Will Revenue Managers Be Ready For Next Downturn?
There are at least a few industries where the price of the product is totally or substantially disconnected from the cost of production.
Oil is a good one, for instance. Even though low prices eventually tend to choke off “new” supply (exploration and drilling), countries dependent on oil revenue generally keep on pumping to make up in volume what they lose on price.
Air travel is partially in this category though airlines have done at least three things to help mitigate the problem:
1. Controlling the supply of seats in a highly disciplined manner;
2. Employing effective revenue optimization to offset money losing fares with money making fares (even on the same flight); and
3. Charging additional dollars for things that used to be free.
Airlines also benefit by being able to centrally control pricing decisions.
Hotels are arguably the most vulnerable of the industries subject to this price/cost disconnect phenomenon. Unlike oil, hotel supply is very “chunky” coming in surges and appearing long after the decisions to finance and build were made. And unlike airlines, supply and pricing cannot be centrally controlled—decisions are spread across thousands of individual owners and operators. We have also trained guests to expect lots of amenities and services in the base price rather than the a la carte pricing model employed by airlines.
Even an industry as decentralized and volatile as restaurants have had the good sense to pass on product and labor increases in a timely fashion—and from my perspective, very effectively. Think back five years and remember the going price for, say, an 8-ounce filet, or what it cost then for a lunch entree salad versus what it costs today. Have customers stopped going out to eat? Hardly—restaurant demand has increased materially over this same period.
Logic and methodology
Having observed the hotel industry for over 35 years, it’s hard to come to any conclusion about pricing other than this: the price that a hotel charges is more dependent on what its competitors charge than any other single factor. Only with highly unique and highly-prized assets is this rule of thumb inapplicable. Today, the notion of value-based pricing outside certain non-commodity products seems almost quaint if not downright antediluvian.
Talk about baffling circular logic: My pricing decisions are dependent on you—and your pricing decisions are dependent on me. Internally generated compression is a key factor in pricing—but groups and other base business that creates that internal compression is usually subject to some form of competitive pricing pressure.
This methodology seems to be partially effective in a growth cycle where market compression emboldens certain operators whose episodic intrepidness spreads like a virus to those around them. But it has led to disastrous results in a down market when fear-based decision making creates a vicious cycle driving down prices throughout a given market. And despite research-based admonitions that discounting does not create demand, we have seen this vicious cycle repeat in every recession and in every oversupplied market for as long as I can remember.
Systems and AI
There are, of course, revenue management software systems that take many external factors into account. But no matter how sophisticated revenue management software becomes, the algorithms still seem to be weighted in favor of this “chase-the-market” logic. These advanced systems are capable of processing more data faster, but at the end of the day, monitoring what everybody else is doing and then acting on those observations, often as not, accomplishes little in terms of increased net revenue. It’s like trying to get somewhere on a highway shaped like a Möbius strip. That may partly explain why industry gurus are still scratching their heads as to why, in a period of record occupancy and low supply, room rate growth is essentially zero (net of inflation).
Is the advent of artificial intelligence the answer to our prayers? Well, since people create the instructions, rules, data inputs and governing algorithms, I’m unconvinced that AI will really be able to “self-improve” past human limitations? Yes, ever more data can be processed and faster than before. And yes, these new systems can learn from their mistakes. But how effective will machines be in determining what data they consider and what weight to place on it? Can they ignore the rules that humans programmed relative to decision making? Or will the speed and sophistication simply result in a faster race to the bottom in a down market? And how will AI deal with continuously changing dynamics like the correlation between demand and pricing, which as of about five years ago, completely changed.
One also has to wonder about the downside of super-fast AI decision making. The stock market has seen the unhappy results of sophisticated programs acting on data movement in lightning fast ways. It’s called a flash crash—an event in electronic securities markets wherein the withdrawal of stock orders rapidly amplifies price declines. The result is a rapid sell-off of securities that can happen over a few minutes, resulting in dramatic declines, like what occurred on 6 May 2010, or 5 February 2018 when a cascade of automated stop-loss sell orders by trend-following investment funds starting in London fed into the steep rout in U.S. stocks. Everyone chased each other down the rabbit hole.
To be clear, I’m not saying that AI would cause hotel rates to fall precipitously like stocks in a flash crash. But highly sophisticated systems with complex algorithms might not necessarily make “intelligent“ pricing decisions when demand is declining and one or more hotel operators makes a decision to “buy” business through discounting. Also, what is mathematically correct is not necessarily logical if formulas are unintentionally designed to seek a sub-optimal answer, or are based on flawed inputs, or are designed around unapt rules or past “norms.” Will revenue managers and others who directly rely on AI driven recommendations be able to think and act independently if they intuit that what their systems are telling them to do is wrong or questionable? Will they have the training, experience, judgement and authority to do so?
My best-case scenario is that AI might help with distribution channel management, business mix and and net revenue calculations, and that would be a worthy accomplishment. But, the further we travel into the future, the less confident I become in the industry’s ability to change our flawed pricing paradigm—regardless of the tools we employ to do so.
Sadly, the evolving meme in revenue management seems to be: “I used to be indecisive, but now I’m not so sure.”