Bruce Orr, Founder and Chief Data Scientist of ProNovos, shares his experience creating value with construction data, joined by ProNovos client and partner Wick Zimmerman of Outside the Lines, a Design-Build contractor.
Bruce_Clipped
Hugh Seaton: [00:00:00] Welcome to Construted Futures. I'm Hugh Seaton. I'm here with Bruce Orr the founder and chief data scientist of Pronovos construction analytics and Wick Zimmerman CEO of Outside the Lines. Gentlemen, welcome to the podcast.
Bruce Orr: [00:00:14] Thank you Hugh
Wick Zimmerman: [00:00:16] thanks Hugh glad to be here.
Hugh Seaton: [00:00:17] So I want to start by letting everybody know what you guys do. So starting with Bruce, let's talk a little bit about what you, what you're up to, and then maybe we'll tell us, and we can dive in deeper.
Bruce Orr: [00:00:27] Yeah. Yeah, sure. So what we do, we're a construction data platform company basically. We just help contractors utilize data to protect their margins. We also focus on improving efficiency for us it's about measuring performance and improving output.
So we're a tech startup and at this phase of our company, we're really focused on helping subcontractors manage labor. So that's pretty much what we do at, Pronovos.
Cool.
Hugh Seaton: [00:00:59] And Wick talk to me about Outside the Lines.
Wick Zimmerman: [00:01:02] Sure sure. Outside of the lines where Design Build specialty contractor and what we've referred to as the Themed Construction industry.
So we do a number of water, features everything from complex show fountains, choreographed and music, and light with fire and fog and, and so forth to natural streams and lakes that you'd see in golf courses and parks to life support systems, artificial rock work, artificial, coral, and so forth, and zoos and aquariums and things of that sort.
Hugh Seaton: [00:01:34] You must have an amazing backyard.
Wick Zimmerman: [00:01:37] Well, you know what they say? The cobbler has no shoes.
Hugh Seaton: [00:01:40] That's right. That's right. It's all crabgrass.
So let's, let's get into kind of the reason we're here. Right? So let's talk a little bit about what you guys do. So we've just talked about what the companies do, but talk about let's, everybody's interested in data and everybody's interested in, you know, getting better at it.
I think that's, you guys are hitting this at the right time in the construction industry where it's just everybody's trying to figure it out. Let's talk a little bit about what you guys do concretely.
Bruce Orr: [00:02:08] Yeah. You know, so I'd like to start this offbecause I'm so fortunate to be in this space at this time.
Construction from a tech perspective has been behind the times, but we've been collecting a lot of data and this is what I do. And I love what I do, it's fun, it's rewarding. And honestly, I've been in this game a long time. So when we started off with an analytic solution for construction, it was based on things that we did for other industries.
And when I discovered how underutilized data is in construction, I knew this is exactly where I wanted to be and working with Wick. He was an early adopter of our platform and understanding how he was running his jobs and his company, you know, I saw an opportunity and so Wick, I, I feel like you don't regret the day that we met, but Iyou know, helped you and your company sort of make some, some good decisions using data because that's why we, why are we here? So I'm curious Wick and Hugh , not to step in your shoes, but I I've been wanting to ask Wick this question a long time about you know how has this impacted his company , and just the jobs that he runs.
Wick Zimmerman: [00:03:25] You know, it's interesting. Bruce, when we first met, I think I told you the story that I knew we had all this data and we'd collected all of this information overall, all history and time, but I had absolutely no idea what to do with it and how to get it. And, and you know, when we would delve into some of these, these ideas, we'd spend hours and hours.
Calling through data, building spreadsheets and Microsoft queries and all that stuff. And then by the time we got done with all of that, we were out of time and out of energy. And so you never really got to the important part, which was focusing on what to do with that information.
And I'll never forget. It was one of the first meetings that we had. You asked me what seemed like a simple question, but wasn't simple at all to me. And it was, you know, what information, do you need to run your business more efficiently? And I sat back and I thought, I don't know. I know I need something, but I don't know what it is.
