S2 E03: Donald Davidoff Talks Competitor Data

Welcome to our latest deep dive conversation on the topic of revenue management. Before we go on, if you haven't already done so please stop and give this podcast a five star review on whatever app you're using. We're making this content to educate the industry and your review will help us to reach more people. Now, onto today's show. Today's guest is Donald David Off, who needs the top introduction. Donald is the Godfather of revenue management in this industry. His chief executive of Reba, which is the fastest growing revenue management platform in the industry, but that's not why he's on this pond today. There's something that the industry has been confused about for some time and it's really come to the fore in much of the reaction to the revenue management lawsuits. There's a time and a place for compessors of data in multi-family and a pricing algorithm is not it. That's the subject of today's conversation. Now, to help us understand why Donald and I delve into our experience of multiple industries to talk revenue management algorithm design and to understand where you should and shouldn't be using comp data. As always, it's a hugely enjoyable conversation. We cover a lot of ground. I hope you find this as useful as I did and now I bring you Donald David Off. These conversations underline the need for revenue management expertise. If you're with an investor or a management company, you may not realise that you have access to revenue edge the industry's leading third party revenue management advisor. If you thought your own options were to rely on your platform's pricing advisor or hire someone in-house, think again. Revenue Edge brings you expert guidance on all the leading revenue management platforms without the overhead or bias. There you go to team, offering high bandwidth and unbiased insights. Want to see how revenue edge can elevate your strategy? Go to revedgesps.com. That's r-e-v-e-d-g-e-sps.com or click the link in the show notes. Okay, so Donald doesn't really need any introduction but I'm going to ask him to do it anyway. Hello, Donald. Hello, Dom. Yes, I'm Donald David Off. Co-founder and CEO of Real Estate Business Analytics. Been in the industry since 1999, which just means I'm old. As many viewers probably know, anybody who doesn't, I've always referred to this as a fly paper industry. Most people come into it randomly and then just stay. I came into it through pricing and revenue management. I led the team to build LRO and actually late 2001, Switzerland's size of the table went to work for Archstone as the industry's first ever vice president of pricing and revenue management. Long history with pricing and revenue management since then I won't bore everybody with that. Well, I guess that's something that you and I have in common is that we both came to multifamily by way of revenue management rather than the other way round, which is how a lot of people build that particular skill set. That's the main reason why I wanted to have this conversation as part of this series. I think when we think about particularly competitive data, it's really critical to think about it in the context of how revenue management works. What I'd like to do during this conversation is just unpack by using different examples, how revenue management systems work, how they help deliver success and critically the role that some competitive data should and shouldn't play in it. My general thesis is there's a time and a place for competitive data and I want to help people understand through this conversation where we think that is. Does that make sense? Yeah, it makes a ton of sense. I mean, certainly, you know, when I came into the industry, it over-indexed on competitive data because there wasn't a whole lot else to go on. And while there certainly has been some improvement on it, I would say that the industry is still probably over-indexes on Comp Data, especially for an industry that has a lot of natural pressure even from state and local zoning authorities to not end up with over-supplied. Yeah, it's said a little bit more about over-indexing on Comp Data just in terms of how the industry talks about this pre-revenue management, generally, how you see it now. Yeah, I mean, I think pre-revenue management, I observe two things. I mean, one is just a tendency to latch onto rules of thumb. So, you know, 95% occupancy was the magic number. In fact, I remember one of the interviews I did in, must have been late '98 or early '99, I was interviewing a property manager where there, you know, one veterans were 98% occupied and two veterans were 93% or vice versa, it doesn't really matter. And I remember asking them whether they thought about raising the rents on the unit types at '98 and the answer was, oh no, I can't do that because then my overall occupancy will drop below 95 and my regional, give me a call. And I thought, well, that's sort of interesting because, you know, the whole point of revenue management is to product segment, customer segment, and sell the right product to the right person at the right time for the right price. And so, ones and twos were different product types with different customer segments and yet, you know, because of that rule of thumb, people latched onto that. The other thing is I think the attention to comps is just sort of a natural human dynamic of not wanting to lose to a comp. In the absence of anything better to focus on, you need something to focus on. So, I think the industry just latched onto that. I took a lot in the early days to get even revenue management adopters to understand is the notion of the forecast, right? Like, don't act. I remember tell the CEO and the executive team at ArchDome, right? We need to get the operators to stop acting like the prospect in front of them is the last living breathing person who's going to rent. And if that's true, then you absolutely have to care about the comps because you don't want to lose some of the comps, right? But, you know, revenue management doesn't exist without a forecast, a demand forecast, and a supply forecast. It doesn't mean you can't price without a demand and supply forecast, but that's just a price algorithm. It's just a price administration system. Revenue management is about going where the puck is going to be to quote the famous Wang Gretzky quote, it's not about where the puck has been in terms of historical pricing or maybe it is today with today's pricing. And that's really, really important because, you know, what I was telling that executive team is we have to teach them to understand and visualize the