This week on Masters of Real Estate, Rich Hughes, head of data science at RealPage, joins us to share his knowledge of data and how it can provide incredible insights into the needs and wants of renters. Hughes shares his thoughts on the COVID crisis and how it's accelerating the use of technology in RE. We also chat about the types of granular data that exist in portfolios, how to dive deep into that data to find rent roll growth opportunities, and real life applications to optimize your portfolio.

In this episode, we discuss:

  • The emergence of data science and how it applies to the real estate sector. (3:30)
  • The importance of collecting and holding historical data for future analysis. (7:35)
  • How data provides insights into customers needs and desires to create a more personalized buying experience. (11:25, 27:30)
  • Using data science to assess your portfolio to improve the rent roll. (13:30)
  • The acceleration of automated and digital RE features led by COVID. (24:05)
  • Using "hedonic construction" to determine rent prices. (37:25)
  • How unit rental speed can be indicative of mis-pricing. (39:05)
  • The multitude of factors, other than price, that influence a unit's rentability. (39:50)
  • Occupancy is bound, but the rent roll provides more growth opportunity. (43:06)

Tips for Current or Aspiring Landlords:

  • Your portfolio's historical data is a good starting point to forecast for the future. (33:42)
  • Look at the seasonality of your portfolio to determine changes in specific attributes, such as price, renewals, and more. (34:00)
  • Get granular when viewing the data on your portfolio. (34:31)
  • Find the point of resistance on prices to maximize value creation. (43:25)
  • Optimizing rent prices increases value significantly during the sale of the asset. (44:50)

Episode Transcription:

(2:02) Justin Alanís: Rich Hughes, welcome to the podcast. Thanks for coming on.

(2:04) Rich Hughes: Hi, Justin. It's great to hear you again. How you doing?

(2:07) Justin Alanís: I'm doing great. Thanks. Yeah, really excited to have you on today. How are things in Chicago right now?

(2:12) Rich Hughes: Oh, you know. It's turbulent times on the streets. We've had some protests and counter protests and vibrant democracy-in-action outside my window. So it's been a show. Put it that way.

(2:24) Justin Alanis: Yeah. A lot of us in San Francisco are going through the same thing. So, obviously wish you luck there. But wanted to start high level with today's conversation. You've been head data scientist now at RealPage for about the last 9 years. We hear a lot about data science out there. But for those who are perhaps less familiar with it, would you mind breaking it down for us real quick and giving us an idea of what you actually do on a day-to-day basis and maybe why it's important?

(2:49) Rich Hughes: Sure. Few questions there. What I do and what data scientists and the two may or may not be related. But we'll talk about that. Data science became fashionable over phrase did a few years ago. I think Forbes magazine made it its, number one exciting job to have for a couple years in a row. So all of a sudden, everybody wants to be a data scientist. And there are lots of people in adjacent fields in technology are trying to branch into it. So it's a very, very hot sort of area. But its always been revered to some degree, but it was accelerated by a few phenomenons that came of age. The first was, is we just live our life more digitally now and record things digitally. So more facts are available on the internet. Sort of data stores. And that allows us to do the sorts of the mining and discovery aspect. And the mathematics behind it were always there. But the sort of the advent of more powerful computers and the low cost of compute, as well as some actual sorts of innovation in methods and algorithms have allowed us to analyze data at scale that really, really wasn't available before. And then I think the the sort of intersection of business with science or empirical methods. Data driven pursuits has also been a large accelerator to the data science field. So, you know, really the mandate for us is to look at the data reserves that we have at RealPage and work out how we can exploit them to create value for our customers. So to make better decisions, or to forecast, or to segment a market into different types that behave differently. And, you know, we're always looking to create insights that are actionable so people can go and take a business activity and do something to make money. If your insights don't create corresponding business activities, they're not really valuable because they're not actionable. Now, with regard to what I do, I collect really, really smart people. I give them tools to be successful, unlimited interference, and then I stay out of the way. That's pretty much what I do on a day-to-day basis.

(4:54) Justin Alanís: Yeah, I can attest to the fact that it's not exactly true. You do a lot behind the scenes in terms of actually digesting and crunching data, although you do hire smart people. And now you talk about this evolution of data science. Now you're the cool kid on the block, right? So, you went from being kind of more fringe science early days at Archstone and AMLI with revenue management. And it's in vogue. You're now the cool kid on the block. How does that feel?