So it's been a learning experience for me. And as you know, as we've gotten into this you know, your eyes open to, "Oh, well we could do this or we could, we could determine that. Or what about this? And what about that?" And so it's, it's been very enlightening for me and it's certainly been beneficial and.
You know, I feel like we've really just hit the tip of the iceberg really with what's possible out there.
Bruce Orr: [00:04:49] Yeah, yeah . You know, so I, gosh I'm so excited to be having this conversation conversation with you. I am a fan. And I was thinking about all of the great things that one can do with data.
And before this call Wick and I, we were just kind of talking about some cool things that that we're doing and, and Wick, one thing that comes to mind when we first started to even go down this path was the whole the data discovery and data visualization piece that is getting hotter and hotter now.
And so what as, as a CEO, what are you, what are you doing with, with, you know, your dashboards and data discovery and visualization?
Hugh Seaton: [00:05:36] If I can throw in, what do we mean by data discovery? Dashboard I think everybody gets, but when you say that, I think, let's talk about what we mean by that. And then, you know, Wick, if you've got some examples of what you've done.
Wick Zimmerman: [00:05:48] Sure.
Bruce Orr: [00:05:49] It's so, so w what I mean when I say data discovery I, I mean, it's, it's looking at , so data discovery is, is really about taking columns and rows of data. Like for instance, think of the, think of the like, like, like a job cost report. I mean, you look at some of these jobs that are huge and it will have thousands, I'll just say hundreds of rows of numbers, your estimated cost you know, your actual cost and just a lot of information. And oftentimes it's hard to look at that data and understand exactly where the problems are. And the trends that you could uncover if you discovered with a visualization tool and data discovery, again, is all about uncovering, hidden trends that you typically can't see with just, you know, looking at columns and rows. And I don't know Wick, if you have another way of articulating that at all or not.
Wick Zimmerman: [00:06:55] I think that's a great description, Bruce. I mean you know, in the previous days we would look carefully at the WIP report, for instance, for job health conditions and you know, focus on as a start on jobs that were way underbilled or overbilled.
You could look at gross profit and then you'd go to your AR report and, you know, look at where you were on receivables and AP and then kind of mentally have to figure out what cash flow looked like. And the great thing about dashboards for me, I'm dyslexic. So numbers are not so bad, but reading is is more difficult for me.
So visual, anything visual is, is very beneficial and you know, these dashboards give you and what's great is it can be customized to what some particular position likes, for instance, for me, I like to locate cash as part of my dashboard, you know, where are we? Where's our cash position today. I like to look at where we are profitability overall, you know, company-wide and and some key factors like that, that then I can drill into if If I see something that's out of whack, or I see that there's a trend of projects that are projecting to trend over budget. You know, what's going on there and drill down into that, but it gives you a great visual for that. And then, you know, that. Operations guy can get a look at you know, a different set of information. Admin accounting can look at a different dashboard so everybody can have kind of their own dashboard based on what they find most important to their to their job.
Hugh Seaton: [00:08:25] So if I, if I get really quick, if I can ask them just to clarify, when you're using the word data discovery, it isn't really unearthing new data. It's really going in and finding out what the data you have is saying. Is that right?
Bruce Orr: [00:08:40] Absolutely. Absolutely.
Hugh Seaton: [00:08:42] That's great. So, so if you think about it, the first thing you're doing or not first, but an earlier phase , of a process for this is to kind of just look at what you've got and start to understand what's there. And I think that's where the visual really helps, right. Is you see patterns visually that are hard to detect when it's just a bunch of numbers.
Bruce Orr: [00:09:02] You know, it's interesting that you say that because visualization and data discovery can mean a lot of things to a lot of people, I will speak about how I've seen estimators use our application and estimators, are ones that they, they just spend a lot of time with data. And, and so looking at data from different places. I mean, you know, you got your accounting system, you got your maybe RS means or wherever they get their you know, the standard costs from, and then if they have labor, just other places where they get data and, and so you load all of that, that data into Pronovos and let it just work for you.
