and developers that build custom analytical solutions for Wyndham’s vacation rental companies and for RCI. She is the author of two books on hospitality analytics: “Hotel Pricing in a Social World” and “The Analytic Hospitality Executive”.