Before I get into this week’s topic, a quick reminder that I really need to hear about what YOU have seen that is new or innovative in hospitality tech. Readers often ping me AFTER I write about a topic to tell me about some product or company I should also have looked at. It’s much more useful if you tell me BEFORE I write about it! And indeed, cool new companies I hear about from readers are often my inspiration for a topic. My contact info is at the end of this section, so please reach out … I’m always happy to spend time learning about new companies or products.
Also, while I usually talk to at least some of the more established players when preparing a topic (and often mention them), the main purpose of my column is to expose you to companies and products that might NOT be on your radar already. In most cases, you don’t need me to tell you about the industry leaders. If I don’t mention them, it isn’t that I don’t think they have good products or that you shouldn’t look at them or buy them. It’s just that there’s not a lot for me to tell you that you don’t already know or could find out easily. If you’re happy with an established player, great – but I still encourage you to find out about good ideas from other companies. When you find one, you can talk with your preferred provider about getting that idea onto their development roadmap, or perhaps partnering with one of the companies I highlight.
Now on to this week’s topic…
Revenue management systems were introduced by airlines in the 1980s and first appeared in hotels around 1990. It’s well proven technology which, used properly, will increase revenue for most hotels. A 10% increment is not uncommon for a hotel that implements a true revenue management system. Yet 30 years later, most hotels still don’t use automated revenue management. A recent study commissioned by Expedia says that only 15% of lodging properties globally are using revenue management technology. The usual reasons we hear are it’s too expensive, it’s too complex, or simply “I don’t trust technology to do as good a job as I can.” Sometimes hotels are fooled into thinking they already have revenue or yield management because they bought the optional module from their property management system provider. These are usually reactive modules, which take actions too late to be very useful, and few if any would be considered appropriate by anyone who really understands revenue management.
If your hotel is one of the 85% without a revenue management system, it’s time to take another look. First, you’re almost certainly leaving somewhere between 5% and 25% of revenue on the table. While 10% is a common rule of thumb, I have seen much higher numbers when a hotel was completely unautomated for revenue management and ALL its competitors were already fully automated. In a world where pricing changes minute to minute and online pricing is fully transparent, you are at a big disadvantage if your competitors can react instantly and you can’t.
Second, there are a lot more options than there were a few years back. Some of the more recent entrants focus on ease of use, simplicity, low cost, and understandability. While there are still “black box” revenue management systems that ask you to place blind faith in their actions, it’s getting much more common to see systems that are intuitive and explain their recommendations (and the reasons) in plain language. Most of them will let you override their recommendations – and the better ones will tell you after the fact whether that helped or hurt, and how much.
Early revenue management systems worked by trying to forecast demand, and then set prices to optimize the hotel’s supply of rooms and rates against that demand. If a hotel could forecast strong demand well in advance, then it could keep rates high until that demand actually materialized – which only happened after competitors had sold out of lower rates. Knowing demand would eventually prove strong, the hotel could count on their competitors eventually selling out first, meaning the hotel could capture a premium in the days and weeks just before the peak period, when late bookers realize they had very few choices. It’s unnerving for a GM to look three months out and see nothing on the books, but often that’s the right decision. In periods of extreme demand that can forecasted far in advance, a hotel could sell all of its rooms at the highest rate in the last few weeks, and never sell any discounted rooms. Meanwhile, most of its competitors might sell 2/3 of their rooms at discounts before rudimentary occupancy triggers told them to stop.
From 1990 to the early 2010s, revenue management systems didn’t really change much, continuing to use these demand forecasting approaches. To be sure, they evolved around the edges to improve things like length of stay controls and to better deal with rate fencing and third-party channel management. But the core algorithms were still built on the basic principles from the 1990s (despite marketing claims of secret sauce, patents, and the like).
Only within the last five years have we seen real changes in how revenue management systems work. What’s different? Well a lot, but the biggest driver has been the ability to capture and process huge amounts of data at low costs.
I looked at both some of the established players and newcomers this week to see what was changing. I’ll highlight some of the things to look for and note some of the companies that have innovative approaches. But this isn’t an exhaustive review. Just because I mention one or two companies doing something, doesn’t mean that others aren’t – there are way too many products in the market now for me to have examined them all in depth. But I hope some of these trends might help inform the questions you might want to ask.
The biggest shift I have seen is the incorporation of many more types and sources of data. The theory, obviously, is that the more relevant data you can bring to bear, the better optimization you can run. And whereas the early revenue management systems primarily looked at historical booking data from the hotel central reservations or property management systems, today there are myriad sources for things like market demand data (national, regional, urban area, and even neighborhood); competitive pricing data; external factors such as weather, airline schedules, travel restrictions, medical travel restrictions, or special events; consumer perceptions of hotels; and highly detailed objective descriptive data for hotels within a competitive set.
On this last point, for example, Beonprice collects about 350 attributes for each hotel in a competitive set: things that matter to at least some segments of guests, such as the size of guest rooms, the number of restaurants, or the presence of a fitness center. It then pairs this with lexical analysis of online reviews and review scores to establish a quality index for the hotel. This can then be compared to the selling price to see graphically the price-value relationships. The driving factors are calculated by segment, since the attributes that matter most for a solo business traveler may be totally irrelevant for a couple or family. Numerous academic studies that show that consumers will be satisfied paying a higher price for a higher value product, but this is the first revenue management algorithm I’ve seen that is built around the concept. Curiously, the analysis has shown many hotels that their true competitive set was not the hotels they thought!