And it can do things like gosh, like, like for unstructured data, it can go in and identify like connections across multiple sources to tell you, you know, if you haven't thought about looking at seasonal type of inputs, when you're doing estimating, you know, you should consider that. ,
I hate talking about this like, it's this black box that is magic, but you know, there's a lot of thought that goes into building this based on construction best practices and it spits out what it thinks, you know, the interesting and useful information could be.
Hugh Seaton: [00:10:31] Let me ask about that a little bit. So you're saying that the system is recommending, you should try this, you should try to, you know, cross tab with that, or is it kind of doing some automated or semi-automated analysis and saying here's what we found or some combination of the two.
Bruce Orr: [00:10:47] Yeah. So it's a combination of what you're referring to is augmented analysis. Wick and I were talking about a new feature that we are working on augmented analysis, it is basically helping Contractors of all positions to prescribe, if you will you know, a particular decision.
So we're, we're, we're doing, we're doing a lot more of that, you know, again, going back to the estimating piece, if you're a public contractor, and you have all this data to bid on oftentimes you know, who is going to bid and you, could later find out what the winning bid is, but what we're trying to do is get ahead of the curve.
What if you are you know, the winning bidder because you know what the bidding bid amount is going to be and don't leave money on the table, you know? So we're working on things like that as well, which is really exciting in this is just some of the examples that that are yet to come in, in construction.
Hugh Seaton: [00:11:43] So you're starting to get predictive, which is the kind of, I don't want to say Holy grail, but it's certainly the step that everybody's talking about getting to is, is not just describing what happened in the past, but starting to give some recommendations about the future.
Bruce Orr: [00:11:57] Exactly. And I will say this, Hugh I'm I'm this is what I do. I run a tech business. I often love hearing how our customers are using a platform. I could keep talking, but I would love Wick I mean he's running a phenomenal business. And so Wick, you know, Hey , chime in. I want to hear all about how you use ProNovos and you know...
Wick Zimmerman: [00:12:24] Absolutely. Well, you know, one of the things certainly predictive analytics is, is of great interest to me because I see a lot of opportunity there, but one thing I want to touch on that you mentioned Bruce, that that is a huge benefit to us is we have all these different software packages and they're all SQL server databases, but they're all independent. So we've got our accounting software, we've got our estimating software, we've got our CRM software, I've got a project management software and everybody's got all these different things. And one of the great things about ProNovos is you can take all of that data and put it in one warehouse. So you have really a single interface to access all of your information. And that's a huge, huge thing for us on the operational side of, of this process. So, you know , having this information available, you know, we utilize it really throughout our entire business and, you know, thinking about how the business flows from business development, into estimating to operations and so forth
Great place to start is, is business development, and you know what we've often tried to understand in order to predict where we're going to be revenue wise what we're going to need in the way of, of resources to manage that revenue and so forth are, you know, the way we look at our funnel is we've got touches, which is contact with some kind of a potential source or lead; an actual lead where a project is identified; and then an opportunity where we're really going to actually bid or put together a pricing or budget on that; and then jobs that are, are actually contracted.
So for each part of that funnel, we want to look at. How many touches does it take to get a lead? And how many leads does it take to get an opportunity? And how many opportunities does it take to get a job? And all of us contractors have always looked at hit rate, you know, how many opportunities to, to get a job.
And that's been near and dear to everyone's heart, but it's really nice to be able to, to look deeper into that, that funnel, to anticipate what that looks like. And, and then. Add the third dimension of time to it. And all of a sudden, now you have a an ability to look down and say, okay, if I want to do X dollars of revenue, I need to bid this many jobs, which means I need this many leads and I need this many touches.
And so I need so many business developers. And then in the estimating arena. You know, how many estimates or how many dollars can an estimator generate in, in a certain period of time. So how many estimators do I need to support that, and project management and so on and so on. So you know, that planning and, and using that information is especially beneficial.