right now imaginary, but a very real line of people behind the person in front of us who are going to come in the next day, the next week, the next month. And based on how long that line is, we'll make decisions. And in a lot of ways, data doesn't matter then because how long that line is and how many of them are buying from us, tells us how we're doing. You're really touching there on the sort of fundamentals of revenue management, which we really want to get into. In terms of context, again, you were part of the team that designed LRO for listeners Donald and I have this shared heritage of having worked for the same company back in gosh, the late 90s. You were right. You were right. I was in Atlanta. So yeah, yeah, which is a company called Taylor Solutions that had created and ran revenue management systems for numerous different industries. I think one helpful idea when you talk a little bit about when you designed LRO because obviously it drew influences from other business problems that had already been solved before Malsie family got on board. Yeah, one of the things that was unique about Talis because we competed with like saber technologies and others in the airline space of Talis space, et cetera. But one of the unique things about Talis and lucky for my career, I think yours as well, is we kind of had a specialty in Greenfield, right? Like how do you take these concepts and bring them to industries that are adjacent, but don't use it, hence the move into the multifamily rental housing space. So when we started the very first study, the natural tendency was to take as much from hotel revenue management as we could because it feels like we're a cousin of that. I mean, all the customer experience work going on, everything else, everybody always mentions what else is very, very natural. It turns out though that from a pricing perspective, the fundamental dynamics of demand and supply and how revenue management software works, we're actually very, very different. So in the hotel industry, a big piece of the puzzle is the metaphorical game of Tetris, right, where you take these different amounts of demand by arrival date, by length of stay, you try to get them to fit. And if you think about that, that Tetris game is really, really important. The classic example is, Dom, you call or come in and present for a rack rate Tuesday night one night's day. And somebody goes, "Whoo, I'm getting rack rate." Except behind you are three people in that business hotel who would come for a week, even though at a discounted stay, if I take your demand for that Tuesday night, then I can't take the person behind you for seven days, filling the vacancy of Friday, Saturday, Sunday night, because I can't house them on Tuesday night, I just gave away the last room to you, right? So that Tetris game is a big, big part of hotel revenue management. Well, in multi-family, we actually simulated them. And what we found was it's practically the minimums, right? Because, you know, according to three days of vacancy loss on an average three nights day, that's massive. But even a week of vacancy loss on a one-year stay is relatively small. And so we quickly realized that, you know what, the Tetris game of hotel and the way we design hotel systems isn't really going to work. Well, as I could have it, at Talas, we had built a couple generations of rental car systems, and I had actually worked very long on one of those. And the rental car vertical was more back than the one we had on Tuesday night. in their technology than hotels, so they couldn't use length of state controls the way hotel systems did. And so we have built more of a true dynamic pricing engine, setting a price for each day, and coming up with the right price based on what the demand and supply characteristics look like. And so, you know, little known fact, the guts of LRO, that very first system that I let the team the built, the belt did, the guts of it are actually much closer to a rental car revenue management system than to a tell revenue management system, at least as they existed back in the late 90s. Yeah, it's interesting. It's funny, I was having dinner with a bunch of property managers last week, and the conversation was about this view that I've developed over the years, which is that most of the most unreliable takes about multi-family, are ones that begin with they do this in hotel, so we should do it in multi-family too. Yes. Yeah, I mean, it's natural, right? I mean, you got buildings, you got people who come and go, it's very, very natural, and I'm not saying there's no lessons to learn, but I mean, the dynamics around competition, the dynamics around the physical characteristics of the demand and the supply are so different, I'm with you, I caution people, don't over index on hotels. Yeah, no, it's true. While we're sort of on the subjects of system design and also hotels, there's one example that I sort of reference back to a lot in terms of understanding this whole question about Compessor Daza. Back in, gosh, this was about 2007, I did this project over the course of a number of years with a mid-market hotel company. Now, as you know, the way the revenue management works in hotels is exactly as you just described it, right? We identify the bottlenecks, right? The times where there's a line of people that would pay more for this room, and we organize our controls so that we take the best blend of demand. Hotels have been doing that for, you know, decades by them, or at least a decade, most hotels. But we found most hotels in the US, they're not like the Marriott Marquis on Times Square, right? Most hotels in the US are sort of 80-ish rooms, they're sitting on some freeway somewhere. And there's this key thing about how they operate, which is that they very, very scarcely sell out. So we were working with this chain, and through analysis, we found that they sell out something like one night in 11. And so we had to design the solution where, if you imagine the scenario you just described, where there's no line behind the person who's. Yeah, yeah, yeah. And so the thing is, in that example, you know, the algorithm you design, it sort of consumes competitive data. It establishes a market condition. Most of the time it just follows the pricing in the local market. And it's only when the forecast is predicting that demand is above or below where it would normally expect it to be, that it starts doing things that deviate from the market. So it sort of flips the whole thing on its head. But what interests me particularly when I think about competitive data and the things that people say about competitive