LodgIQ ingests more data of the types mentioned above than most systems and uses machine learning to find the key relationships. While others also use machine learning, LodgIQ takes the statistical approach to new levels, basing their design on the approach (proven in high-frequency trading in financial markets) that machines with enough data and processing power, and able to make and implement decisions, can beat humans consistently. With a laser focus on measuring accuracy and learning from the outliers where they missed the mark, they claim they have achieved 95% accuracy on forecasting 90 days out. Their approach also enables them to tell a revenue manager WHY they made a particular recommendation, for example that city demand data accounted for $5 of a proposed price increase.
There is also a trend toward faster, more frequent analysis. Where historical revenue management systems often ran optimizations once a day, and some have increased this frequency to a few times a day, we now see companies like Pace now doing it hourly, and Atomize claims they can do it in real time, reacting to every new reservation (provided they can get a real-time reservation data feed). In yesterday’s analog world, this probably didn’t matter, but with more and more hotels optimizing more frequently, speed of reaction can become critical especially on high-demand days or for hotels where demand may materialize or disappear very quickly.
A few trends are serving to make revenue management more approachable for smaller hotels, for which historical options may have been both too expensive and too difficult to understand. Machine learning can be used to compensate for a lack of historical demand data. Atomize, for example, requires no historical data to get started, and LodgIQ can be deployed this way if needed, relying only on external data for the initial months. In both cases, the system can start collecting data as soon as it is installed, and after a few months, there is enough data to start to make good recommendations.
Some of the newer systems are much more cost-effective for smaller hotels, both in terms of subscription service fees, and in the custom work required to get them set up and interfaced. Normalization and analysis of historical data was often a major factor in the historically large implementation fees assessed by traditional system vendors. Better automation of this, together with the ability of some systems to start without any historical data, has significantly reduced this barrier. For example, RoomPriceGenie charges $6 per room per month, with a minimum of just over $100 a month, meaning that even a 20-room property is above its minimum. It also can connect directly to channel managers, so there isn’t an additional interface cost incurred with the property management system.
User interfaces are being streamlined and simplified, so that an owner, general manager or rooms division executive at a small or medium hotel can see what’s happening easily on a tablet or phone, review recommendations and their reasons, and approve or override decisions. Atomize is designed principally to be used on mobile phones. This can be a big advantage when the person overseeing revenue management isn’t full time and sitting at a desk, but doing revenue management “on the run” between other tasks.
Many hotels use BAR (Best Available Rate)-based pricing. Historically, most revenue management systems determined an optimal BAR, and the hotel (or system) would then set other rates a given amount or percentage above or below the BAR. However, some systems – and particularly the heavily data-driven ones – are opening up new options. The BAR approach was efficient in the days when computer power was more limited, but suboptimal because the supply and demand factors for different rate products often depend on different factors. A $10 discount for a nonrefundable rate might be right on average, but totally wrong for the mix of business expected on a particular day. IDeaS has moved away from BAR-based pricing, and Beonprice does this as well. Having said this, there are hotels that may not be ready for so-called “open” pricing, where every rate is determined separately, so this functionality is typically optional. There may also be rules to enforce consistency even when the algorithm shows otherwise, for example ensuring that a club-level room is always at least some minimum premium above a standard room.
Cendyn (which acquired Rainmaker last year) and IDeaS both talked about optimizing around profit rather than revenue. This is an important trend, and a place where some of the more established players are ahead. However, I found only pieces of the complete profit puzzle in any one solution; we are still some distance away from a solution that truly optimizes the bottom line. Some of the systems can adjust for distribution costs by channel, which is critical. Cendyn looks at profit margin by category of revenue and covers multiple revenue sources besides rooms, which is also a big improvement; but it uses average margin per category, whereas profit is impacted by incremental margin (which varies both by type of product and by guest behavior). IDeaS has done quite a bit to improve profit contributions from group quotes as well, noting automated interfaces with Delphi to help sales managers quote the right rate for an opportunity.
Perhaps one of the most interesting companies I found in the revenue management space is Waylo. It’s not a systems provider for hotels, but rather a company that has built a business model around arbitraging the money that hotels leave on the table by not managing revenue as well as they should. They basically make bets on hotels whose prices they can predict will fall over time, selling rooms to the public at a discount to the hotel’s current selling rate, and only actually booking them later, after prices fall. The model works only if their predictions are right often enough – if prices don’t fall (or worse if they rise), they’ll lose money. The rates are fenced behind a login, nonrefundable, and don’t earn loyalty points or privileges, so hotels haven’t been too concerned about cannibalization.
Waylo uses serious analytics to basically do a better job than hotels of predicting the market. With 85% of hotels not even using revenue management systems, it’s not hard to see why this can work, but even with hotels that do have modern systems, they seem to be able to make it work. I had my doubts, but as I was writing this week’s article, Waylo was bought by eDreams Odigeo, which you can bet did thorough due diligence.
The question is, if Waylo can do this and make money, why aren’t hotels doing it themselves? Yes, hotels try to fence our discounts to avoid cannibalization, but fencing them from the entire OTA community (they are all moving in this direction) won’t be practical; they account in aggregate for too much demand. The answer is for hotels to get into the game and capture the revenue they have been giving away to competitors and channels by not pricing optimally. This means getting with the revenue management game. It’s time.