why your 2026 ai strategy is not going to work

Small Improvement Business Operations Podcast, Small Business Operations Podcast in the entire world. How are you today?

Good. I took a look at our numbers. I think we're up to 4 billion downloads last month. So it's always great to see those numbers go up into the right.

Yeah, that's right.

Yeah, that's really incredible. So, uh, other than your absolutely lights out, fantastic new Christmas sweater, thank you, Gemma, my wife, I could devote a whole episode to What are we talking about today?

We are talking about— well, we're continuing our discussion about the theory of constraints, and this is going to be a multi-part series that's going to extend into first quarter of 2025, 2026. But specifically today, we're discussing the 4 steps to rolling out new technology in a way that doesn't flop. And it's very applicable, uh, because a lot of Listeners and a lot of businesses right now are considering their 2026 AI planning and strategy, and there's a couple things you need to get right or it's really not going to work.

Before jumping into what is the main course for today, I want to kind of circle back slightly, which is you and I probably got 20 or 25 messages about our episode on constraints.

We did.

Mind-blowing, like, like, oh my God, this was so useful. I've already listened to it 3 times and all that stuff. And it's cool. First of all, it's cool to get that kind of feedback from people who I can, who I respect and admire, like not schmoes, right? Like smart people being like, I never heard it explained that way or whatever. And you actually said you had a good idea, which is putting it in the AI and, and, um, and, uh, asking why do people like this episode so much?

Yeah.

I think I think it really— we talked about this on our non-recorded podcast, which is just when you and I talk all the time yesterday, because we basically did a podcast, but it was just me and you talking and not recording— is I think there is absolutely a direction that we could take a majority of our content, which is I think what you and I are really good at, humbly, is the same thing that Socrates was good at, which is he was a popularizer. He was a popularizer. Right, which was he could take a dense text and say, here's what this really means to the masses.

Yeah.

And I think why people like that is because the goal— and Eli Goldratt's stuff, he's a physicist, he's a doctor in physics— it's dense. And if you're not smart, you might miss how profound it is. And that was just an example. And I think today is going to be another example where you and I could take a concept like a book or an article or something that's really dense, and you and I both are very good at saying, so what does this mean practically speaking?

Yeah, I mean, I feel like there's a long tradition of taking ancient wisdom and retranslating it, not literally translating to a new language, but reinterpreting and re-explaining it for today. And like every generation, what does What is theory of constraints for a TikToker? Exactly. Every generation has their own filter through which they view the world, their own language, their own culture to a certain extent, their own patterns of work and play and talk and writing. And so if you can take ancient wisdoms that have been useful across many generations that are timeless, you can explain them in a way that resonates with a new generation of people. I feel like that's a tried and true formula. That's not something we set out to do, but it's just sort of an accident that that Theory of Constraints episode— that is what comes up for me when I think about what did we do in that episode. We took an ancient text from the 1980s, timeless text, yeah, uh, you know, from the late 20th century, the late 1900s, way back in the late— we, uh, you know, reinterpreted it, you know, we got out our translation dictionaries And we went over with, you know, microfiche and, uh, you know, really, uh, translated it for the modern listener.

I love that. Yeah, I love that. And, you know, my— I would say, and I think you're the same way, which is one of the things that I like the most about this, of course, is when one teaches to learn. And so the reason why I, you know, I'll listen back and I'll go like, I'll learn. It's like, oh yeah, Peter did say that in a good way, or that is the right way to think about that, right? So for me, very selfishly, I also enjoy breaking apart, like, you know, I was mentioning, I want to just grab one of my timeless books from my bookshelf and say, Peter, you and I are going to talk about what we take away from E-Myth. And you and I could easily talk about that, unpack it, modern— guy talks about uniforms way too much. I still remember that like 30 years later. Like he talks about uniforms because he ran a fast food restaurant.

Like, And yeah, he also uses like a bakery as a, as like one of his main examples. No one's bakeries today. We need to do like, substitute that for the thing that everyone's starting today, like an agency or whatever it is. Yep.

Yep. So, so that, expect more of that because I'm, I'm, you know, I would do another one of those episodes this afternoon. Like that's how excited I am about that journey because I'm, I'm, I'm learning this kind of like alongside. So, okay. So let's jump into, um, Let's jump into the topic today. Now, I would say first and most importantly, we just need to, in two sentences, talk about what the theory of constraints is, just in case somebody didn't hear the last episode. And essentially what it says is you as a business owner are not trying to optimize every sub-function, every link in the chain in your business. You have one link that is the weakest, and there's only one because it is the weakest. And you need to direct your time, resources, focus, energy, team on strengthening that weakness to increase the throughput of the whole chain. Not saying I'm trying to get every individual link as strong as possible. So that is my short summary of theory of constraints. What do you want to add or clarify?

I mean, what a relief, right? In some ways to know that you don't have to work on everything all at once. And in fact, it's not optimal. Instead, you need to take some dedicated thinking time, CEO time, as John would call it, and really identify what is the constraint in your business? What is the bottleneck? What is the thing that if you fix throughput of sales, of production, of fulfillment, would drastically increase. So that's what we call the constraint. Everything else is irrelevant. Optimizing non-constrained resources is a complete and total waste of time. That is the main idea behind the theory of constraints. Now, Dr. Eli really spends a lot of time in his ancient text from 1984 discussing this very old concept called a factory. And the fact, you know, this is the whole thing is about a factory floor, basically. So it's, you know, not necessarily intuitive to translate what does that mean for me running a property management business. I'm not producing a product. So that's sort of what John and I've been wrestling with quite a bit and sets the stage for how we're thinking about the work.