(5:22) Rich Hughes: Well, you know, being cool in a sentence there are always gonna be strange bedfellows. So yeah, thank you. It's kind words and from you. You've certainly made your mark on the industry and done a lot in the analytics space as well. So I'm not sure who should be interviewing who at this point. But I do thank you for the kind words. It's great to be recognized for what we can do. Actually to give Steve Winn, the company's leader, credit by moving to a SaaS platform early and being able to centralize, memorialize all this information, rather than have them live on site in databases, really gave us a foundation to work against. That's spectacular. In data science, the data that you learn from, the facts that you read in order to work out how things work, that's called the corpus. And our digital corpus of things like least transaction history, and so on and so forth is unparalleled. So we can look and see how rent rolls perform through recessions and their boom times and all of those things. So there's probably very few events that we haven't seen in our historical back catalogue. And obviously, that becomes incredibly powerful; the understanding and modeling of things. But yeah, more information.

(6:36) Justin Alanís: RealPage is a 25 year old company at this point. Something like that. And it's almost, for its history, has lived in the cloud. I'm sure there was a transition at some point a dozen or so years ago truly to the cloud. But that corpus of data has lived through now, the Great Financial Recession, and now obviously, you're collecting data through COVID. But also the corpus has grown over time with the client base as well. And also with probably the different products that RealPage is using to collect that proprietary data for its customers.

(7:05) Rich Hughes: No, that's a very astute observation. And it's funny. Our strategy around data has probably evolved a little bit. It used to be that storage was expensive. So keeping all this data around was at some point you purge. And sometimes there's regulatory or legal reasons why you have to purge data, but we would say, if it's more than a year or two years old, throw it away. And actually now, I think the the mindset is, let's keep everything because the cost of keeping it is so, so low. And even if we can't think of how to exploit it now, it doesn't mean in the future, we'll see how it becomes predictive or indicative of something. Or we can stitch it with some other data to discover a new frontier. So to anybody out there who's on this sort of data discovery journey, information and money are pretty much the same thing at this point, so please don't throw any away. Worst case, send it to me and I'll keep it for you and then we can share it. That would be okay.

(7:58) Justin Alanis: Yeah, exactly. You derive insights for it and it helps their business. A lot of different directions that we can go here but I'd like to dive really quickly into the different verticals of data science. So data science is kind of a broad term. And then artificial intelligence is thought of as machines starting to learn behavior on their own. And there's different subcategories of artificial intelligence, like machine learning and deep learning. Where are you guys right now at RealPage and how do you think about the evolution of the data science field related to AI?

(8:27) Rich Hughes: Alright, so some great phrases in there. And, of course, like any great phrases, they'll be grabbed by executives, misused and become largely meaningless in the sense of things. I'm teasing, but you remember the cloud and big data? That we're all taught that they're very exciting, but no one's really wrestling it now. First of all, AI is not as scary as it needs to be. I think the common sense definition of AI is a machine doing something that's so difficult that normally it takes a person to do. So whether that's translating a language or identifying different types of pitches, or, being able to spot a good credit applicant from a bad credit applicant. These are all pretty involved things. And actually, now I wouldn't say we don't mature in the AI space, but we can do a lot of things now digitally that people could never do. And I look at Elon Musk is certainly very much in that space. He's got rockets that now land themselves backwards onto landing pads. People can't fly a rocket that well. That's just impossible for us. So, in certain very narrow domains, AI becomes really, really powerful. Certainly much better than people now. I think a lot of people conflate a sort of very narrow, task specific AI, which tends to do very well if it's trained properly with the more general intelligence that people have. AGI, artificial general intelligence, is in that field. Where you know, can I continue to know what it is? Or can you learn a random assortment of unrelated tasks quickly? Probably not. That's a very difficult human level task. But when we apply it to very narrow domains, very specific fields, and teach it with a big history, most AI outperforms people if deployed properly.

(10:12) Justin Alanís: Walk us through maybe a good example of how that can be deployed within RealPage and within real estate investing and/or asset management and management.

(10:21) Rich Hughes: Sure. What I like about real estate is it exists in many, many levels. That's probably the most minute, is someone on the web doing something and you're trying to interact with them in a way that creates the most value and entices them to your offering, all the way to the lease, which is a financial transaction that's actually very meaningful. For most people, it's about the same as buying a car. So it's not a popping down to the shop for a Snickers type deal. It's something that you would throw a lot of time and care, love, and analytics to get that right. To the unit types in the units themselves, to the communities, or maybe what markets you want to be in your, data tells stories and give insights at all of these levels. And so there's not one right place to do data science. And they all feed into each other, right? Better interactions gets better residents which allows you to fill up nicer units. Better communities making more money and maybe being in different markets. So there's not one entry point or exit point for data science. We try and work in all of them. Because of RealPage's history, our original bread and butter was in the sort of leasing and rent space, The who is in what unit and when do they pay rent, when do they move in and move out. And from that we've got into revenue management, and then we've sort of grown out that way. And now, we're in all different areas. Some of our competitors, say for example, started more in the financial space, and some started maybe in the customer journey in terms of attracting residents. So there's been lots of entry points into the business but they're all amenable to data science.