data, going back to your original point about overindexing on it. The thing I sort of learned from that is that putting competitive data center stage A, you only want to do it if, as you say, you've got nothing else to guide your decisions. But second, you need this guard rail that stops you from following your competitors when you shouldn't. And in that situation, which is very, I mean, it's not analogous to multi-family because multi-family properties generally don't suffer from that problem of, I can't fill my units. But yeah, it's interesting to think about it that way in terms of how you have to strike this careful balance between competitive pricing and. Yeah, I mean, it's. It's different from your own. It's another reason why, depending almost exclusively on comp data, it's such a bad idea in our industry. I mean, aside from the fact that it's looking out the rearview mirror instead of out the front windshield, which we talked about just a moment ago, it's exactly the scenario you just described, right? Most quality multi-family real estate, right? Certainly A and B class. Even some C class were the demands high enough. It's going to fill. I mean, the local boards that are proof zoning, approve entitlements, etc. It's really bad to have housing that isn't filled. In the hotel world, it's not that it's great to have an empty hotel, but remember hotels, especially in city centers, hotels are all about a massive day of week seasonality, if you will, as opposed to just week of year. So in downtown most places, it'd be impossible to build just for selling out on the weekends, because then you wouldn't have enough for the business people in the week. That's why the business people pay more in the week in the middle of the week, because it kind of covers for the downtown of the weekends. But once you built the supply, right, you start to get creative from a marketing and a pricing perspective on how can you avoid some of that weekend vacancy loss. We just don't have that problem in our world, okay? So the funny thing is the way you describe the bifurcation of that hotel market, right? The merits of the world are like what I just described. These 80, 90, you know, room hotels in tertiary and even lower markets behave the way you did. You just described where they infrequently sell out. That's actually the theory behind the original use of or lack thereof of comp data in LRO, which is in situations where we had very high availability and very low demand. You had to care about where the comp's rad, because you didn't want to lose that one person in front of you. But most of the time, you don't care about the comps, right? So I remember telling execs anytime they asked me a question about revenue management, I could answer it with the same two words. It depends. So how important is comp data? It depends. Right? It depends on the situation depends on the scenario. The reality is in most of multifamily housing demand typically equals or it seems supply. Right? The independence answer is most of the time it doesn't matter much at all. Yeah. No, no. So now you're touching on algorithms and we are going to come and deal with that in detail after the break. Radix is the source for complete, transparent and accessible multifamily data for professionals and consumers. Whether you're an owner, an operator or an advisor, Radix provides timely and accurate data through the entire life of a deal from sourcing an acquisition through to management and disposition. At the same time, renters get unprecedented access to Radix's metrics, helping them find housing that meets their needs and preferences. Transparent markets help everyone to learn more go to radix.com or click the link in the show notes. So at the end of the last segment we were talking or we were beginning to talk about algorithm design. Let's start off with a nice broad question. We've talked about several different business concepts. You've designed a few revenue management algorithms in your time. Tell us what are the essential elements of a good revenue management algorithm? Yeah, I mean a good revenue management algorithm looks at the business objective. Right? So a hotel or an airline or a car or company has a different business objective than multifamily. What do I mean by that? Right? A hotel has very short lengths of stay. So it's kind of a very short term optimization and then you're just rinsing repeating and over the course of a year you achieve your numbers. With rental housing you really have to look at like what do I want to achieve over the next year. So I always talk about optimizing rental revenue over the next 12 months. At the same time, you know rental housing also has some other interesting components like well what about the property that's going to be prepped for sale. Because all the sudden the business objective is no longer about optimizing over the next 12 months. It's how do I set up a rent role that's going to get me the maximum sale because a couple points of extra operating revenue aren't nearly as important as if I can get a five or six cap on a little more rent. Then you know that's going to pay off in spades on that deal. So again, you know what makes for a good revenue management system is it is it understands the business objective. As I mentioned earlier, it's not just looking at the revenue mirror is how to how do we get here is not just taking the temperature of today. You also got a forecast and you know demand and supply forecast component so that it is trying to get you to where the puck is going to be. And then it implements pricing recommendations in a way that can be marketed and promoted. And that way the process and the math all fits together. That's why I just to go back to some I said earlier, right, we designed something for the rental car industry different than the hotel, even though they had length of stay because they couldn't implement like a state controls. So we had to implement something that would work in their business process. And when we first developed LRO, we spent a lot of time in the field understanding how people sold how people least so that we didn't build a sexy math model that couldn't be implemented in the real world. Right. And so, I mean, you have a very interesting perspective on this like having having been in this sort of from the start, but I mean, let's talk about what are some of the things that make multifamily difficult. I mean, all of these industries have their nuances, right. Yeah. Yeah. Yeah. A few things make it difficult. I mean, the long lengths of stay on relatively small buildings, you know, 223 or 300 units. But, splinter. one