Excellent. You and I have also walked down more of Dr. Eli Goldratt's writing, and I Again, what I find very great, what I really like about him is because he was a physicist, I find him to be very intellectually rigorous, which I enjoy. It's not reading some nonfiction book from some jabroni who I could just like walk holes through their arguments. You know, he's very disciplined in how he lays out arguments and say, well, if you believe this, then it follows like, and I can follow. He's got very methodical, clear thinking, which I kind of really admire. And so even, I would say, something that's even easier to implement than theory of constraints, even easier, which theory of constraints is a very simple idea, but in practicality, it takes some soul searching and problem definition. And really, you know, the problem is not just laying around, you have to go digging for it. He also talks about the The Four Stages of Technology Deployment. And he talks, you know, in chapters and for an hour in one of his books that you and I both listened to or read about why technology rollouts fail to increase productivity. And he talks about ERPs and he talks about, um, I think MRPs, which was some old school version with like punch cards and material resource planning is what that is.

That was all about like how much, uh, inventory, how much, uh, What raw material should you order for the upcoming month?

Yeah. And the, the, I think we have to start with kind of like what is the presenting symptom, which I think is ubiquitous in small business, which is you buy a new technology and it fails to increase throughput. Like that is the presenting symptom. So you bought Salesforce and you said, guys, this is the year we get serious about CRM and you equip your team with $100 per seat Salesforce Salesforce, and you notice you still need to increase your headcount linearly. They still can only process 10 X's per day, and you go, well, what the hell did I just invest all this money in technology for? They invest in LeadSimple, but there's no significant change within their business, which is a like ubiquitous phenomenon.

Yeah. So, so to really put a fine point on this and make it real for any property managers listening, like, how many times have you heard a property manager say, yeah, I switched from PropertyWare to AppFolio and it was a lot of work and it's a little bit better in some areas, but everything's kind of the same and it was really disruptive? Or, yeah, I switched from Rentvine to Buildium You know, it doesn't even matter the specifics of the software because it's the same results, which is a lot of work, a lot of money for basically net zero, sort of where you were before. Now, I'm not even— I'm not blaming the software providers. I think they're releasing tons of really interesting, time-saving, legitimate products and features and even some AI-powered stuff. Uh, but it's, it's the way we go about it, which is what we're talking about, these technology deployments, you know, whether it's a CRM, whether it's a new showing software, whether it's a new process and workflow tool, whether it's your accounting software, uh, you need to not enter into that lightly. And you've— anyone who's been listening to me for the last couple years has heard me say many times, because people always ask me to be blunt, why haven't you switched from Buildium to AppFolio? And I always say, yeah, AppFolio hasn't paid you enough.

Yeah.

I always say it's not the constraint. You've heard me talk about that long before you and I got rehooked on this book, by the way, John, it's not the bottleneck.

Yeah.

You know, it's not, building them is not what's keeping me from reaching 1,000 doors. It just isn't.

Yes.

So until it is the thing that's blocking unit growth, which is sort of, yes, that is, that is so profound, dude.

That is so good that you've, that means you've internalized the concepts, which is. Would it make us better? Sure. Is that the constraint at RL? No.

Yep.

Some fires you gotta let burn, right?

Yep.

It, it also, you know, as you're talking, I, I put this in my newsletter that I just wrote, um, uh, which was, it also provides a verbiage to me. Oh, I'll actually say two things. First, I think you can actually broaden the bit. Uh, I think you could actually broaden the definition of technology. I think you would actually probably put the word global talent in there.

Sure.

I think you could. I think any sort of innovation— why didn't this innovation work? Well, it wasn't against the hard part.

It wasn't against the hard part. Yeah. Global talent's kind of a cheat code though, because in some instances, like when an employee quits or gets fired and you replace them with global talent, you're basically just by default saving, you know, you're getting a third of the salary or whatever.

Yeah. Well, this is the second thing I'll say, and I put this in my newsletter. It also— have you read about any of these? I've seen some kind of like private stuff from private equity guys about this, but have you seen a lot of these series, these people talking about essentially venture-backed companies buying local service businesses, you trying to use AI and then trying to basically turn them into software companies?

Totally. That's been a huge theme in property management.

It's been talked— it's talked about incessantly, right? And there's venture-backed players trying to run it, which I think are the ones who are in the most, do the hardest. But then there's private equity and, okay, holding company guys, whatever.

Okay.

Basically, we're going to buy this company and use AI and okay. But I realized why I, in many, many instances, hate that strategy. And I'm going to pick on one in particular. There's some person out there who bought a pool cleaning business, like a couple, and they raised $20 million. And I saw the press release and they're like, we're technologists and we're going to come in. And I went, My brother and Jesus.

We've been laughing at these people in property management for a decade.

Until you can tell me why AI is going to help solve the $21 an hour person who's in a rich picky person's house, I am uninterested. Unless you— oh, they send text message follow-ups when the techs leave the jobs. Who cares? Well, it lets us close our— who That is not the hard part about running a pool cleaning company. It is not following up on leads that didn't close. So until you can solve that problem, you're going to run into the exact same structural limitations that everybody else does, except with way more debt and way more expectations. And so unless you can say, This AI helps me, we make them wear AR goggles, and we monitor their behaviors, and so we can fire them quicker if they clean the pool wrong. Okay. Franchising, by the way, I'm on a roll now, sorry. Franchising solves this. What's it do? It makes all the labor issues local, and it basically, it extricates some amount of revenue, and they solved the constraint because now who has to deal with the drunk pool cleaner? The Z, the franchisee. So that works because it's addressing the constraint, which is managing low-cost local labor. AI in pool cleaning?

Yeah. The fax machine ain't the constraint.

Correct. Correct. That is not why that business hasn't grown. It's because it's got terrible working capital dynamics and you can't find high-quality X in Y geography. Unless you can show how technology solves that, I just think it's a waste of time. So that's the other reason why I like what we're about to talk about, which is it also gives me a verbiage to know instinctually. I was like, I think that's a horrible idea, but I don't have a clean way to tell you why. And now I'm like, I know exactly why. Because step 1 in the 4 steps of technological deployment is actually, sorry, That's step 2. So let's go step 1. Step 1 is identify the power of the technology. This is the easiest step because the vendor tells you.

