Pace Yourself When Utilizing Pacing Data
Monitoring and controlling pacing has long been a tactic in hotel revenue management. Essentially, properties can track how fast rooms are selling for a certain stay date (pickup) as a flow metric and how many rooms have been booked for a certain stay date (pacing or booking curve) as a stock metric and compare that to the pacing and pickup of several other entities (e.g., the competition, the market, their own property the previous year) to adjust pricing. If the property is pacing ahead of the desired booking curve, then raising the price will slow pickup. If the property is pacing behind the desired booking curve, then the price can be lowered to increase pickup.
As with many things in the vacation rental industry, this method was adopted wholesale along with the rest of the hotel revenue management techniques we see in the space today. And it’s popular! We’d be hard-pressed to find a property management company that didn’t use some sort of pacing method to make pricing decisions, forecast financials, or plan budgets.
But is the pacing method even the right method for vacation rentals? Or better still, is it even needed for vacation rentals?
Challenges to Implementing a Pacing Strategy
Vacation rentals face several challenges to implementing a pacing strategy that hotels do not. First, pacing is based on the idea that as bookings occur and occupancy changes for a specific property or unit type at a specific point in time (seasonally, day of week, lead time, etc.), insights can be drawn now to inform future pricing and availability restriction decisions for that specific day and property. However, the possible occupancy percentages for any stay date at a vacation rental are limited. It can only be 0 percent or 100 percent. While hotels can have incremental occupancy anywhere between 0 percent and 100 percent because they have multiple identical unit types, by definition, a vacation rental is a single property. A vacation rental only has binary occupancy: 0 percent or 100 percent, booked or un-booked.
Second, gathering enough historical data to draw significant and robust conclusions about a vacation rental is much harder to do compared to hotels. Hotels have multiple identical unit types. This means every booking day, hotels are gathering a data point about every stay date across multiple rooms. However, a vacation rental is only one unit so can only gather one data point every booking day for every stay date. If a hotel has 10 identical rooms, it can gather data 10 times as fast as a vacation rental. If it has 100 identical rooms, it can gather data 100 times as fast. Using the statistical rule of thumb that a minimum of 30 observations is needed to reach statistical significance, it may take a vacation rental 30 years to draw conclusions with the same level of confidence that it may take a hotel one day to.
Getting Around Pacing Method Obstacles … at a Price
To get around these obstacles, vacation rentals often employ a few techniques that make the pacing method more useful but also less accurate. First, vacation rentals and dynamic pricing software will use “horizontal occupancy” so that an individual property’s occupancy can go from binary to incremental. Horizontal occupancy is when the occupancy of a single unit is taken across a certain time period of multiple days. This allows occupancy to be shown as a percentage between 0 percent and 100 percent instead of only as 0 percent and 100 percent. It’s called horizontal occupancy because it includes multiple days on a calendar or expands horizontally across a calendar. For instance, you may look at horizontal occupancy over the next seven days, the next month, the month of October, or weekdays and weekends. While this may be useful for measuring large-scale trends and identifying potential huge misses, it assumes that different days are the same or at least similar enough that conclusions can be drawn about all included days. If the next seven days has a horizontal occupancy of 29 percent, then it could mean Friday and Saturday are booked or that Monday and Wednesday are booked (or one of 25 other possible combinations). The insight and subsequent action taken from the two 29 percent occupancy scenarios are extremely different. At a minimum, horizontal occupancy always requires further investigation before drawing conclusions and taking action.
Second, vacation rentals often create “clusters” of units to create vertical occupancy (or occupancy on the same stay date through multiple units). To be effective, the units within the cluster have to be extremely similar so that conclusions are cross-applicable to all units within the cluster. However, the more similar units in the cluster are required to be, the harder it is to include multiple units and the fewer observations that can be gathered.
Third, vacation rentals often rely more heavily on market data where there are many more data points and it may be easier to gain insight. These insights often sacrifice usefulness, as market data is much harder to utilize for vacation rentals than hotels. Vacation rental market data is often scraped, which introduces more opportunities for error. Market inventory is often variable, with changing market dynamics from day to day, and just as vacation rental properties are unique from other properties in their portfolio, they are also unique from other properties throughout their market. Market data expands the number of data points available to vacation rentals for analysis but also makes the data and analysis less reliable for particular properties.
So, if pacing and occupancy are so challenging for vacation rentals, what else can we do?
The first alternative approach is often used in conjunction with other methods and may be familiar to a majority of property managers: pricing based on the competition and a property’s competitive positioning to competing properties. Competitive-based pricing methods are employed by many properties, especially those in competitive markets with lots of similar properties vying for similar guests. Competitive pricing methods make a lot of sense since each individual vacation rental represents a very small percentage of total market share. So, most vacation rentals are price takers and not market makers. Every individual vacation rental is strongly influenced by the other vacation rentals in their market.
Another alternative approach is creating and utilizing a probability function for which one constantly tries to optimize the expected value of a particular property on a particular stay date. A probability function is often some sort of logit/probit model that uses several inputs to estimate the probability of each outcome in a binary situation. For vacation rentals, the two binary outcomes are booked and un-booked, and one of the inputs is always rate. By using this booking probability function in conjunction with a specified price, an expected value for each stay date can be calculated and maximized.
EV = P(b) * P
Expected Value for a Stay Date = the probability of getting a booking at the proposed price multiplied by the proposed price
A third alternative approach since the onset of a mostly digitally distributed product is monitoring and controlling booking velocity. Much like any other website, online travel agencies and direct booking websites can track views, clicks, and purchases in relative real time. Much like variable pricing on Amazon, the changing of prices on the stock exchange, or surge pricing for Uber, pricing increases and decreases can be made to raise or lower the booking velocity for a certain property on a certain stay date (or for certain kinds of stay dates). Changing the price allows a property to control its booking velocity and adjust it to the desired level. This approach is similar to occupancy pacing, but it focuses exclusively on controlling pickup and ignores occupancy on the books.
Finally, the last method is similar to managing booking velocity and clustering but utilizes a specific technique to manage prices for different properties. The tree-and-branch method uses a cluster of units, but instead of using the cluster to create vertical occupancy, it uses each unit’s booking velocity and booking order to inform pricing adjustments. Essentially, the cluster is given a base price. This base price can be moved up or down to increase or decrease the booking velocity for the entire cluster. Each individual property also has a modifier to the base price, which adjusts that specific property’s rate up or down compared to the cluster’s base price. This allows for the unique features of a property (like bedrooms, average review score, number of reviews, maturity, picture quality, amenities, etc.) to be considered and accounted for. A property’s base price modifier can be increased if its booking velocity is higher than the rest of the cluster or if it usually books earlier than other properties in the cluster. Likewise, the modifier can be decreased to raise a property’s booking velocity or to make it book earlier. The tree-and-branch method is a great way to incorporate more data points for more robust conclusions while allowing for individual property variability.
It All Depends
Ultimately, the decision to implement pacing for your vacation rental should be based on your individual circumstances, goals, and market conditions. Pacing can help inform decisions and improve a portfolio’s ability to optimize revenue, maximize occupancy, provide a good guest experience (by creating more lead time so guests are familiar with the arrival process), de-risk future revenue, and generally conduct sound revenue management. It’s important to assess your property’s unique characteristics and the dynamics of your local rental market to determine whether pacing is a strategy that can help you achieve your objectives. But also, be sure to consider alternative methods as well. Just as every vacation rental is unique, so is every revenue management strategy.