(11:50) Justin Alanís: And it can be manifested in a real world application. Something like a recommendation engine. So you talk about revenue management and how RealPage got its start in the leasing and rent sector. So you had all of this tenant data, what they were paying, what the renewal patterns were. And ultimately started to understand and look at market characteristics and more aggregated information to be able to draw inferences and recommendations for resident managers and managers to be able to say, this is what you should rent your unit for. Is that how it works and that's how a real application of data science?

(12:24) Rich Hughes: Yeah, it is. You could argue maybe credit screening and revenue management were the first true data science applications in real estate and I wouldn't really argue with that. They're certainly mathematically complex tasks that a person with Excel struggled to do at the level of what a machine can do right now. That's an interesting example because right now, and I don't know how much gratuitous product plug I can do on this wonderful Switzerland of a webcast you've got here. Alright. Well, we're always looking to fine tune our algorithms and the problem you sight is someone going to renew. We're deploying massive amounts of AI against that right now. There are lots of different signals that impact whether someone is going to sign a renewal lease. Obviously, the price you offer, the price they pay, fair market price nearby, and how many months they've stayed, the type and size of the unit itself, the month they moved in, climate. There's a load of different things that are going to impact that. If you can use this information to forecast ahead and say, okay, in this batch of renewals, it turns out there was a lot of short term and corporate leases, so I know that I'm gonna go down. For various reasons the model says that, and I don't think I can maybe get the 50% renewal rate that I was going to get because of the composition. I'm only going to get 30%. Then obviously, that impacts a lot of things operationally. You might not actually discount to try and get them because they're unsavable. Quote unquote unsavable. So you might keep your pricing recommendation for that batch of renewals higher. At the same time you may go, right, I need to spin up some marketing and demand generation because there's going to be a shortfall in my rent roll. A big hole where these people were but aren't anymore. And I'm going to need new people to fill those. Or you might stagger out some lease expirations, offer some people some early renewals in a different batch. So it all plays together. And it can work at many levels. You can solve a discrete problem with data science, but you also have to appreciate its part of a broader ecosystem of problems that really sort of manifests as a rent roll.

(14:28) Justin Alanis: Yeah. And then you can this data that's sits at the root level. It really feels to me like having the root level data, so the unit information, the tenant information, what they pay in rent, the unit square footage, and then you roll that up into a zip code level or maybe a zip plus four level and a sub market level and a market level, and then even a state level. And so you can then look at different patterns that are happening across those different levels as you indicated earlier. So we talked a little bit about how revenue management delivers insights at a unit specific level, but then how do you then aggregate that data up and start looking at it on a broader basis? And what kind of insights can you derive from that kind of information?