Yeah. And going back to my example of like, you know, why we haven't switched off Buildium, this, you know, the audiobook that you and I are listening to by Dr. Eli Goldratt goes on and on about ERP rollouts and, and property management software is essentially an ERP. It's the central software. It's, it's highly verticalized and tailored to our industry, but it's the central thing around which everything else is built and operates. And what I think is interesting is, you know, I just gave a bunch of reasons about why we're not switching, but if you and I walk this all the way through, at the end, what we should have is a framework for how to successfully— for when to and how to successfully make a large change like that in a way that is hugely meaningful for you and your company. And also, by the way, the vendor should love it too, because they're much less likely to churn a customer.

Yes.

Right. The last thing AppFolio wants is somebody to switch preemptively or prematurely in a poorly thought out way. And then that person churns off 9 days later and is now not only not a customer, but is actively, you know, badmouthing AppFolio or Buildium or whatever, Rentvine, you know, whatever software they churned off of. So I don't want people to think we're saying you should never switch softwares or these big companies are a ripoff. Far from it. It's just, it has to be done in a thought-out way.

Yes. So step one is identify the power of the technology. What's the fundamental capability that this technology provides? And this is easy because it's on the website of the vendor, right? You must define exactly what the technology can do. You know, it allows us to instantly be notified when somebody's late on their rent payment. It allows us to have multiple people within RL. Understand the case file. So not just the property manager, it lets us deploy a pod-based structure, not a departmental structure, because information flows quickly between our lease-up team and our maintenance team. Great, right? Like, those are the kinds of things you want to start with.

Yep.

It allows us to— right, okay. So that's step 1. Questions or comments on that? But like, what is it at its core as the first step? Yeah.

And that, you know, that's pretty straightforward. What's the power of the technology in question?

Perfect. That's the easiest one by far. Okay. The second one is identify how it diminishes the constraint.

This is the hard part.

Yeah. Which is right now you have one constraint at RL Property Management. So before you pull the trigger on that SaaS you read about on, uh, Twitter, that's gonna help you administer pet protection plans, you say, awesome. So the biggest inhibitor to RL's growth is administering pet protection plans, right?

Wrong.

And you say wrong, and I say, okay, I don't know if that's where I'd be deploying your precious discretionary OpEx. You're right, you'd make more money. You're right, it would be easier, and Tammy, who currently does this, but Is it applied to your constraint?

Yeah.

Which I think is just, you know, it's just a wonderful worldview.

Right.

Questions on that?

Yeah. I mean, the phrasing here is a little awkward. Um, when I first heard him say it, identify how it diminishes the constraint. I'm like, wait, why are we diminishing? Okay. Wait, no, that was the constraint. So diminishing is good. So you're like, okay, because you're not used to hearing about diminishing things in business, right? That's not a— but, you know, so it's kind of an awkward way to say it, but it forces you to get clear on what is the constraint and how is this impacting that constraint. Our constraint right now is growth, unit growth. Our churn's under control, we're profitable, not as profitable as I'd like, but we're profitable. Um, we got a great team in place, the technology stack is fine, we have in-house maintenance, like we're doing a good job at property management. We just want more customers and we have capacity for more customers. So switching softwares, whether it's showing software or whether it's process software or whether it's, you know, our main accounting software, that is not going to get us more customers. So this would— we would fail at step 2, which is Rentvine is not going to diminish the constraint. The constraint is need more customers coming in the door.

An external constraint.

Yeah, right. So yeah, I mean, I think the, the thing to realize here is this doesn't mean you do nothing. Like, there's obviously small wins all over the place that are worth pursuing, but when you are considering a large technology change, a large rollout, swapping a SaaS are starting with a new one or, you know, a total reorg of the company. Um, those are things that you need to run through this rubric.

I will put a small footnote in because we could process this later, but I wonder if you say external growth, I would— I'm about to throw a hand grenade into this conversation and I don't want it to derail it, but if you looked at this quarter's rocks, that you just ran through with EOS, how many of those apply to your constraint? Or is it a bunch of local optima you've identified? And I think it's a really important question, which is everybody's moving around a lot and everybody's chasing local optima because they have rocks. And, but how many of those are directed at what you have identified as your constraint?

Yeah, I think there's two problems with EOS. That this sort of uncovers or reveals. The first is EOS does a great job of having you set goals, right? You set your 10-year, your 3-year, your 1-year, and then your rock should support your 1-year. And so there's no issue with the constraint when you're setting goals, setting 10-year, setting a 3-year, setting 1-year. That's how you figure out what you're trying to do. Um, but there's a missing step here because it has you go from— it has you looking at your 1-year goals and then it has you creating rocks. And they do explicitly say the rock should be supporting that goal. But what the step that's missing here is there should be step 1, pick your 1-year goal. Step 2, which I'm about to identify now. Step 3 is select rocks. Step 2 should be identify the primary constraint that's preventing you from hitting that goal.

Yes.

And now step 3, all of your rocks should be focused on that constraint because what ends up happening is when you're like, okay, we're in rock setting mode, you sort of, you're like, well, everyone should kind of get one, one to two rocks. And so they end up being spread out around all these different departments.

Wouldn't it be great if we had a welcome video for new employees?

Exactly.

Great one. I'll build some welcome videos.

Yeah.

And you're just like off in la-la land. Yeah.

So what should actually be happening is there probably should be like 2 to 3 rocks, which the good EOS people do try to limit the number of rocks very aggressively. And those rocks, should not be like one in sales, one in marketing, one in operations, one in HR. No, they should all three be on whatever the constraint is. They should all be attacking it and everyone should be collaboratively going after that.