bedroom, two bedroom, three bedroom. That leads to a really low transaction density. And so the airline in hotel world, which is where all this stuff started, right? Those algorithms inherently depend on very large data densities, right? Lots of robust statistics. So we had to invent things. In fact, I still think only the two systems I've been involved in designing have actually implemented what I call dynamic pooling, which is you actually look at how much data do I have for my one bedroom's at this property? And if I have enough, great, I calculate the statistic. But if I don't have enough, what do I do next? Do I add some two bedrooms to it in the pool? Do I go to other properties and add data, et cetera? So in LRO, we designed it to be able to add more data where it was necessary, but it was still very community-specific. The systems I've worked with now, learning from that lesson, and I'll like, oh, wait a moment, pulling three bedrooms with two bedrooms, that doesn't make a lot of sense. Why don't I pull three bedroom data from this property with other three bedroom data in the market? That's going to get a more robust statistic, right? There's a lot of detail and nuance in how do you deal with the small data sets. So that's one big problem right away. The other problem or set of problems with multi-family compared to the traditional airline hotel, et cetera, is we have a lot more special cases, right? In the hotel and car rental world, you're not renovating a unit in the middle of somebody's day. OK? You don't have a lease up in the hotel world, per se. Things like that. So special cases like lease ups and renovations are really important. Small count unit types. I mean, there may be the occasional penthouse or something in a hotel, but you don't have like the three bedroom versus the two in one challenge that we have. Also, the whole role of concessions. I mean, in the late '90s, when we were building LRO, everybody thought revenue management, net pricing, we were going to get rid of concessions, right? Makes me remember the '70s, '80s, where computers were going to get rid of paper. How'd that work out, right? So how did it work out with revenue management getting rid of concessions? I mean, not at all. Concessions are as prevalent today in the right circumstances as they were back then. And so there were a lot of things-- when we built the system back then-- there were a lot of things that we didn't know. Like, concessions were going to be here to stay. And so we would have designed differently if we'd known that. And then some of them, we even intentionally just avoided. You know, we're hacking through a jungle to create a V1. Well, we're not going to worry about the 4% of units or whatever the number was that are these steps. We're not going to worry about the 6% that are renovations and restores. I'm not going to worry about the 1% or 2% that are very low count unit types. So a significant chunk of the business-- not a majority, but a significant chunk-- was intentionally avoided in those early algorithms, because it just would have been too much to swallow. I guess the other thing that we didn't talk about yet is this sort of trade-off between complexity and the ability for people to understand how the thing works. I'm on the risk that I've seen when you consult with revenue management companies is that you encounter more and more details of the business problem. One way to solve that is to append more and more stuff to the algorithms that are setting prices. It's like a recipe that keeps adding ingredients to. But you do that at the trade-off of absolutely nobody can understand how the thing works at the end of it. I mean, how did you generally think about stuff like that? As you know, TALUS was very, very mathematical. I mean, we had PhDs and operations research and math majors and engineers, et cetera. So we spent a lot of time with LRO trying to simplify the complex. And I think there are a couple places we did a great job with it. And I think there's a few places we did. And I think one of the lessons from the legacy systems is how do you take the complex and make it simpler? I do believe, while there's some trade-off, I think it's a false narrative to say it's one or the other. In fact, when I look at the three legacy systems out there, LRO definitely opted more to the complex side and then tried to make it as understandable as possible. At the other extreme, maximizer now called revenue IQ, they really took a-- we're going to have a simple model that's easy to explain quick for people to understand. And that's kind of the segment that they went after. You'll start somewhere in the middle. In fact, I don't know that most people know this, but you'll start started with a JPI student housing problem to solve, which is why that system is so dependent on the lead time curve or the booking curve. And it really-- it's price recommendations have everything to do with how is it going against that lead time curve. Explain that concept of a lead time curve a little bit more. Yeah, so lead time curve. I mean, it comes from in the hotel world, the booking curve, because people book a hotel before they arrive, you can establish this curve that come the end of a rival day, I've realized 100% of my demand. However big or small that is. So what percent have I realized one day in advance, two days in advance, all the way to 90, 180, et cetera? So booking curve, the term booking and reservation wasn't consistent with multifamily speak. So we invented the term lead time curve. It's synonymous with the booking curve in our world. Because we're similar that way also. I mean revenue management works in places where people reserve or buy the product in advance of using it, of consuming it, because then we can make adjustments on a day to day basis, right? If everybody goes in and buys a bunch of watermelons at King Super today, they can go order more. They can decide to raise the price tomorrow. But there's no chain. I mean, people grabbing a bunch of watermelons today doesn't really tell you much about what's going to happen a week or two weeks or a month from now. In our world, if people rent a whole bunch, well, we have a good idea of how much more demands left to come. And we have a fixed supply, right? We can't just order more watermelons. We can't just order more apartments. So that whole concept of lead time curve really matters. Having worked in the hotel industry a lot, I completely get how that works. Because you're turning your supply over so often, like what occupancy am I going to have on this day? That lends itself to looking at