(15:14) Rich Hughes: So yeah, great leading. I will struggle to give an answer as good as the question. One thing about information is, there's probably more depth to it than you might initially think. And that sounds cryptic so let me explain. A unit in a community, it has attributes. So it has how many people are in it, square footage, and maybe when it was built or renovated, what its current rent is. And then the community itself has attributes. Its location, when it was built. Maybe what we would say its quality as class A, B, or C, which is something that we we derive. Maybe the community type is a high rise garden, mid rise, the vintage, what decade was it built. And maybe even the operating model. Mixed use. Is it a residential? Is it affordable? Is it military? So all of these become dimensions or attributes about things and you can slice across these dimensions. So if you get a big enough pool of information, you can start slicing. It's like it's a magic cake that you can keep slicing in different ways. But you get to keep the cake, which is of course the best cake. The cake that you keep. So you can say tell me about you know, Class A in Atlanta. Class A in Atlanta built since 2000 that's high rise. As you get into these sorts of little subcategories or the dimensionality increases, you start to see little pockets of overachievement and underachievement, over subscription and under subscription. I mean, I think one thing that we would all acknowledge is in the what was the golden era that we certainly just lived through, there was a lot of class A real estate built  and probably more high rise product than ever before. And you could argue maybe too much, but it looks like it was driven by, after the financial meltdown, real estate being viewed as a very good risk adjusted return across all capital investment classes. And a lot of foreign money coming into the markets. And of course, they all want something that's a bit of a marquee product to put on front of their annual report. Here on state Maine is our wonderful grass and steel high rise apartment community, which is spectacular, but I think the world only needs so many of those, and we probably got enough. So now we're seeing the more modest offerings. The more suburban less new stuff is actually doing really, really well. And when we look at renting a data science problem is to look at, there's obviously other ways to live. You can buy a home or you can rent or you can get a roommate and co-occupy. So we have to be aware of the things that drive those decisions. Obviously, the affordability of money, savings, but also things like the population pyramid and the demographic breakdown of America. There's an old trope about as the population changes, schools become hospitals and hospitals become schools, as you get those sorts of ripples of large and small groups of demographics. I think that the Boomers are the ones that are cited as being the very populous one. Interestingly enough now, they're quite old. They were very wealthy, more wealthy than that cohorts ever been, although I'm not sure that's still true after the recent financial events. But that being said, a lot of operators started looking at multifamily and said I'm doing it to get into extended living or a sort of product that's geared more towards that specific age group because there are lots of them. At the same time, the number of renters, or the people that historically would enter the workforce and would rent probably before they buy, that's not a particularly big cohort for us right now. And it isn't going to be big for the next few years. So these decisions drive data science and probably operational decisions. Now, one thing I will say is that I've made very broad brush national level statements there. I am sure there are pockets of America where the complete opposite is true. And more people moving to here than ever before, or migration is fantastic from here to there. So even in a down market, there are stocks that do very, very well and I think the same is true for real estate. You have to A, be able to find them, or even better, B, predict them. Because once it's happened, everyone knows it's happened and you pay the premium. If you can guess what's going to happen or, guess is probably a bad word, predict what's going to happen, then you can get there early, buy the thing and have the appreciation. I think, and you deal with real estate people a lot more than me, I get the feeling that holding real estate is this unfortunate thing they have to do between buying and selling it. Everyone likes the deals, right? Everyone wants to buy something, make it better, flip it, change it, renovate it, and then sell it at a nice premium. And I think that's sort of really the fuel of real estate.

(19:47) Justin Alanís: Yeah, it is. And we've seen that both at Rentlytics and this new business at Awning. We're seeing that. We're now elevating our discussion to market level information. So this points to how people want to invest and how they deploy capital. And when you're looking at a market level, we're talking fundamentally about the give take of supply demand, and then the extra element of risk. And so this is reflected in things like cap rates across different markets and cap rates are meant to be reflective of that supply demand and risk that you're taking associated with the asset. But ultimately, the information that starts to dictate that is obviously what's happening within real estate and operations is a huge factor there. But you started to talk about these other data elements. Things like immigration patterns, jobs and the major employers that are there. Certainly during COVID things like unemployment rates and things like defaults and tenant evictions, and whether there's distress in a particular market. So you're now starting to fuse data together with now the main corpus of RealPage, which is the data that you have at kind of the the lowest level, the unit and tenant information. Now you're able to then bring that up to a more market level and start to drive investment decisions, which is what you say the buying and selling of the real estate becomes so important.

(21:12) Rich Hughes: Yeah, I think so. And as you bring up COVID, I can just speak to that for a minute. So, modeling in general works when tomorrow looks a lot like today, which was a lot like yesterday. So, I think it's called the Copernican principle, but in the fact that things tend to be pretty similar across time is good for us. Because that means that I can create rules on yesterday, and they'll apply today and they'll apply tomorrow. Now, the crazy thing from a data science standpoint is we've been through the mortgage crisis, then the best time for real estate ever, and now we've got a global pandemic. So it's hard to argue these are normal times, although maybe there are no such thing as normal times anymore. But when you train your model, the corpus that you give it, and say, okay, learn about real estate from this history, you have to be careful that you give it a history that's going to be applicable across a lot of different time possible futures. Rather than, if you have to train a model just on the best ever time, you'd have the most aggressive rent and valuation models ever because all they've ever seen is incredible growth. So when we train models, we do try and get some of the good times and the bad times in there. And now that becomes a question of, would we even allow this sort of COVID time into our training corpus because it's certainly not indicative of normal societal operations.

(22:37) Justin Alanís: Yeah. That's really interesting that you might not allow that data set in. But you would probably use something like the Great Financial Crisis data in order to start to predict out what may happen in COVID. Though COVID represents a uniquely different situation than even the Great Financial Crisis because this recession is going to hit mainstream a lot more. It's gonna hit working class citizens and so maybe have more pressure on C class assets. Although I hear through a lot of what RealPage is saying and other industry professionals that actually the A class is getting hurt right now. So how do you use that historical data from things like the Great Financial Recession and apply it to a situation like we have with COVID?