Hugh Seaton: [00:15:22] And, you know, one of the things that we talk about in the industry is is skilled labor shortage. And I mean, what I'm hearing you talk about is you're using among other things, using data to more effectively plan for how many people you're going to need. So you're less likely to run into a shortage because, Oh my gosh, we forgot we need more people.
Wick Zimmerman: [00:15:39] Absolutely. Absolutely. And you know, you, you take that information from operations. And one of the things Bruce did for me early on that I thought was absolutely amazing was I was convinced, although I had no idea how to do it, but I was convinced that, for each different type of job that we do there's some kind of a curve of how you spend those dollars and everything in construction, accounting is done on a cost of completion. So as you spend money that percentage spent divided by the estimated total cost is percent complete. So everything's kind of driven off of costs. And you know, when we would do our projections. It was really a complete wag carried out to two, two or three decimal places. But it kept dawning on me that there's gotta be a better way to do it.
Well, sure enough, there is. And, and Bruce said, look, let me do this. I'm going to take a job that you're just starting. I'm going to look at some past data, and then I'm going to tell you, I'm going to give you a projection of how I think that you're going to spend those costs over, you know, whatever the duration of the job is . Our, our typical job is six to six to seven months, probably a year plus jobs are pretty unusual, in our line of business anyways.
Bruce did that exercise and that his projection was within 5% of the actual job costs by period. It was absolutely amazing. And so it just goes to show that. You know, using this historical information and having the ability to compute that and project that forward is incredibly powerful.
Hugh Seaton: [00:17:22] Do you guys mind if I ask a little bit more about what you just said? Cause that's, that's great. And, and that, that example you can think of it, you know, not just in spending, but in, in, you know, a bunch of other data data points from safety to, you know, time spent and so on how many, how many jobs did you feed into the model to get to that level?
Bruce Orr: [00:17:42] So luckily most contractors like Wick, they have an insane amount of jobs that they've been running for a long time. And so what we would do, is look at the past five years the past five years have been, you know, the industry has been looking pretty good.
But when we look at that data, we know that every job is different, but there are many commonalities that you will find from job to job. And what we would do is look at those things that are in common labor, right? We all have, you know, labor, there's a cost associated with it.
There's a number of people, you have some that are higher paid. You have some that are, you know you know, maybe apprenticed, but we, we look at all of that information and the more data we have, the better we can predict. And, and it's really interesting. I love this aspect of like construction, right? It was hard for me to understand everything is about controlling your costs and protecting your margin.
And I didn't understand it as much until I started working with Wick. Nothing gets past him. And when I was able to show like, "Oh yeah, you know, this data is showing that. You know, based upon the trajectory of the job next month is going to end up at X". So when the project manager is providing his inputs to Wick, he has another piece of information that could help him guide his decision.
When he says, I think my cost at completion is going to be X.
Hugh Seaton: [00:19:26] So better context. I asked, I asked that the question about how much data it took, because one of the things that I know slows people down with any technology is feeling like that doesn't work for me because I don't, I'm not, I'm not whatever it is. And in this case, everyone is in a little... honestly, everyone across the economy is nervous about the state of their data. If you go and talk to a retailer, they're freaked out about their data, even though their data is usually pretty good. You know what I mean? Like everyone is nervous about it and understanding that, that, you know, most people, most contractors have data that is at least good enough to do what you just described is really important. How much did you have to do with, with the state that the data was in to feed it into your system? Did you have to scan a lot of things? Did you have to kind of rework stuff or was it, was it not so bad?
Bruce Orr: [00:20:16] You know, luckily I think going back to what I was saying earlier, I've been doing this a long time and technology has gotten so good that it didn't take long to build out our models, understand what our objectives were, understand what Wick's objectives were, and basically just, just pull the data in.
Now, one key thing that I'd like to say, because I don't want to have someone believe, Hey, you go grab a, a data scientist, build a model, and it's going to work. There is another problem that you have to solve and this goes back to a modern data warehouse. You have to take all of this data from different sources and you've got to clean it because data quality is, it can be an issue. So think about it. If you have bad data coming out and you want to predict something, you're going to predict something that might be wrong.