Yeah. Fantastic. What you're talking about, I read a great book. It's esoteric, uh, but it's called On Operations and it's a military book and it talks about the history of the operational level in warfare, which is there is strategic and And then there was tactical. And basically, in like the 19th century, this idea of linking the strategy, which was invade France, and the tactics were like, here's how you take a bunker. But there wasn't this operational level that linked invading France with how to operate your rifle. And it talked about the development of a general staff and how the operational level linking strategy and tactics is this very rich place. And it's exactly what you just said, which is we've defined the strategy, grow 2x. We defined the tactics, which is here's how you turn a unit. But that linkage between our strategy and our day-to-day behaviors requires coherence, like real coherence.

Yeah. So, so the two, the two flaws in EOS is like, one, it misses that step two of identifying the constraint before setting rocks. And the second error it makes is like the nature of EOS encourages everyone to have one or two rocks.

Yeah.

Which the way most companies are organized, that, that ends up spreading those rocks.

Local optima.

To a bunch of local optima. Yeah.

I would, I would also say as kind of a corollary to that as well is when I look at the problem identification process, you know, idea, IDS.

IDS. Yep.

IDS. I know one of them is when you get diarrhea a lot and one of them is EOS.

That's IBS.

Oh, IDS.

Yeah.

Okay. BS. But I also think that not all problems are created equal.

Yeah.

And so it's like, hey, I've got a problem. We did— it's like, some fires gotta let burn, man. And in the EOS architecture, I find problems that are against the constraint are not given— they should be giving 20 times more effort than somebody who— an email went to spam with a customer. It's like, bro, that does not need to be next to— we're not calling leads back quick enough.

I've said for years that one thing I hate about to-do lists and to-do list apps is all of the tasks have the same font size, and it gives you the perception that they all have equal weight and equal value to whatever your goals are. I want to design a to-do list app that lets you set each item with like a different font size and weight. So like, there'll be one item that's like, remember to pick up milk on the way home, and right below it will be like, solve the marketing problem. It'll be like 10 times the size, uh, so just to reiterate in your mind that these are not equal. These are not of equal value for you to solve right now.

So can I go on a small, small esoteric aside?

Okay.

Before we get to the second two steps of technology, because it ties in what you just said, uh, a book from 2008, I read it when I was in college and it was called Gaming the Vote: Why Elections Aren't Fair and What to Do About It. So I'll read you like the first sentence. Uh, electoral system is fundamentally flawed. But there's a simple and fair solution. At least 5 US presidential elections have been won by the second most popular candidate. The reason was a spoiler.

That's gotten way worse since 2008, by the way.

A minor candidate who takes enough votes away from the most popular candidate to tip the election, right? Ross Perot, George H.W. Bush, Bill Clinton, right? Okay. And this book goes, uh, electoral system by electoral system and saying first across the line, uh, multiple votes. It looks at all these different votes.

First past the post.

Yeah, first past the post. All these different ones. And it Basically, one of the core issues is what you just said, which is there's no gradation. I like George H.W. Bush 8, Bill Clinton 6, and Ross Perot 2. That is not captured in our— and after that, at the end of this whole book, it's so brilliant. He said the best single voting system is hotornot.com, which is you, you go to every candidate who's running and you can give any number you want between 0 and 10, and you can give out as many 10s as you want, as many 0s as you want. You could go 10, 0, 0, 0. You could go 10, 9, 1. You go 6, 6, 6. You can— and you add all those numbers up, and that is the most reliable way to match voting methodology to preference. And it's exactly because of what you just said, which is there's no, like, stratification in preference. Super interesting, right?

Yeah, that is interesting.

Like font sizes. Yep. Sorry for the political science.

Uh, no, that's good.

You know, uh, cold sec. Okay. So back to, uh, back to the, um, 4 steps. Okay. So number 1 was what does the technology do? It should be on the landing page. Number 2 is identify the limitation diminished. Now the last 2 are the brilliance. The last 2 to me are the brilliance. You could have inferred the second one. These last 2 though are so stunningly good. Okay. Number 3, identify—

look, for anyone who's listening and hasn't read this audiobook or listened to this audiobook, like, what would you think the next step would be, right? You would think like, okay, the next step is like roll out the thing, get a good price or something. Yeah. Like, you know, roll it out and then measure the results, right? You kind of would think that would be like step 3 and step 4.

And have a smart goal.

Yeah.

But with the technology's use.

So, but, but what it actually is is not that it's kind of not that, not something you would have thought of intuitively.

Step 3. And this is where I'm just like, oh my God, my brain hurts. Identify the old rules, the accommodations. What rules, policies, or behaviors did we develop to live with the old limitation? Long before the technology arrived, your team created workarounds to cope with the limitation, by the way, implicit or explicit. For example, if data was slow, a rule might be to hold large weekly meetings to manually update them. These habits become deeply embedded in the culture. I need to go smoke a cigarette.

Wait, wait, I don't smoke. Better for you, John. Do you know why this section is in here? No. It's because the old ways of doing things were accommodated because of enormous transaction costs.

Oh, don't you say— don't say that word.

So when the new system comes in and it lowers the transaction costs of data, information, communication, those transaction costs are no longer there or are immaterially small.

You're exactly right.

So you need to rethink about how these systems improve.

Well, that's step 4. But at step 3, you have to say, we— like, I'll give you an example. Now that we've switched to Appfolio, The old rule was we only did owner disbursements monthly because we had to do all of this manual reconciliation on maintenance bills plus our markup. And so we run— we, we— an informal rule based on the how high the transaction costs were was we run owner disbursements once a month.

I want to— let me give you a literal real example from my business.

I have one too.

