No Room at the Inn… or IS There?

No Vacancy sign

How to use a powerful technique (stratification) to find answers hidden in a large pool of data

Suppose you are a hotel manager and your occupancy rate for the first five months of the year has been steady: 72% – 71% – 73% – 70% – 71%. Corporate headquarters has just announced large incentive bonuses for all hotel managers who deliver an occupancy rate of at least 85% in the coming month.

What will you do to get to 85%?

I’ve been posing this question in seminars for years, and heard both obvious and unorthodox answers:

Reduce prices 15%
Partner with local vendors / companies to offer package deals that include meals, entertainment, etc.
Shut down one wing of the hotel and only count available rooms in the occupancy calculation. (Now that’s creativity!)

While many businesses certainly do take a ready-fire-aim approach to solving business problems, this is a tailor-made application of the analytical technique known as stratification — breaking a large data set down into component classes to see what can be learned. For example, what if we broke the overall occupancy rate down by day of week and got the below:

This gives an average occupancy of just over 71%, which is line with the year-to-date data. It also demonstrates crystal clearly that weekends and weekdays are dramatically different – which is the profile for a hotel in a business district. Think about the impact of the idea to reduce prices by 15%- it is a lose-lose proposition. Since the hotel is basically full 4-5 days per week already, a 15% reduction will cut into revenue-per-room without adding significantly more guests- ouch. And a 15% reduction on the weekend is not likely to draw a lot more people to a business district. This is a great example of the fallacy of proposing a global solution to solve a specific problem — low weekend occupancy.

What if the stratification by day of week didn’t reveal anything useful? Still valuable information because it enabled us to rule something out as a potential cause for variation. But since it doesn’t solve the problem of getting to 85%, we can stratify other ways. Suppose you choose to look at historical data stratified by month and got this picture:

This would be representative of a hotel that is in a summer resort location; jammed full in the hot weather months but steady at a lower occupancy rate in the off season. So, if headquarters is telling you that you’ll get incentives for hitting 85% in June… what will you do? Spend time and effort to set up package deals with local vendors? Not unless the deals are profitable on their own, because you don’t need it to get to 85%. Close down a wing of the hotel so the occupancy denominator is smaller? Only if you want to take a significant revenue hit to meet a target you are certain to meet anyway.

So is the “solution” to sit back and do nothing? Of course not! You need to get busy creating a narrative that supports the fantasy that the occupancy rose in June specifically as a result of the actions you took- and then taking credit for the rise when it happens. (Sorry, was that out loud?)

The point is that stratification is an extremely useful technique, whether executed via sophisticated algorithms on large and complex data sets or via pivot tables in Excel. It can help you target specific problems and formulate specific solutions… and hopefully get those occupancy incentives!

—More from Ralph Smith

Stratification is one technique Orion teaches in workshops like:

Each of these workshops can be customized to meet your company’s specific goals.