So what part of what we do is this concept called master data management, and it ensures that the data quality is good, so that when we predict we can get within that 5% that Wick talked about.
Hugh Seaton: [00:21:28] And how much work was that? I mean, that's, that's an, almost an unanswerable question, but let's just say a little bit or, it's manageable.
Bruce Orr: [00:21:38] You know, that we invested money in building the technology upfront. So when we began to engage with contractors like Wick, It wasn't a big, you know undertaking and the more we do it, the better we get. So I think it probably about 90 days, we were up and running with Wick. So you know, now we're, we're, we're getting, we're getting really good. So it's like half the time now.
Hugh Seaton: [00:22:03] That's really cool. I mean, I've heard some of this also is just how the data sits, but how it gets entered. But you've heard stories about, you know, trying to get project managers to enter in data the right way. Can take, you know, many months, but that's another story, Wick you were saying.
Wick Zimmerman: [00:22:21] Yeah. That's a, that's a whole different story. Yeah, exactly. You could, you could do a couple of podcasts on that. One of the great things about it though, is, you know, you have this upfront investment and, and, you know, it's been a few years since Bruce did that for us. And, but my recollection is it seemed like it happened really fast, but once that's done, you have all of this information at your fingertips. And so the, the it's limitless, what you can do. And really what was kind of fascinating to me was. Yeah, Bruce would come back and say, Hey, have you ever thought about looking at this or would this help you at all in, in your analysis? And, and in a lot of cases, it's stuff I had never even really thought about.
And I was like, well, yeah, that'd be pretty cool. Let's take a look at it. So once they have done all that upfront work You, you know, you're, you're pretty much unlocked to do whatever it is that that you want to do with it.
Bruce Orr: [00:23:17] And, and just, just to add to that Wick, it made me think of something I was reading earlier and it was talking about purpose built applications and I think purpose built applications are good.
For an example, maybe you have an application that does... tracks safety very well or time entry really well. These purpose-built applications tend to be better when the data is cleansed and it can see across multiple business units within the firm. And then add the analysis and insight that something like does.
An example, let's talk about safety: let's say you're tracking safety really well. You know all of the misses, near misses, and struck bys and all of that. Well, when that data is on the ProNovos platform, it can tell you the probability of, you know, the job's health regarding safety. So, you know, it's those purpose-built applications that helps drive the fuel to allow us to do what we do.
Wick Zimmerman: [00:24:27] And ProNovos provides that seamless integration of all those applications, where normally you'd be going to five different places to get that, that information. And just kind of going back to what Bruce was talking about on safety. There, that's a huge opportunity there. And even beyond all of the observation and the reports and things that we typically put into you know, safety or a project management software on a daily basis for the site inspections and so forth, you know, we're we're as a course of business, taking hundreds of photographs a day on projects and you know, the ability to be able to take those photographs somewhere down the road and, introduce artificial intelligence/ machine learning, to be able to look at those photos and say, "Oh, there's a safety issue" or, "Oh, there's a quality issue" or, "Oh, you know, there's two guys in that photo that are standing around there's a productivity issue" and being able to get all these additional data points from something we're already existing that we're already doing in an existing system.
You know, that's some of the really cool stuff that's, that's out there. Not, not in the too distant future that can really help with that because like anything else, the more data you have have regarding something that the more accurate your predictions are going to be, and the more you can use that to analyze specific situations or outcomes.
Hugh Seaton: [00:25:56] I love what you just, you said a couple of times, both of you, how you know, Pronovost is able to pull data from multiple systems. I mean, one of the things that is a perennial issue in, in honestly everywhere, construction, a lot of construction folks think that they're the only industry where different systems don't talk to each other, but that's honestly pretty common that's what IT departments are for, but nevertheless, it's an issue and we're all kind of learning and feeling, feeling our way across the river, as they say. With with regards to IT, the fact that right now you have a platform where some operations might not be connected, but the picture is so you don't have to wait for, you know, one system to talk to another, to be able to run analytics against it.