how a booking is going to accumulate over whatever, six months prior to the check in date. I get it for student housing as well, because they sell all of their stuff at one time. It's a ship that leaves once a year. Right. Exactly. So in multi-family terms, I mean, how analogous is that? Given that you're constantly just leasing a few apartments at a time of your time. It's honestly, it's one of the reasons why Yieldster is so vulnerable and it's pracing. I mean, if you ask most users, they will tell you they recognize that price can shift a lot one day to the next, and even flop, flop a couple days in a row. And that's just not consistent with the business. I mean, things don't change in multi-family that much one day to the next to just a five, five, or six percent shift in price. But it's an artifact of the model, because it's clinging, cleaving to that lead time curve so much, those switches happen, those big changes happen. When you suddenly go from trying to stimulate demand because you're below the booking curve, to when you're trying to discourage the man because you're above the booking curve. So if you're just staying below or staying above, those are the days that you see relatively small changes in Yieldster. But when it flips, that's when you see a big change. And that's an artifact of the model. It's not descriptive of the business. I think one of the reasons it doesn't-- one of the reasons a booking curve alone doesn't work so well is, again, go back to an airliner hotel. The airline's got a booking curve to a single flight. The hotel has a lead time curve, a booking curve, to a couple of days or a week if it's a resort. You get a chance to reload the supply on an airplane every time it lands. On a business hotel every month. I have a dinner half, yeah. Right, day and a half on a resort every week. Well, in our world, we reload more on the order of a year. So that's why I think that while a lead time curves an important statistic, if you-- my favorite word, apparently, in this webcast is over index. If you over index on that, the problem is you're missing the fact that, well, I'm not just trying to get to a lead time, a set occupancy. Use their word, a sustainable occupancy. In a given week, it's that week and the next week and the next week and the next week. What I do in a hotel for this week has no bearing on next week. What I do for a multifamily this week has a lot to do with next week. And I think the lack of an understanding around that in the model is part of why it's as volatile as it is. All right, so let's end this segment with-- if you're talking to somebody who's not that familiar with the stuff that we've just described and they're asking you the question, knowing all that we've learned over the last 20 years of doing revenue management, describe to me in plain English how a good, multi-family revenue management algorithm should work. I'll go back to something I've said to a lot of people. When we were going to build a contemporary new system, after watching the legacy systems stay essentially the same for more than 20 years, right? So this is back in like 2022. I said to myself, what does 2022 Donald know that he wishes 1999 Donald knew? The 1999 Donald didn't know back then. And I even asked other revenue managers, right? Brian Pierce at GID, Heather Cooper at Equity to name a couple people. Hey, Heather, what is 2022? Heather want to know or wish that whatever year when you started knew. And at least for Donald, I'll tell you what the answer is. Back in the day, especially coming from the travel hospital in the world, We thought it was 80%. math 20% process. And I know now in multi-family it's 50/50. It might even be 60/40 process, right? For a couple of reasons. First is multi-family housing is micro economically a much more inefficient business than hotels, airlines, etc. There's not enough transactions. Frankly, there's many more different business strategies out there. So the notion of a precise optimal rent is a bit of a full-sale. The other thing is those industries either have very sophisticated general managers of hotels or people who don't care about the pricing, the pilot of the airplane doesn't care. Right? In our world, we live with our residents, right? Our community managers, our leasing managers, deeply and emotionally care about their residents and therefore the pricing, etc. So it really is about winning arts and minds, it's about process, about psychology. And so a good system today, it still has good math, right? It doesn't mean 50/50 doesn't mean you can have crappy math, but it means you need to pay a lot more attention to winning the hearts and minds, how easy is it to explain for people to believe in it, right? If a leasing associate doesn't believe in the price, it won't sell. And so bringing in the process and psychology is really important. And also because multi-family has relatively small ranges in which it can be successful on pricing, paying attention to the cost of ownership is also really important. And the legacy systems have never looked at the total cost of ownership. And so I think a good revenue management system today has math that is as good as any system out there, but really pays attention to how easy it is for people to understand, you know, how you win the hearts and minds, what the psychology of the buyer is, and how that plays into the pricing, aka, concessions, etc. And then lastly, pays attention to the whole cost of ownership makes it easy for users to use so that a revenue manager can handle many more communities and the cost per community to run the system goes down. So the cultural points are really interesting. And I wasn't going to talk about this, but I do find this interesting. I wrote an article last year in MFE about the way that some multi-family talks about pricing increases. So one of the things, I mean, a multi-family for me was, I think it was the sixth industry that I've done revenue management in. And the thing that's really different between multi-family and the others is people sort of fetishize rent increase here. I sort of guess it, it's like we're more of a deals industry than any of the other industries are and rent increases have this high salience. But there are probably too many people in multi-family who think the point of revenue management is rent increases. That's the problem. And there's a lot of that is about, you know, it's probably a lot of vendors telling buyers what they want to hear about this. But you know, for example, if you would walk into a hotel revenue management meeting with the general manager of a hotel, not a revenue management