Okay. So, and I hope my team doesn't take this the wrong way. It's just, it's just how business goes. But so for years and years, for the first 6 years of the business, you know, we did all in-person lease signings. Okay. So someone applies to rent an apartment, we review their application, we approve it, we contact them and say, hey, your application has been approved. Come on in to pay the deposit and sign the lease. So this is like from 2013 to 2019. Maybe 5% of our leases would, cuz we would give them the option for e-signature, but nobody would do it cuz they want to come and see and they want to sit down and we want to read them the lease. We would really push the in-person lease signing cuz we wanted to explain the lease to them. 2020 hits and COVID. Okay. No one wants to come to the office. No one wants to be in the office. E-signing takes off like a rocket. We go from 5% e-signing for leases to literally 100.0%.

Wow.

So that change happens and a couple years goes by. Now, back when we were doing in-person signings, we would schedule the signing with the prospect and then that would go on the calendar. And so there would be a calendar event on like our company calendar that would say lease signing 10:00 AM on Tuesday, and that conference would be blocked. So fast forward to COVID. And even a couple years after COVID, I got to looking at our calendar for the company. I'm like, why do I see all these calendar events for lease signings? Aren't these all electronic now? And they, and the answer came back, yeah. Oh yeah, those are all electronic. And I'm like, so why do we have a calendar event for an electronic lease signing? And they said, well, that's a reminder for us. So that's when we actually send out the lease for signature. And I said, well, isn't it the case now that when we have an approved applicant and they go ahead and they're ready to pay the lease and pay the deposit and sign the lease, that there's no, like, there's no real reason that we would need to schedule.

We'd be instant.

We want to send them the, you know, the lease for signature as soon as possible. Right. This is a customer we're trying to sign. They could be touring other apartments. They could like, who knows what could happen, right? Time kills deals. And they said, well, that's true, but, you know, the system was working and it really hasn't presented a problem right now. I was like, Guys, I wouldn't be doing my job here if I didn't just insist that we do away with the calendar events and we send the lease for e-signature within the hour that they are approved and agree to move forward. Like, that is how you handle a customer who's ready to sign. You send them the thing for signature and you collect the deposit. So that was an example of the old rules and how they were just left in place because that's how we were operating. With a constraint around e-signing. You could pretend that we like adopted some e-sign technology or something and it wasn't COVID.

I mean, you, you start, yeah, I mean, you essentially did because of COVID like it was like a new technology even though it wasn't a purchase. Well, let me give the fourth, because third and fourth are related, and then I want to give my example. The fourth, of course, is identify the new rules. What rules should we use now that the limitation is gone? This is the most critical step. If you install technology but keep the old rules, the technology becomes a burden, not a benefit. If you now have real-time data, you must abolish the large weekly meeting and replace it with real-time response. Or if you now have e-signing, sending leases, right? The example I'll give, and again, I'm victim to this, this was like 4 or 5 months ago. We changed our process so that I think the first thing our candidates have to do is upload a video. Um, because we just saw that that's like a sticky wicket in our process, which is we need that. We can't even. We can't even look at anything until they get that, right? Mm-hmm. But our talent advisors were, because we didn't used to do that, were going and looking at raw candidates, trying to figure out who we should invite to the next step. And it was this huge volume of candidates and it was sucking up 50% of their time instead of saying, guys, do not look at anyone who has not uploaded a video because that is now the mandatory first step. But we forgot to update the rule set. With the new video upload requirement. And so we didn't harvest what I knew would be the gains, which is, look, 30% of our applicants are gonna upload their videos. And so I was able to reduce workload at that screening stage by 70%. Yeah. But only if I updated the rule set.

Yep.

And I forgot to update the rule set. I was just like, hey, video uploads now. It'll be great. And we didn't harvest it in terms of productivity improvement. So that was, again, this was 100 days ago or something like this. This is very much a live issue for us. Yeah, but this just, this blows my mind, the wisdom and simplicity. It blows my mind.

Yeah, I mean, it really puts to words something that a lot of business owners intuit, but it's very difficult to articulate to your team. And it's difficult to even to explain to somebody else, like you and I talking before we had these words, like, you have these strong intuitions around Why isn't my team getting it? Like, don't they understand that when we did this new CRM, we wouldn't need to like physically be taking notes and then scanning them in? The whole point is there's an app on your phone. Like, how is that not obvious? Right? But no one cares like the business owner does, and no one has the context that the business owner does. And so having this framework is a great way to put words to something that's been frustrating for me at times., in dealing with a complicated business with a bunch of people?

Well, I like to say values drive policy. That's right. And policy drives— I forget the middle word— process. And process drives technology.

That's right.

Did I get that right? Yeah. Is that a Peter Lohmann original?

No, um, that wasn't me originally. That, I believe that's a Jordan Moeller original.

I like that. Yeah, I like that. He's got a future. He's got a future.

He's going places.

He's going places. Um, so there's two kind of corollaries off this that I think are really worth like diving into. Uh, where was it? Okay, the, the first is that I'll give it a specific example, and I think it explains something that we're seeing every day, and the data shows this. There is wide adoption of ChatGPT in the workplace, wide adoption. But people and business owners and consultants are seeing it is not driving increased outputs from the workforce. Why? Because writing emails to tenants is not the constraint. Mm-hmm. Unless you can show me that the reason we don't have more units is because we can't write 4-sentence emails, or, or the time of back office personnel is our most limited resource and we can't find any more back office personnel.

And so we need to deploy them with technology that helps them respond to correspondence faster, because that's the hard part. Mhm. And that's why you're seeing this mass adoption, but you're not seeing it drive better revenue per headcount or better unit growth or lower churn, because it's not against the constraint. And I think there's going to be this— and pandemic of but people and business owners and consultants are seeing it is not driving increased outputs from the workforce.

