Results from HT’s 2015 Lodging Technology Study indicate that big data in hotels is still in its infancy, but gaining ground. While only 31% of operators reported moderate maturity levels for use of big data and 9% reported high maturity levels, this is an improvement from 2014, when not a single hotel reported high maturity. According to HT’s 2015 Restaurant Technology Study, 30% of operators currently use big data technology and 24% are planning to add it in 2015, indicating that data tech is gaining steam in restaurants.
HT: What are some best practices for using data to predict trends?
BILL CARLSON, SVP Analytics, Choice Hotels: You need to start by ensuring a strong understanding of the key drivers of the variable or trend you are trying to predict. You need to assess if there are any changes in these key drivers that are likely to occur over the forecast horizon and, if so, assess options for how to account for these changes. Also, some outcomes are easier to predict than others; as a result be upfront about all the assumptions and document them accordingly. Finally, model various scenarios to associate a probability related with each scenario. It is always good to remember that even with all the big data and sophisticated modeling tools available, forecasting is both art and science.
DION HINCHLIFFE, Chief Strategy Officer, Adjuvi: Focus on investing in what I call “data supremacy,” meaning having a better and more updated understanding of the market and target customer behavior than competitors. This will be the priority for hospitality organizations that want to perform at the highest level. Master data management will also be a key technique at a tactical day-to-day level.
DEBRA MILLER, VP Business Analytics, Church's Chicken: Data is foundational to the business, but interpreting how data should be used to predict, and then applying management to that predication, is an art form. Data analysis is about the blend of art and science. Science represents the data and knowing the business is an art. You can’t run the business on just data; you have to have an understanding of the business as well.
HT: What is the importance of common platforms and/or
systems integration in order to compile valuable data?
CARLSON: Common platforms are critical in our efforts to leverage more data on a real-time basis. This is why having a common hotel PMS like Choice Advantage is so important to our success with analytics. With this cloud-based system across our portfolio of brands, we are able to readily access comprehensive guest information. Choice’s hotel and guest analytic databases also reside on a common platform that produces one version of the truth and enables various analytic efforts including proactive alerts, predictive modeling, and integrated management dashboards.
HINCHLIFFE: It’s very important, though flexibility is important too. Businesses should have an enabling enterprise architecture that strongly prefers open solutions to make their data accessible. This used to be very challenging, as hospitality firms had to decide between having access to open data or using the best technology product, but increasingly vendors are getting the message and opening up. Insist on products that have APIs at the very least. Also insist on system integration products having a basic understanding of the hospitality domain, and that offer standard integrations for common hotel IT systems and hospitality scenarios.
HT How do you facilitate collaboration and break through departmental silos to ensure data reaches appropriate parties?
CARLSON: Aligning our analytics team to mirror key parts of the organization allows the analytics group to engage with the business and fully understand their priorities. These teams also understand the full array of analytics and voice of the customer tools we have available so they can customize solutions for the business. Examples of these customized efforts are building an attribution modeling solution to optimize digital media spend and developing an alert system to identify individual hotel performance opportunities for our Franchise Services group.
HINCHLIFFE: There are now tools and approaches that can greatly facilitate communication and collaboration, especially in the very disparate geographic silos that exist in the hospitality industry. These include enterprise social networks, data aggregators/search engines like Xen.Do, and corporate portals, if used well. But it also takes a sharing culture, and that requires a commitment to transparency and data decision-making from the very top.
BRUCE BEDFORD, VP Analytics & Consumer Insight, Oberweiss Dairy: Data, analytics and reporting aren’t about technology at their core. Many people get confused and they think they are, but ultimately it’s about people. At the end of the day the data has to support the people doing the jobs. It has to support the activities that people pursue in order to help the company achieve its strategic objectives and stimulate people to ask and answer questions that might not necessarily have a direct impact on today’s business, but could impact tomorrow’s business. The data has to facilitate some aspect of exploration and creativity. At Oberweiss, our technology is about transparency. If someone wants a report, all they have to do is ask. It’s important to have a culture of data transparency and not being scared to share data.
HT What are some key ways hospitality executives can and should use data and analytics to set business strategies?
HINCHLIFFE: I find that most executives are lacking situational awareness due to the key data being submerged in their IT systems. I recommend executives demand a high degree of data accessibility/analysis and make sure that IT is providing simple, easy ways to monitor the health of the business, from rooms booked to customer satisfaction levels, in real-time if possible. This allows one to make decisions better and faster than competitors and detect important changes in market and property conditions. It also enables real forecasting using trailing trend data and statistics.
MILLER: Analytics will help set both strategies and tactics to support marketing planning, local planning, development, menu item optimization, advertising effectiveness, pricing impact, and operations. The resulting analytics will assist as each of these disciplines attempt to identify program effectiveness, opportunities, areas requiring change, what’s missing, and what can go. Ideally, all forecasts must be integrated — from unit volume sales of each item at the restaurant level, through to the supply chain and distribution channels. Then, there’s pricing, the competitive set, and the impact of ops scores. Forecast models can be extremely complex and expensive, so go for keeping it simple. Make sure you have the proper data, in the proper format, that it’s easily accessible and, most critical, that it’s clean.
HT: What are the benefits of using data visualization to make information easily digestible for all employees?
HINCHLIFFE: The biggest benefit is better use of data to make decisions. Often, it’s too hard to access or too difficult to interpret the data on hand for workers to routinely use it to make decisions for the business. With effective visualization, one can take an organization’s data flow and make it much more actionable. Combined with good user experience and availability/delivery on mobile devices, hospitality organizations can turn data into real value on a more regular basis, particularly if business processes are designed around the data feeds themselves.
BEDFORD: One of best ways to present information about attrition to people who don’t understand the underlying analytics is to show it in a graph. There are classical ways to do this such as survivorship analysis, but data visualization is very important because the notion of survivorship analysis is quite complicated from a technical standpoint. If we tried to show that in tables and numbers we wouldn’t get any play at all, but because of the ease of understanding the problem when you look at it graphically, we’ve been able to make significant changes to the way that we promote service to customers that required sales force buy-in. We couldn’t have done that if the graphs weren’t so compelling visually.
HT What do you think is the most misunderstood aspect of data analysis?
HINCHLIFFE: That it requires a math or technical background, or that one has to wait for the IT department to enable data gathering and analysis. Self-service SaaS can really help with this, though I always recommend trying to get the IT department involved to help. But these days, it’s not absolutely required, though one must perform due diligence to ensure corporate data is kept safe, private, and secure.
MILLER: Unless you are someone who works with data, you don’t fully appreciate the time it takes to gather the data, perform integrity/reasonableness checks, cleanse the data, analyze, prepare the fewest number of charts/visualizations to communicate your findings to a diverse audience, with diverse expectations and make sound recommendations. I like to use the analogy of data analytics to that of a plumber. The business analytics team is like plumbers. We’re wading through the “clogged drain” in the muck of numbers to get to running water. When everything works well, no one knows we exist, but when something goes wrong, everyone is aware that there is a problem. The plumber then becomes a crisis manager.
BEDFORD: People have expectations of data and analytics that are unrealistic. The output of an analysis is only as good as the input data and the appropriateness of the analytical tools that are used. The first problem is that people might use the right analytical process, but feed the wrong information into it. The second problem is if there is good data, but an inappropriate analytical technique is applied, the result is an outcome that isn’t necessarily useful. To do analytics properly you need to have the right data and systems, but you also have to have people who are trained appropriately and disciplined enough to follow the right procedures and apply the right analytical techniques. Don’t expect more from the data than it can provide, but make sure you’re using the appropriate analysis for the data you have and the question you are trying to answer.