specialist, but the person who runs the whole thing, and you would talk about, well, look at the rate increases we got. You would immediately put everybody on edge in the room because everybody in the hotel in the industry knows you can't take rates to the bank, right? That's this mantra that they tell you. Everyone gets the rent. Right, everybody knows that these are two ends of the same rope. And I think to your point about hearts and minds and getting people making it easy for people to understand stuff, I think a combination of this has probably been an effective way of selling software and the algorithms and workflows have not really been designed to make it so that people can easily understand what's going on. It leaves you in a scenario where there are just too many people that think about this as, as a rent increase machine. Yeah, I mean, it's boomerang down the industry. I mean, you look at the DOJ complaint, you look at the civil plaintiffs, all those complaints are written as if the only purpose of revenue management is to raise rents and quote-unquote beat the market. I was beat the market. I mean, you know, 70% of people think they're above average. Right. I push that closer to 100. I still haven't met these people. But you know, only 50% of the companies can beat the market. And 50% are going to lag the market. Like, what is the market? I mean, what we did with LRO in the early days was we proved that using the system, and we did really be testing using the system, beat not using the system. Right. And that's the measure. But interestingly, I mean, some of the biggest successes of LRO were in the 2002-2003 recession and the '89 great recession. And those were not times where we were pushing rents. What we did find, for example, there was a classic case in the 2002-2003 recession and Atlanta property where we were lowering one bedroom prices while we were keeping two bedroom the same. Because two things were happening. One is, in retrospect, we should have more two bedrooms. But we had the property we had. Also, as most people in this industry know, the inter-recession, one bedroom suffered more because people tend to double up as roommates or go back and live with mom and dad. So mom and dad, they're out of the system, double up in roommates. The two bedrooms actually don't do as poorly as the one bedrooms. Well, in the old manual world, 95% was the mantra, right. You would lower prices across the board. You'd peanut butter spread all your rent decreases. Well, LRO led us lower the ones more than we might have to short up. Meanwhile, we didn't lower the twos. It's not like we were raising rents on the twos. But just by not lowering the twos while we were lowering the ones, the average performance beat the cops by 400-500 bips. So our biggest success dealt with how and when we lowered, not with how and when we raised. And again, you're right, the industry does fetishize rent increases. We should celebrate targeted rent decreases because that's what shores up and avoids the vacancy loss. Yeah, and I can't remember where I said this, but the way I think about this revenue revenue management is about rent increases in the same way that salad is about not having a heart attack. Gentlemen, you can definitely see the logic that would make you enthusiastic about salad for that particular reason, but there's a lot more to salad than that. It's funny you say that because I remember the old sizzlers, this back in the 80s, and I remember I was in the Air Force, I was in Southern California and I would go to sizzlers for lunch. And you would see, and I don't mean this stereotype, but you would see this stereotypical, right, female, young Southern California, put whatever filter you want to put on that. And they would go to the salad bar, and there would be their lettuce, and there would be their bean sprouts, and there would be their alphausal sprouts, and then they would reach over and lock three ladles of ranch dressing up. And my cheeseburger with bacon had less fat than calories than their salad had because they lay on the. Yeah, well, so just to sort of finish this segment, because I think the thing that really, really gets lost is that you get to increase rent by doing revenue management when the market permits you to do so. But in order to capitalize on those opportunities, you have to do a lot of stuff well that isn't increasing rent. Just what I mean. It's more about laying groundswork. So that on those occasions, when you have pricing power, you get to capitalize on that. Absolutely. And to sort of the theme of this session around where do continents apply and where do they not. If you pay too much attention to comprehensive, that's an appraisal model. And the key to finding the opportunities to raise rents is to lead. We used to have the mantra at Archtone from Scott Sellers was, "We want to lead markets up and follow them down." We will follow them down when we have to. Again, contrary to some of the propaganda from the government and others who have their own agenda, software doesn't make the market. If the software price thinks too high, people would go rent single-family homes, they'd stay with mom and dad. They'd find alternatives. So software doesn't make the market. Good software can help you understand the market better. And that's exactly the point I think you were making is when our demand forecast is strong, we need to ignore the comps. If you over index on the comps, you don't ignore the comps when you should. And that's a fundamental point of good revenue management. Okay, we're going to go to a break and when we come back, we will talk about in detail about the comp data. Pricing is one important piece of the multi-family revenue picture, but we mustn't forget that revenue depends on having great and happy residents in your community. In today's multi-family industry, that means being better at understanding income stability and keeping out the bad actors. That's where snapped comes in. The industry leader in fraud detection now offers income verification and ID verification as part of its all-in-one applicant fraud detection suite. To learn more, go to snapped.com.snapp.com or click the link in the show notes. Alright, so Donald Twosy and the last segment you talked about, you characterized comp data in a sort of appraisal scenario. We have a lot of different kinds of comp data out there. Let's talk about what different resources there are that you've seen that people use. Yeah, I mean what I've seen obviously is the old school call-around. Just ask what somebody's charging and share what you're charging. I've seen the