Why? Because writing emails to tenants is not the constraint. Unless you can show me that. The reason we don't have more units is because we can't write four sentence emails or the time of back office personnel is our most limited resource and we can't find any more back office horses came out. Yeah, I mean, this is a tale as old as time, is like, how do we— what, you know, rolling out of new technology, which is what we're talking about, the 4 steps in the framework here, how do you apply a new technology within a business in a way that's actually meaningful? Whether that technology is a desktop computer, or is the internet, or is a cloud-based solution, or a Model T instead of a horse. Yeah. Right?

And so, so that's where I'm just like, I, I'm like, oh my God, I'm so happy I read this right now because we're, we're doing some big stuff in AI and automation. We're about to come up with some more services around that. But this gives me such a great way to not fall. He says it in the audiobook, but he says the early adopters of ERP were like kings of the world because they understood the first principles. But all the late adopters would spend millions of dollars on implementation and would get no, no increase in productivity. Activity because they didn't understand this sequence. And I'm like, oh my God, there's a lot of wisdom here for product design as well. Like, think about the original iPhone, how it was— it did away with keyboards, right? I mean, the, the old phone manufacturers were operating with that old set of rules, which is that, well, touchscreens either aren't available, aren't affordable, or suck.

Therefore, we need a physical keypad Physical key for every letter and then a bunch more on top of that. But when good affordable touchscreens became a reality, it was time, which is a major technology shift, it was time to reexamine the rules against the limitation of input on a, on a device. Building on what you just said, the one that I'm spending, that's keeping me up at night right now is when, let's pretend you have an external constraint. I'm gonna use your example cuz some of the stuff we're doing at Sigm right now, I, I think is on the very cutting edge right now. And it's in 10 years it'll be table stakes, but I will be hopefully on an island somewhere by then. That's how, so let's look at your company and you say your limitation right now is door growth, right? Let's pretend that that's what it is. And you see this, you know, AI and automation and LLMs and vibe coded whatever. You, you see that. The question that I'm having right now is, when you think about the services you provide to your owner investors and how you go out and acquire them, what things are now easy and fast for you that rewrite the rules of what a property management company does and how they acquire their customers? For example, You could build a custom mobile app for every single one of your owner investors for nothing. You can redefine what the value proposition is. Yep. And you could say, by the way, uh, when you sign up with us, we build you a custom mobile app. And you know, that offer is going to cut through the noise of somebody else giving a free rent report in Columbus, Ohio. And you're, because you've You've redefined the value proposition and how you go out and reach out and touch your customers because you understand that historically building a mobile app would have been $50 grand and now you can— that's where my mind is like, is redefining value propositions and customer acquisition.

Yeah. You want to find things that are easy and cheap for you to do, but of high value to a customer or a prospective customer. So just to take your example to the extreme, we could just build an app for every prospect, correct? Just say, here's your app, we built you an app for free, just here, have it. It's got your property address loaded in, it's got your county tax record in there, we shopped around some insurance quotes for you, it's all pre-built, take it, use it, have fun with it whether you use us or not. Now if you sign up with us, it's going to get automatically populated with a bunch more data, water bill in there.

Yeah, and what's that app cost you now to build $50, right? And you now start reaching out to real estate agents, and instead of just saying, send me your customers, you say, hey, I want to build free— I want to build free apps for all of your pe— now you're like, oh my God, you know, what would that do to your customer acquisition more than, you know, join our weekly webinar to learn about rentals?

Like, and the point isn't if that specific idea is good or not. It's more about like, given generative AI technologies, whether that's ChatGPT or Replit or Lovable, given that is now affordable and fast, what can we do against our constraint? Yes, constraint being new customer leads, closing rate of those leads, relationships with affiliate partners and referral partners. Yep. Like, how do we stand out in the marketplace using some of this new technology? Because you're exactly right, the market rent report is so antiquated and just everyone's getting a free market rent analysis with every property manager they talk to.

What if you go around and using Google Maps, you take pictures of a house that you know is third-party owned and operated, you go into Nano Banana and you have it, uh, redone as Rembrandt style, and you print it out on thick stock in your office and you mail it and you said, We think your house is a work of art. You deserve a property manager who will do it. And you send it for $15.

You're so good at this stuff, John. I'm so glad this is recorded because last time I got you on a roll of good marketing ideas, it wasn't recorded and I was furious because it was like gold in there and I never could recreate it.

But that's an example. But they think you hired an artist to hand sketch in charcoal and you say, We believe your house is a work of art, and so we wanted to give you a piece to commemorate it. If you're ever thinking about, you know, upgrading your property manager, we'd love to earn your business. So good. Here's my cell phone number. Now, what did you just do? You just took 1/10 of a cent of generative AI and a screenshot of Google Maps, and now you have a no-shit way to, in an interesting way, get in front of these owner-investors, which right now you're like, oh, I don't know, we'll do a webinar.

Yeah. It's a conversation starter, right?

Yeah. It's just, I love, I love taking this stuff, technological innovation, and using it to push the limit of what's capable, not just making faster horses. Hey, we, we give you a live bang, bang, bang within 24 hours of when you join. Like, I just love pushing it, pushing it, pushing it, pushing it. That's where I get a lot of satisfaction against the constraint.

Yeah. I, I think predicated on being able to successfully do what you just described, which is creatively brainstorm realistic ideas against the constraint using the new technology. You have to be quite familiar with the new technology. And this is, I think, a problem for a lot of operators is they're only familiar with the new technology to the extent that the salesperson did a slide deck. And so when you, if you're considering, you know, switching property accounting softwares or rolling out LeadSimple or some big new marketing campaign, like you have to be familiar, deeply familiar with that folio or LeadSimple or ChatGPT and, and some of the other LLM services, because if you don't have at your fingertips and sort of in your intuition, what, like, what they're capable of that wasn't previously possible, you're not going to be, you're not going to be able to identify the rules that no longer should apply, or potentially even like the power of that and how it can apply against your constraint. So coming back to that step 1, I think there's a little more to it these days. Like, you can't just read a slide deck about a new machine and understand that it goes from 10 widgets a minute to 20 widgets a minute. You actually really need to understand, like, what actually is the difference between Buildium and AppFlow. It's not, it's not easy. Like, you can read speeds and feeds and stats and figures about, like, you know, the product, you know, the bullet points of what the products can do, but I, I don't even know how you develop that. It's like you have to play. I can tell you.