(23:19) Rich Hughes: So I think what we're always trying to do is to take an outcome or an empirical event and try and look at all of the factors that may have contributed to it and work out how much they all contributed to that event happening or not happening. And some may accelerate the likelihood of the event and some may decelerate or inhibit the event. And what you have to understand is that, although we've seen lots of things historically, we've probably never seen a combination have events that are exactly the same as now. So although we've been through economic cycles that are good and bad, the population was certainly different. I don't think we had the rapid digitization of society that we're going through now. I mean, one thing COVID is, I think someone beat to the phrase, but they're talking about the distance economy or the touchless economy. And it's electing a lot of people, or rather forcing a lot of people to have a digital experience where they would have had a physical experience previously. And I think what we'll find is A, people that are good at digital will do very well in the short term, but B, we're training cohorts, maybe the older people who aren't so tech savvy, to use technology. And if it seems seamless and easy they'll like it because it's convenient, and they won't go back to physical experience. So what does that mean for the tools and the on-boarding of prospects? I'm not sure, but I do know digital is now more prominent and I don't really see a revert to a more physical experience to the extent we had it before.

(24:52) Justin Alanís: Yeah. I'm a big believer that 90 plus percent of owning and managing real estate can be automated. Whether it's receiving packages. Whether it's door codes and automated entry points. Whether it's the leasing experience in a virtual leasing experience. Whether it's the maintenance experience and having a digitized experience where tenants are able to interact directly with their maintenance people and allowing them into their home through some sort of smart home technology. Being able to track the residents behavior within that smart home technology and understanding it. And so it comes down to, and this is where a lot of what you're doing can help companies like RealPage and tenants and landlords, is that it doesn't just live in the data where you're helping landlords make better decisions and charge more rent, but you're also in a lot of ways using artificial intelligence and data science to produce automation at the site level and create that more seamless real estate ownership experience from end to end.

(25:51) Rich Hughes: No, that's very well put. I think historically we've been focused on the sorts of slower data. By that, I mean things that if you record it overnight, it's fine, rent roll, who's in what apartment, those kinds of facts. But the frontier for us now is clickstream. So when somebody comes in and clicks on a floor plan how can I create digital intimacy with them to entice them to leasing with me. I guess if someone's just shopping for an apartment now and they're quote unquote, top of funnel for me, they're top of funnel for a lot of other people as well who are also fighting for their business. So for leads that maybe the I can A, assess and think that they are higher quality than average or there's something about them that I can see or induce that makes me think they're a really good prospect. And then if I can find facts about them, and engage them in languaging terms and concepts that they're looking for, I can steer them towards the purchase event with me much quicker. And therefore, financially do better than the people that miss out on that great renter. So that real time frontier for now is where a lot of emphasis and now you can't wait to run an algorithm overnight and see what you should do. They click on them in a millisecond you have to say, have I seen this entity before and you may not know their name or their phone number, but you may be able to see this device I've seen before or this IP address or from this point of origin, I've seen traffic before. And last time they were looking for two bedrooms and they had a dog or I think they had a dog. I inferred it from what they were doing, or maybe their cookies told me they have a dog. There are lots of ways you could get the information. So all of a sudden, they get a website that's tailored to them. That's highlighting the dog park and the wonderful two bedroom floor plans. So a big trend in digital has always been customization. It used to be, we'll create cohort groups of five types of renter and we'll do five types of experience. Now, I think when someone comes through the door, you should be creating their website that's unique to them, highlighting all of the things that you can do that resonates with them. And by the way, if they're an absolutely terrible fit for you, knowing that early is great as well. Because while they may rent, the probability is very low and you're role resource constrained and time constrained. So you're probably going to throw cycles at other people, or at least they'll be a little further down the list because the fits just not right for them with what you're offering.

(28:24) Justin Alanís: Yeah. I think that's a really big deal. And we see that a lot in the technology space with software as a service companies and marketing automation, where they're starting to learn people's behaviors. Both through their online presence and activity, but also through selection and choices that they make as they move through the customer experience. And you can then tailor through technology and through automation in a lot more efficient and better way than you could as a solo landlord or a property manager out there and trying to distill and distinguish between residents on your own. And so it's a very similar way, and probably actually more applicable in the marketing space, because you have so much more volume of traffic in terms of helping your staff and your team understand, well, is this a resident or a tenant that I should potentially pay attention to? What does this mean from a credit perspective and their ability to pay, all the way through their entire leasing history is it shows that you can start to optimize every single experience. And it's not just about making more money for the landlord, obviously that's part of it through optimization, but actually you're delivering a better experience for the tenant themselves because they get an optimized experience of how they search and curate properties and space and establish preferences. And then also when they move in, if you can become a better landlord by understanding their preferences and what they want from an amenities perspective, and maybe if this particular property is not working out and you have another property that you can recommend to them, it's more about servicing the customer in a very real way and creating intimacy and customization around that resident experience.