secret shop. The secret shop's really important in times of stress where the call-around or even a web-based price may not be the price really being offered. So you're trying to find out what's the actual price as opposed to the headline price that obviously gives you the third example of You know, back in the 90s, it didn't exist. On the best, you could like look at the print guides or things like that, but those were pretty out of date. So, you know, within internet world, it's pretty easy to see what people are charging, whether it's a human looking at it, or a bot doing some kind of screen scrape. You know, I'll mention it. Obviously in my career, I stayed away from it, like a third rail, and I'm grateful for it now. If you have access to actual leases, you can see the actual lease rent, as opposed to the asking rent, but that's only available through systems, because there's no manual or call-around way of getting that reliably. It's an area I've always been advised to stay away from the best the optics are weak. - I mean, what are some of the good use cases for using a comp data? Because there's lots of it out there. There are lots of these resources. You know, you've got supplies that lean more heavily on call-around. There are some, as you say, that's, I mean, there's some increasingly sophisticated tools out there that are, I think, guessing better and better at characterizing the gap between asking rents and factoring in concessions, and so on. What are some of the good use cases? - Because of the opportunity for outsiders to misunderstand what is happening in a call-around, I would tend today just for optics reasons, avoid call-arounds. And the fact is, they were the legacy because they were what was done before we had the internet, before we had the ability to get data in a more automated way. You know, screen scraping is the primary way. I would grab data today, but I would put a big asterisk on it because I know it may not be exactly what's really happening. There's all sorts of different ways that that data could be off a little bit. And particularly, any place that has highly itmentatized units, right? You don't know what the amenities are. And so you could see the starting at price for one bedroom or two bedroom changed dramatically one day to the next, but it might be an availability mix. It's not necessarily representative of an actual price change. So gather the data that way, but be very careful about how you use it. The useful uses of that kind of data, I think you and I've talked about this before, it's really more about a guardrail than a guidance, right? So I don't wanna cause the market to go down unnecessarily. So I might use that data as a guardrail to be careful that I only go below certain numbers if a human's involved. I might wanna use that data as some knowledge, you know, in the first segment you talked about that hotel example where they were only sold out one night out of 11. If I've got a property routinely at 88, 89% occupancy, maybe I'd sell it, maybe I'll rehab it, but if I've got to operate it, I'm gonna pay a lot more attention to the competitors. But if I'm running that 94, 95, 96% sort of typical occupancy, at least of the kinds of properties I've been involved in my life, it's rare that the content is gonna matter. And the few times that it does, it's usually been like when I think in my career, when did I care a lot about the comps? Almost always comes back to a recession, like the great recession, or even with the COVID days, the early days of COVID when we all thought it was gonna be falling knife. We didn't realize that demand was actually gonna grow during COVID. So in those situations, those are the ones that stress the publicly available asking rents the most. That's where the secret job comes in. So I'm very intrigued. As I say, I see solutions out there. I'll bring some really smart AI-driven technology to capturing the screen script data. Again, going way back when in the hotel industry, I was part of a startup that was scraping and processing hotel data. It had some very smart capabilities, like normalizing for hotel amenities, triangulating across lots of different websites. The problem that we always found though was that if you do the most perfect job of characterizing every bit of information that's out there on the web, the thing that you're still reporting is actually how good the property's marketing is. Because it's not a true representation of what my supplier, my demand is. It's sort of almost is. But you can't get around the fact that if some competitors in the market are less good than others at representing their stuff online, you have a systematic inaccuracy in the data. I think that's a good point. The attracting, attempting to use that kind of data to set the price is basically tying yourself to the quality good, bad, or otherwise of your competitors marketing and sales on average. If you're better than them, you're missing that opportunity to reap the benefit of being better. If you're not as good, you're actually overpricing and you might end up with vacancy. And while nobody wants to ever believe they're not as good, again, 50% of people are below average. So I do think it's not that that data's not a useful data point. It's not the driver of a good revenue management system as we talked about earlier is the demand and supply forecast and how I auto react with my price to that. Not the thing itself. The one point I'm going to make real quickly is I would also say the use of an appraisal model is highly relevant to buy and sell decisions. Because buying and sell decisions, you want to know sort of on average what's likely to happen over time. And so a sense of what the market is doing pricing-wise, appraisal-wise, is very, very useful when you're making a 5, 10, 15-year development or buy decision or sell decision. What people have to remember is when you're pricing, it's a very, very micro decision. What do I want to price this bedroom for this move in week, at this property for these kinds of people? That micro decision has a lot more nuance to it than just a simple assessment of what's going on in the market. Yeah. So it's worth pulling a few strands of this conversation together. So I think thinking historically about the multi-family market, just from my own background, I've always characterized-- LRO is the only orthodox revenue management system in the market. The others do revenue management, but they just do it in a way that's not consistent with how most revenue management industries do it. And I think the use of competitive data is very consistent with that vision, because