I can give you a way. Well, you just said the first thing you just said is the word is play, right? The word is play. Now I'll tell you why this is even more important with AI and automation. The technology builders themselves don't know what they're capable of. Yeah. So there's no, you know, we see hiring requests all the time, like knows AI. I'm like, bro, nobody knows AI in roofing. Nope. Like how to deeply use AI in an innovative way in roofing. The guys who built it don't know what it can do in roofing. Hey, did you realize you could take screenshots of roofs and pipe them into an AI model and identify which roofs are old and which ones aren't? They don't know that the model can do that because literally 3 months ago, the resolution on images got good enough for an LLM to read. You think they're tracking a list of roofing use cases?

Mm-hmm.

Of course not. And so it's even more exacerbated by the fact that, that the builders themselves don't know what they're capable of. And so if you're interested in pushing this, You have to get in there and play and try and fail and say it was unreliable at this, but it was reliable at this. If you're at all interested in pushing this limit early— now, in 5 years, there will be a productized service that— and they're going to charge you 50 times and it's going to become table stakes. And— but at this stage, there is so much, which is why I'm being careful what I say. There's so much greenfield to go out and build a billion-dollar company before anybody even realizes that AI can look at roofs from Google Maps images and you don't have to be a hedge fund anymore.

It's so early that if you're the guy who figures out how to read roofs with Google Images using ChatGPT, you're probably better off productizing that solution and selling it to other roofing companies than you are just using it in your own roofing company. Like, that's where we are in the cycle. And so that's kind of terrifying because it's just, there's so many unknowns. But it's also a huge opportunity for people who are entrepreneurially minded to be like, okay, I'm going to— I am going to figure out how to use new technology to grow our leads faster. Um, once I do, there's an argument of like, okay, well, we're going to pull that into CRANE and we're going to make that available to CRANE members now, um, as just an added on, you know, member benefit.

The business model to exploit that is really important, which is a whole nother topic.

Yeah. Yeah. So yeah, that'll add, you know, 30% growth rate to RLPMG. But if we now make it available to Crane and it halves our churn rate at Crane and people are hyped about it and now more people are joining, like, yeah, I mean, that's probably a different discussion of like what vehicle is right to exploit some of this in.

But which I've spent a lot of time thinking about too, because I'm on the front— I consider myself at the very, very tippy top of actual— I mean, you've seen what we're doing under the hood. Would you say in the recruiting space as a non-technology person, would you think I'm top 10 at AI and automation in the recruiting space.

Yeah, definitely.

I know I am. I think I'm probably higher. Like, not I, we collectively, like in terms of what we're pushing. And I'm pretty decent in other sectors too. You know, I could, hey, roofer, talk to me about your problems because I know the tools so well now. I'm like, no, no, don't do that. That won't work. Try this. Use Google. No, no, use, use Replit for that because they have a built-in database. Like, I'm pretty good at like that, but it's just come from hours and hours of playing. You know, and colliding it with kind of like my day-to-day work.

Yeah. So it, it requires being intimately familiar with your constraint and intimately familiar with tools that might solve it.

You, you gotta, yeah, the sector and the constraint. Yeah. And then the tools that are potentially available to solve it. Like I'm telling people things that they didn't even know AI could do. And I'm like, really? This is like 4 months old. And they're like, I'm like, oh yeah, I spend 50% of my waking hours playing and learning about AI. Probably 50% because I see so much, you know, whatever. Okay, so I'll wrap this up. In summary, number 1, identify the power of the technology. That's easy. The vendor will tell you what does it say it can do. Number 2, identify which limitation it is working against or make sure it's working against a constraint, the constraint ideally. Uh, number 3, identify your old accommodations. What stuff did you used to have to do because you could only sign leases in person? Well, we scheduled them on the calendar. Well, we had only 5 interviewers, and so we only invited people that had already been manually screened. Okay, that's the accommodation. And then number 4, identify new rules. Now that this is no longer a constraint, the new rule is that as soon as somebody signs the DocuSign, we immediately send them the official lease. And it's instant. It's an automation now. It is not scheduled by a back office person.

Right.

So that is in summary what we talked about today. What are you smiling at? Did somebody leave a comment?

I guess I don't know what I'm smiling at. I'm just, I'm just happy to be on the show with you, John.

You're just excited about the sweater? Yeah. Okay. Um, I like doing it. I kind of maybe want to start doing this like the Emmys. Like, who are you wearing today? Who are you wearing today? Okay. Yeah. Like, where's that from? I love it. Oh, it's Versace. Uh, okay. Uh, in closing, final thoughts, comments, questions, technology.

Um, so I'm consuming like in parallel, an audiobook, a Kindle book, and a physical book all having to do with theory of constraints right now. But at the same time, I'm reading about, um, I've been reading a lot cause you've kind of, cause you've been sending them to me, but I've been reading for many years anyway. Uh, the Common Cog articles about becoming data-driven and the weekly business review that Amazon does and stuff. Um, I really want to talk about this XMR chart. Did you—

you were digging?

I was, I was texting with you.

Uh, yeah, you were talking about it, and then I was talking to Cedric, the guy who wrote it, basically about saying— yeah, I mean, the XMR charts.

Yeah, I'm pretty sure Cedric is, is talking at Holdco Conf, by the way.