(29:57) Rich Hughes: That's really well said and I think a frontier that you're alluding to there is when we talk about marketing, most people actually think of sales. So they think of how can I put my current product in the best possible light that people coming through the door will rent it? And that's well and good and laudable and we all think that's a good use of time. But you speak to marketing as really how people and products interact, which is a better definition. And as well as finding residents who are a great fit for your product, you could create product that is a great fit for residents. And I think that's a frontier that really hasn't been mined to the fullest yet. Operators in the future who get good at this will have a scary advantage over people building vanilla stuff. In some ways, standardization is good, from a cost perspective, certainly. Now we're seeing with prefabricated inventory and standardized sizes, things can be made cheaper, more standard, replacing them becomes easier, you hold less inventory. All of that is good, but we have to remember that home, which is really the business we're in, is an emotional product. It says to the world things about you, and you want your personality to be part of the home. So I think letting prospects have designed decisions about their home is great, because what's the worst is going to happen? They love it and want to stay there for a long time? That would be okay as a landlord. So I think letting customers co-produce digitally the experience they're going to have, and the size and the shape might be fixed, but colors, or the packages that come with it or how they live their life or how technology's part of it, or their broader living experience. These are all frontiers, I think that that can be exploited by data savvy operators, or even what sort of apartment should I build in the future. Not only how big and rent levels but really who's it speaking to? Who would want to live here? And it's not just one person. It's probably groups of similar people. And not everybody wants to live with everybody else in the same type of apartments. So I think by trying to appeal to everybody, you might appeal to nobody. So find the sweet spot and get really good at it.

(32:15) Justin Alanís: Yeah, absolutely. And so I want to talk really quickly. So RealPage does, obviously so much with their data, but RealPage really sits in the enterprise space for institutional landlords now. RealPage recently bought a company called Buildium, which operates a bit more in the, what we call small and independent landlords segment of the market. And I'm sure that you guys are going to bring some innovation around data and data science and artificial intelligence to Buildium. But while you guys are ingesting that acquisition, I'd love to talk through how you think, I mean, you've been at Archstone, you've been at AMLI. You've built systems before without a huge corpus of data. Obviously, AMLI and Archstone have bigger corpus of data than a small and independent landlord. But what are some practical tips that DIY landlords can use where they don't have a ton of data, but maybe they still want to take a more mathematical approach or more systematic approach to things like marketing and leasing and renewal processes? What are some tips that they could use and some really quick calculations or formulas that they can maybe apply to their own business, whether it be a mental model, or whether it be an actual Excel document that maybe they put together that they can use in a similar way? Obviously, not as good, but at least taking a similar approach to what you might do on a much larger scale?

(33:32) Rich Hughes: Okay, so good question. The first thing is if you try to work out if an event is going to happen, look at all the historical instances of it might have happening and work out what percentage did happen. So, the historical take rate for an event is probably a good naive starting point to forecast into the future. The next thing I look at is periodicity or to be property management specific seasonality. Do some events happen more at certain times of the year? So do I find at least more apartments in the summer than the winter? Is there more demand in the summer than the winter, can I get more of a price increase in the summer than the winter, and this will be different for Florida than it will be for Minneapolis. So do the math yourself and work it out. And then the final construct is granularity. I have a community and I've got units and they're one, two and three bedrooms. They're not all behavioral the same. People rent them for different reasons. They have different needs. So, because of that, their rent rolls will be different and how their rent rolls evolve will be different. So I need to treat them as separate entities. When I talk about portfolios, when people talk about portfolios, generally they talk about your groups of assets. And no one just looks at their portfolio as one big asset and runs it and says, okay, rent's going to be 1,000 dollars for everybody and mandate that out. That would be silliness. So at the same time, why not go more granular and be smart and say, okay, my one, two's, and three's are my portfolio? How can I manage those? How are they different? Now, I understand when you get down to small units, and maybe for a smaller operator, it becomes difficult because you don't get the safety of large numbers to make decisions. If you've got one vacant after four units, I understand that's 25 percent of your rent roll. So your risk tolerance is probably less than a bigger operator, but I think you still don't assume the worst. Go and look historically and see, okay, what rate of these things happened, and see if that can give you comfort to make better decisions. And if you don't have that data, maybe there's a benchmark you can get somewhere that would help you. I mean, the most basic thing we do with apartments is we go and look and see what other people are charging for similar inventory. And we use that to guide our own pricing process oftentimes. Things like that are great, and I think there's a lot more information available about not only sticker price rents, but also effective rents, renewal rents, average lease terms, percentage of rent roll that's month to month. It's all out there, so don't see what other people are doing and presuppose that they're rational. You too can partake in however well they're doing, or if you're smarter, maybe even a bit better.