You're able to, to kind of, kind of almost be up a level and pull it all together. For one, I've heard someone use the expression, a single pane of glass. I love that one . How have you used it for for HR.
Wick Zimmerman: [00:26:52] You know, for HR, there's, there's obviously an incredible amount of information that you generate on a regular basis there through timecards, which we actually do timecards in the field on a iPhone or iPad. So we have the ability to get real-time job productivity and cost information in our system. We're not waiting for it to be entered by payroll week later or whatever, but you're also capturing vacation days, sick days in a various HR functions that you can then look to see, you know, is there a safety violation, you know, is there a particular position that's more prone to do that than another?
Or, you know, we've learned is that the chances of somebody having a near miss or an accident within the first 90 days is 80% higher than at any other time in their employment. So how do we focus on that? Well, it could be an orientation or onboarding or you know, additional training or whatever. So, you know, that's, that's one piece and it's kind of that's, that's that historical piece of information that You know, most people think of as what data is all about is it's collecting all this history and reporting on the history and that's important, we have to do that. And we can use it for scorecards, you know, the balance scorecards kind of a buzz term used, but you can collect, you can set up the metrics that you want and you can collect that information and have it report to you without having to go a different bunch places to look at it.
But one of the key things I like about it is, you know, something you touched on Hugh, which is it does allow us to predict our personnel and human resource needs down the road, whether it's based on job and, you know, our labor budgets, taking all of our labor budgets and the man hours that we have projected for a certain period of time, comparing it to the manpower we have available. You know, are we going to be in a pinch and you can go a level deeper because for us, our guys, like we have some guys that are good at plumbing. Some guys are good at waterproofing. You know, some guys are good at carving plaster and other guys are good at painting and, and some are good at many of those.
But the ability for us to have the right person doing the right task on the right job at the right place, because we work all over the world is, is very important. So it gives us the ability to look at people's skill sets and match those with that needed on a particular job at a particular time.
Hugh Seaton: [00:29:20] That's amazing, especially in, in sort of today's world.
Bruce Orr: [00:29:23] Hugh I was just going to add to that. It makes me think of two examples. I actually the correlation between pay you know, like, you know, your, your average hourly rate and the average hourly rate in the location where you're working. So It makes me think back to there was this project.
It was I think it was like Nashville music center and one of our customers there we're based in Atlanta, Georgia. So we have a customer that was based in, in Georgia and they won a big job there. I mean, again, this is a huge job, so they were basically employing people in that area, paying them a wage that was based here in Atlanta and this, this was a multi-year project.
Well, they were underpaying the people. And so we had at the time, now this is fully deployed, but at the time we were playing with this narrative analytics, that'll basically describe a situation to you. And the situation was this. You're paying your employee in Nashville $30 an hour. And the average hourly rate for that Trade is $35 an hour. You're losing people on this job because you're underpaying them. If you increased by X dollars, the net correlation Your, your attrition will essentially get better. So so, so it made me think of that. And this the same example is down in Florida, it was like the I-9 corridor is something multi-year project and we were just monitoring the attrition.
So we have to keep the people that are good on, you know, so yeah.
Hugh Seaton: [00:31:07] That's a really, I'd never thought of that because you, you often hear that it's that whether it's attrition or whether it's you know, various elements of performance, that it's always an issue. And just making sure that you're not underpaying people is a huge point.
Well, gentlemen, thank you so much for a really cool conversation. Where can people find you?
Bruce Orr: [00:31:27] You know, for us you can go to www.pronovos.com and always just search for ProNovos on, on Google or, and I'm Bruce or, and yeah, that was great Hugh. Thank you. Thank you. And Wick?.
Wick Zimmerman: [00:31:48] Yeah, and us, you can find us at www.otl-inc.com and check out our website and certainly contact us from there if if need be and Hugh, thanks for including me enjoyable. And it's always great Bruce, to to engage with you on this topic. It's fascinating.
Hugh Seaton: [00:32:06] Yes it is. And it's just the beginning of the story.