‘Big Analytics’ for Hospitality


In last week’s post I proposed that the hospitality industry move past debating whether they have a big data problem and move towards admitting that they have a big analytics opportunity. In this post, I want to give you some more specific examples of the benefits of “big analytics” (and “big data analytics”).

In Natalie Osborn’s SAS post from earlier this month, industry experts weighed in on the opportunities and challenges associated with analytics in hospitality in today’s market. The overriding theme was around accessing data and turning it into meaningful information quickly.Big analytics is all about speed to decision making.

For hospitality, big analytics comes into play in two high-level areas:

    1. Making analytical decisions in real timeThe guest is standing in front of you, on your website or even just passing by, and you need to get a relevant message to them right at that moment. Your manager is on the floor trying to figure out what do next based on current operating conditions.
    2. Running scenarios, evaluating options, and testing alternatives. Even if you don’t have to make a decision in the moment, faster answers give you more time for analysis. Multiple runs with adjusted parameters or dynamic what-if analysis provide opportunities to do a more complete evaluation before making a decision.

Big analytics for real-time analytical decisions
Personalization has always been a hot topic (and big challenge) for hospitality companies. Delivering that memorable experience, which cements loyalty and increases return likelihood on a mass scale across huge global enterprises with high line-level turnover, is a seemingly insurmountable challenge.

But, if you can interpret behavior and deliver a relevant recommendation to the employee (or device) that the guest is interacting with while the guest is right there, “mass personalization” becomes achievable. If you get it right, you have a “surprise and delight” moment. If you get it wrong, you have an artificial, seemingly scripted interaction that makes the guest feel like one of the crowd.

Big analytics lets you run predictive models based on current, observed guest behavior compared to their past behavior (or that of similar guests) to determine what you can offer, mention, or provide to encourage the behavior you want, in time to make it happen. You can uncover cross- and up-sell opportunities, fill empty restaurant seats, or just wish someone a happy birthday or a welcome back — all right in the moment.

Big analytics also comes into play when your guest is on your website. Matching what you know about them, or how they compare to previous visitors, with their current click patterns can help the analytics predict what content should be surfaced to maximize their likelihood to convert. Extend that to the mobile strategy with location data (and, by the way, now we’re starting to get into some big data — I’m just saying…).

What about operational managers on the floor who want to make real time, proactive decisions about what gaming tables to open and close, how to deploy staff or which restaurant to send guests to? You don’t have time to wait three hours to calculate an updated forecast, when you are trying to decide whether you can save labor cost by cutting a few servers early. Is gut instinct enough when high revenue impact decisions like raising and lowering minimum table bets are on the line? Big analytics helps you to calculate the advanced predictive models that support your management decisions at the speed of business.

Big analytics for scenario testing
Many hospitality companies may argue that they either aren’t ready for real-time or don’t need it. Whether the scenarios I described above are relevant or achievable for your organization or not, there’s a very strong argument for using big analytics to help your analysts make better decisions faster. After all, good analysts are in high demand, so increasing productivity — and making better decisions — with your existing analysts will have a huge positive impact on your business. Here are some hospitality examples where big analytics moves the needle, even if “real time” isn’t the goal.

As the decisions that marketing departments must make about their campaign plans become more complex, marketing departments are turning to optimization to help. Marketing optimization algorithms output a contact strategy for each campaign that meets a campaign goal (maximize response, minimize costs), considering all relevant campaign constraints (opt-in preferences, channels, contact strategies, guest eligibility, prevailing rate, blackout dates, budgets, etc.).

As the number of available offers increases, along with the size of the customer database and the number of channels, so does the size of the problem. Without big analytics, the answer comes back in hours. This could be fine, if you don’t have any options within the problem. Frequently, however, the parameters of the problem have ranges (guest can be contacted between two and four times a month, channels can be added or removed, eligibility can change). With big analytics, you can run multiple versions of the problem in the same amount of time, and determine whether relaxing or removing a constraint gives you an even better answer.

Revenue management is the hospitality industry’s classic big analytics problem. To come up with a pricing recommendation, detailed forecasts are calculated and fed into a complex optimization algorithm. The detail and complexity requires intensive processing power and many passes at the data to solve. In order for results to be relevant, at a minimum, these forecasts and optimizations need to be able to run overnight so that prices can be updated once a day. If it takes any longer, with new reservations coming in all the time, recommendations will be out of date before they are produced.

Without big analytics, even to only meet this overnight goal, revenue management systems would have to make sacrifices and take shortcuts — summarizing data, restricting forecasting methods, and replacing true optimization with heuristics. With big analytics, harder, more complex problems take minutes or even seconds to run. Revenue management algorithms can utilize more and more detailed data, forecasting options can be expanded, true optimization can be utilized, and the entire process can run more frequently, providing better pricing recommendations faster.

Even better for revenue managers, with big analytics, scenario testing and “what if” analysis is now possible. Revenue managers often see the need to override the system based on their unique market knowledge (or demands from the general manager). Without big analytics, revenue managers make the adjustments and have to wait until the next day to see the impact on price recommendations. With big analytics, revenue management systems can support “what if” analysis. With the push of a button, revenue managers can see the impact of parameter overrides on the highly interrelated pricing problem before they put the overrides into production. Not only can they better support their own decision making (and learn how the system reacts to certain types of changes), but they also have evidence to support their position when they disagree with the GM’s demands!

Clearly, big analytics presents big opportunity for hospitality. These are only a few examples. Can you think of any other areas where better answers, faster, would benefit you? We’d love to hear from you!

Have you seen the “Hospitality research you should be using now” webcast?

This originally appeared on the SAS blog, The Analytical Hospitality Executive.

Thanks to allanalytics.com

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s