(36:20) Justin Alanís: Yeah, and I think a good way to do some of that, you talked a little bit about it, is comp research. So if you look at what other people are charging from a rent perspective, the best way to do that is first to establish a comp set and look at buildings that are most similar to yours based on location. But also In terms of profiled asset, year it was built, renovated, and size of units. And then to go out and look at what other folks out there are charging on those units and then figure out whether you should charge less based on your sensitivity to vacancy or charge more based on maybe a profile of your unit that might be a little bit better. And so is there some sort of profile or some sort of basic algorithm that people can do basically by saying, alright, I've got 6 comps, this one has a pool, so I'm going to deduct 25 dollars for that, and this one is closer to Whole Foods, so I'm going to add 25 dollars to that. Is that a rubric that you think is applicable in certain instances?

(37:19) Rich Hughes: Yeah, it is and the methodology you've talked about there's real name is called hedonic construction. So we think of price like Lego bricks. So the base price of this unit is 800 dollars and then they've got a balcony and that's 25 dollars. And maybe it's a nice unit so that's 50 dollars. And we just stack all of these values together, and it comes up to the final price. If you've ever tried to build a PC online or went shopping for a computer you start off at 1,000 dollars. And by the time you've tweaked it all you've got a 3,000 dollars PC in your shopping cart, because they're very good at sort of upgrading and allowing you to do that. The same is true for apartments, except we should never talk in those terms to the customer. Only as an internal construct to help understand value. If you tell people, this is your apartments price, and here's why, you're inviting a multitude of arguments as to why they disagree with the value of a fireplace, balcony or a better view than the next unit along. So, I think what you have to do is do the best you can and there are certainly some statistical tools that you can use to help with that. One thing we do and actually, I think the first person to do this was Donald Davidoff, was to look at the days on market of individual attributes of units. And if the construct says if something's great, it flies off the shelf. And if something's horrible, it languishes on the shelf. And in a well run community, if you're pricing all parts of your value proposition correctly, units should more or less move at the same speed. So if you find you've got some attribute, where its days on market is very low or very high, clearly, that's a price preference being exhibited by some people. They think it's great value or terrible value. And in that speed difference is an opportunity to maybe right price the units, which will give you a multitude of positive outcomes. That unit will move at an appropriate pace, which means that the community itself will actually run a bit better because you don't create gluts and surpluses of inventory because you're just mispricing one tiny part of the value proposition.

(39:23) Justin Alanís: Yeah. And it could also lead an investor or an owner to make a decision around upgrading units or changing their marketing and looking for where they are struggling in that leasing process. Is it not getting enough views? Is it not getting the conversion? Are people touring, and then they're not leasing? And so you can start to pick apart the funnel and start to understand where your bottlenecks are. Because Rich, it is true, isn't it, that price isn't always the way to move a unit? It could be some other leading indicator that landlords should look at to say, am I doing this whole thing correctly?

(40:00) Rich Hughes: Yeah, so pricing's my background. And I think I see what I would call a lot of apologetic pricing. Items solve any problem with price, but I probably shouldn't. Say it's just a bad floor plan or it wasn't turned properly or there's something about it, I can discount until it moves. Is that the right thing to do? No, oftentimes, there's a better solution that isn't price related. And by using that, I can keep price the same and obviously, then make the money that I'd like to make. If I haven't marketed well enough, then I just simply don't get enough eyeballs on my price. So it doesn't really matter what the price is. It might be great value. Because of the limited eyeballs I've got, everybody loves it once they see it, and they lease there, but maybe if I spent twice the money on my marketing, although I would discriminate. Only half the people would look at the price and think, oh, that's a bit expensive for me. I've got so many eyeballs on it, that there's a big enough population that will take that higher price. So I think price is where we cash out a lot of decisions. Where we realize the value for the analytics we've done and the decisions we've made, but it it should be a reward for making good decisions. It shouldn't be a mechanism to account for bad decisions.

(41:14) Justin Alanís: Yeah. Completely agree with that. And I think it's really important to first look at how you're doing things and look at the people you have, the promotion you have, the actual physical product, and whether it shows well. And so you have to go through maybe a series of checklists and have a mental model around how you think that the best way to go through that entire process before you look at price as kind of the final one. Rich, before we go, I want to ask you, what is one element that you've seen in your data at RealPage that you guys hammer on with your customers over and over again? Like, if you don't do this right, then not much else is going to go right within your portfolio? Is there one metric that you say to landlords, you should be looking at this? Is there an equivalent of like an NPS in technology? Where if your customers are happy, and you know that they're going to give word of mouth, and there's going to be a referral process, and they're going to stay in the units, they're going to continue to use your software and pay higher prices as they renew. Is there one number like that in real estate, especially apartment ownership, that you can think of that can be a signal for guaranteed results?