if you're trying to set prices-- and that's the kind of example that I talked about earlier with the hotels that never sell out-- as you just said, you don't want to be looking at the external market to guide your pricing. In a multi-family property, I've got 300 units. The pricing of the other $299 is more relevant to my consideration of pricing than what's going on in the external market. The time when the external market, other than if you don't have enough demand to fill your property, the time that competitive data is useful is to evaluate revenue management decisions to identify if you're systematically out of whack with the market, which is a bit of an edge case. That's not normally going to be the way that it works. Two areas where this bias probably leads us down the garden path a little bit. One is that people tend to think that the way to price stuff is just look at the external market when it just isn't. The way to price stuff is to understand those people standing in line behind your prospects. Exactly. Yeah, I mean, again, look, a methodical use of appraisal data to have a disciplined approach is going to outperform people just taking fingers in their air and making human intuition decisions all the time. But that's not what the comparison should be, right? It's how do we be better than just that? And the way we're better than that is to implement algorithms that look at the demand and supply cues of your own situation, right? Again, if you want to talk about Bysel develop, you want to look at the demand and supply cues of the entire market or some market. And in that case, a lot of comp data can help understand that. But if I'm trying to set my one-bedroom price today, my studio price today, et cetera, I want to look at the cues of how has my property, my asset, my people been performing? And I'm going to see all that data in the leasing, in the lead gen, all those statistics. And I want to take that first party data to do a detailed forecast. And from that forecast of demand and supply, I'm going to make decisions about what is happening right this moment. And that to your point, it's not about the comps. It's about me. And again, I go back to, I've had a couple of people ask me, well, how can you possibly know your prices any good if you aren't explicitly including the comps? And I go, well, let's examine the scenario, right? Let's say for a moment that I price too high. What happens if I price much higher? Way too higher than my comps. Well, my leasing velocity goes to zero pretty quickly. My availability starts to grow. Wait a moment. I can see that immediately in my own data. Similarly, if I'm pricing too low, my leasing velocity shoots up, my availability shoots down. People at Wall, is there a lag? OK, maybe there's a little bit of lag, but by the way, it's two variables, right? So I get to see the availability drop and the leasing velocity grow. So a properly designed model is actually very sensitive to that. And in fact, in a system I've been involved in building more recently, we've taken a real close look at the lead at how the lead time curve works. So the minute like a marketing program kicks off and we get more leads, the legacy systems really have to wait until the leases happen before it realizes something's changed. Well, we can design a system so that it sees the lead count jump right away and now starts to realize, oh, wait a moment. The fact that I got more leads yesterday means I'm going to get more leases tomorrow, I can adjust in A for that. I struggle sometimes to wonder if I'm being clear, but to me, the difference between-- between relying too much on comps and using comps appropriately is back to that constipity of guardrail, not the centerpiece of how pricing decisions are made. And people have to get used to this concept of, it's a deal driven industry. So many of the leaders are very deal oriented, hence appraisal sounds good 'cause they're used to buy sell. But we're making a very, a very specific micro market decision today that depending on how that works out, we can change again tomorrow. And again, they have to that, et cetera. And that's just fundamentally different than what appraisal does. - It's a really good way to describe it. I mean, it's like you want, competitors are pricing to be the salt and pepper, not the meat and tasas. In fact, the problem with some of the other algorithms that we've talked about during this discussion is that for whatever reason, comps are various types, became part of the main dish. And that's ultimately not a good way to design a revenue management algorithm. The meat potatoes is leases and leads. - Yes. - Right? That's the meat potatoes. And if I knew nothing but the leases and leads, I could design a better revenue management model than if I knew everything perfectly about the comps. Right? Now, if I sprinkle a little bit, guardrail here, opportunity there, to your metaphor, it's the salt and pepper. But the meat potatoes is leases. - Okay, I think we're gonna need to stop there 'cause now I'm hungry. - Okay. - There we go. Donald, as always, it's been a pleasure. Let people know where they can find you before we sign off. - Certainly, LinkedIn, you can find me, just search for Donald Davidov. There aren't too many of us out there. It's the one that's at Reba. Obviously you can go to our Reba website, www.getreba.com, G-E-T, r-E-B-A.com. Somebody named Reba has the Reba.com address. And or email me directly first and initial last name. So D Davidov, at getreba.com. - And I assume you and some of your closest friends will be at Optekina. - Oh, we'll be at Optek. - We'll be at Optekin Spades. In fact, we're two years ago, it was a startup booth. Last year was 10 by 10 and this year it's 10 by 20. So. - Excellent. All right, well, look forward to seeing you there. And thanks very much. This was a great conversation. - Thanks Tom. Appreciate it, as always. (upbeat music) - Thank you for tuning in to TechTalk with 2020. We hope today's conversation was insightful and useful to you. Please make sure to subscribe to the podcast if you haven't done so already. So that you never miss an episode. If you enjoyed today's show, please share it with your network and consider leaving us a five-star review wherever you get your podcasts. If you didn't enjoy it, well, just forget I asked. We also publish a lot of thought leadership content and white papers, which you can get by subscribing to our blog at 20420.com. You can connect with us through the site or with me, Dom Beverage on LinkedIn. We'd love to hear your thoughts and your feedback. Until the next time. (upbeat music)