Oh really? He's— yeah, Tim Ludwig got me onto him. Yeah. And I'm like, okay, this guy's—

if I can— this XMR chart thing Yeah, like this is new to me. I didn't get this far down in this 10,000-word article last time I read it. Um, I— it's like I just discovered candy or something. I need to start putting everything on XMR charts. I don't even understand how they work yet. I just read enough of it to be like, okay, this is what it's really—

I can, I can describe it to you. And I spotted an issue with it. I was very proud of myself. I spotted an issue with it, and he wrote a whole subsequent offer, to talk about metrics. So basically what he says is when you look at data, um, oftentimes it looks up and down, right? And that's just essentially normal variation, right? You know, whether your sales guy closed 6 leads or 3 leads, you don't, you know, yank the joystick to—

So my solution to this for a long time has been, uh, we look at everything in terms of like, cause the week to week variability in all your business metrics is all over the place. Correct. So my solution for years has been, We look at a trailing 2-week or a trailing 30-day, which smooths it out quite a bit.

Yeah. And you can layer a line on it. But one of the things he says is the issue with layering a line on it is you remove one data point, the line looks completely different. Like it's very, you know, if you don't have massive amounts of data, it's a super sensitive analysis. So what he says, and then I'll tell you the issue and then the answer he gave, which was, he says, what you do is you track your data points and you create a range in the form of an XMR chart, what accounts to non-intervention-driven variation. Because he defines knowledge as your ability to predict things that are going to happen in a business. And if there's normal variation, that's throwing a dart, right?

It's normal.

Did you— is it because you changed the commission plan that that number went up, or normal variation? And without an XMR chart, you can't tell the difference. Oh, it went up and to the right. We did, we intervened successfully.

Right? Yeah. And there's a bunch of science behind how these estimates are constructed, having to do with like 3 sigmas, or I mean, it's some, some sigmas.

So what he says you do is you basically go back, and you backfit a range to what accounts for normal variation within this number, right? So it might be 6, but it might go between 13 and 2, with normal variation. Right, right. And then when you're planning interventions, if it doesn't drive a change that lifts you out of the normal variation, it is an ineffective intervention on that metric. But if it's within that number, you go, that's just noise. That's statistical noise. Yeah. So my intervention to change the commission plan was unsuccessful, but it's very clear to tell the signal from the noise because you've now established— XMR chart's too fancy of a word. Just call it a range, a normalized range.

That's what it means. Yeah, the thing that I didn't understand or haven't dug into yet is there's two— actually two charts. There's the xplot and then the mrplot. That's where the xmr comes from.

Yeah, x is the data and mr is the range.

Yeah, I don't understand the mr chart.

It's the range, I think. Whatever. But the issue that I had with it was I said, yeah, but this doesn't work if in the last 12 months you've been trying interventions. Because that gets baked into your normalized range. So what happens if I changed commissions 3 months ago or launched a new market? And he basically said, you either have to throw out data points. He wrote a whole article about it because I spotted it right away. I was like, what if I changed something 6 months ago? That gets baked into my backwards-looking range. And so he said, you either have to pull those data sets out or look for a period of time when you didn't do any interventions on the data set. I was like, okay, that makes sense. But that was why I was like, okay, this doesn't make sense to me. But it was a fascinating topic. We should unpack more of his stuff.

He's got a subscription, he's got a membership. We should— one of us should join it, like a paid community membership thing.

I actually have this idea. You should rip it off. You should license his private stuff and give it to Crane members. I think that it's a really good idea, licensing gated content and giving them—

we got one of those in the works.

Peter's, Peter's working on something like that right now, so that's something. Okay, final thoughts.

You should license it for Sagan members, dude.

You just want that because you're a Sagan member.

Yeah, totally. I can't afford the $90 a month.

Yeah. Uh, okay. Uh, final thoughts, tips, quotes, technology.

That was my, that was my closing thought.

Um, my closing thought is, um, I don't like being on my phone when I'm with my kids or my wife in particular, but I also have anxiety if I'm disconnected. And I never really had a great answer to that. I tried to do a separate phone, but nobody called that number. They just called my main line. And I actually think that the single best use for an Apple Watch is to be connected without having access to Slack or email. And so anybody can text me, anybody can call me, and I'm out on a walk with my kids and I do not have my phone with me.

And that's cool.

I like that. The other thing I will say as an added bonus is I now officially have the fastest capture in the world, which is I I reprogrammed the side button. If I hit that, it's a voice memo right away. So I now have one button fast capture, which is— if you go back to our GTD episodes, it's something I'm obsessed with.

Okay, so let me ask you this. You are obsessed with it. I have literally a printed out tweet from you from like 5 years ago talking about fast capture. Like, that was one of the first things I ever read that you wrote, actually. Um, great. But let me ask you this. When that voice memo So you record the voice memo, I get it. But then what happens?

It goes into the cloud and my voice memos effectively are an inbox that I routinely—

Okay. I hate that. Why can't it, can't it fire into a Gmail inbox or a Todoist or something?

If I built an Apple shortcut that said every time a new voice memo is added to this folder, transcribe it with Claude. And then automatically add it to my— dude, that's the answer I've been looking for.

Absolutely.

You can do that.

That's easy.

That is cool. But that's what you do is you just create a place where the uploads or whatever go to, and then you just have that be the trigger of a Make automation that takes it, transcribes it, categorizes it. Oh, it's an audio file.

Pretty, pretty straightforward. Yeah. I need to set that up.

But I love it because it's just like when I'm out in the ocean or whatever, I can just Or if I'm—

yeah, when I'm in the ocean too, I always am thinking about the same thing.

Sorry, sauna.

Sauna.

When I'm— when you're in your new barrel sauna.

Good, good episode. All right, good episode. Talk soon.