(42:25) Rich Hughes: We look at obviously, the rent roll is really where the empirical test of if you did a good or a bad job, and there are lots of ways to get there. I guess from a revenue management standpoint, we're always being asked rent or occupancy? Rent or occupancy? That's the eternal question. And a lot of people like to run incredibly full, thinking that the occupancy is clearly, it's much more powerful than rent. Because, if I go from an empty unit to a full unit, that's another 2,000 bucks or whatever, 1,000 bucks, whatever the number is where you are. And that's far more powerful than getting, you know, 10 dollars, 20 dollars incremental rent increase, and that's true but occupancy is bound. You can never be more than 100 percent full. And for you to grow your enterprise in a meaningful way to the future, the only mechanism that can do that for you is the rent roll. So you should always be testing price and trying to advance price in the understanding that there are really two ways, at least two ways. You can do new rents or maybe in a down market when your rents are a better mechanism for you. And also offering some optionality with regard to the lease terms and month-to-months may let you get more demand than you've historically been able to get that. So, with regard to that the rent and occupancy conundrum, sure they're both super important but I actually lean more towards rent. And the reason for that is in a similar industry, airlines, when the plane takes off, if they've got one empty, they think they're pretty much there. Now, that doesn't happen anymore because of overbooking. But the idea was, we got to a point where it was expensive enough that we were almost full, but that one last person couldn't quite pay that amount. And that might be at 400 dollars or 500 dollars a seat. You could be full at 300 dollars a seat, you could be full of 200 dollars a seat, and you could be full 100 dollars a seat. And you wouldn't know which of those was the right price. So finding that point of resistance, and a little push back on price is actually good, because it tells you you're operating in the right domain in terms of value creation.

(44:26) Justin Alanís: I think that's really relevant to our audience as well. Because these small and independent landlords, when they have 10, 20 units and they have one unit that goes vacant, it hurts their bottom line to a higher degree than an institutional landlord who can always float at 95 percent and optimize a bit more fluidly. But I think one of the things that is really important to remember is that when you're pushing rent, that you can then capitalize on that at a 10 x rate upon a sale. And so if you continue to push rents up, then you're actually going to be able to sell that asset for more in the future. And when people underwrite these assets, they typically underwrite it towards around 95 percent and have an understanding that there's going to be some amount of days vacant that the unit sits. And so as a landlord, if you can take the near term hit of cash flow in order to continue to accelerate rents in a bit more aggressive way, or at least get them up to market more regularly and not carry a huge loss to lease the difference between market rent and your in place rents on your books, then you'll be able to capitalize that on the backside. And actually create way more value than you would just by creating this occupancy increase that accelerates cash flow while you own it.

(45:39) Rich Hughes: Oh, absolutely. And we see even when an asset is going for sale, people do things to the rent roll. They know they're not going to get credit above 95 percent occupancy so they're not going to be more than 95 percent occupied. The oldest trick in the book is, you run heavy, so you get up to say 98 percent, and then you put the rents up really high. You get a couple of extra leases at very, very high even though your occupancy goes down, but when someone looks at the rental, they say, okay, I'm at 95 percent, and my last 5 leases were super premium leases, which is great, and they draw this line through it, that really isn't true. So the operators have been playing games around that for a while, but even at the more basic level. if you've got a waitlist, you're underpriced. There's just some signals that show you that you could be driving price more. I think Malcolm Gladwell did an article around realtors saying that, when they say I can sell your house in 2 or 3 days, what they're really saying is, I will underprice your house by such an incredible amount of money, I'll sell it in 2 or 3 days. So I'm sure that there's a too fast in property operations as well. So just be aware of that.

(46:46) Justin Alanís: Yeah. Always good to find that equilibrium. The right place between how long your unit sits vacant and optimizing for price. And I think that's a really good lesson for both institutional landlords as well as small and independent landlords. And all of this has been really a fascinating discussion, Rich. Your experience and the amount of data that RealPage has obviously has been really valuable to you guys at RealPage and your customers, but I appreciate you sharing that with us and our listeners. I think that there's a tremendous amount to take away from this that's real and applicable that can benefit their business, so really appreciate you taking the time.

(47:23) Rich Hughes: Thanks very much. It's been a pleasure and of course, we're all expecting to see more wonderful, great things from you as well, Justin. So great to chat with you again. Take care. Bye

(47:30) Justin Alanís: Same with you, Rich. Thanks so much. Bye.

Thanks for listening! Make sure to: