Why Your AI Isn't Paying Off

**Mickey Anderson** (0:00)
The real competitive advantage isn't just great people or great tools. It's how you put it together. It's the recipe, the secret sauce that you have that gives you that competitive advantage. And most organizations are not leveraging that or investing in their people in that way.
Most weeks, I sit with a business owner who is partway through an AI implementation that is not delivering what they expected. They bought the tools they were told they needed and hired or borrowed AI capability, and they picked a few areas to automate. The pitch was efficiency, headcount reduction, cost savings, and faster output, as you'd expect. But six to 12 months in, the team was still struggling with adoption, the data was sitting in silos, the token spend was higher than the plan said, and the growth numbers just hadn't moved. In the next 10 minutes, I'm going to show you why this is happening across every kind of company right now, and what you can do from your seat to make sure that the AI you've bought or are about to buy actually pays off. One thing I want to note before we jump in is I have some really exciting news about UNLEASHED and our future. I will explain that at the end of the episode. I'm just bubbling with excitement about the future here. We've got so many cool things on the go for you to help you build a business in a decade no one is prepared for. Let's dig in. Most of the time, we're not choosing wrong tools. The technology isn't bad. The AI isn't the problem. Usually, we're just not set up to be able to actually leverage that speed properly.
Now, I'm sure you've experienced this. I see it pretty much every day I'm working with every company that I work with. The AI is doing what AI does, speeding things up, and it's accelerating what your team is already doing. Now, the trouble is that the team and the organization haven't got the foundations in place to actually leverage that speed. It's like getting a five-year-old on roller skates and giving them a turbo engine. They're going to go fast for a couple seconds, and then they're going to skid out and fall, right? We need to prepare them to be able to handle that speed, and that takes time and growth and foundations, which is exactly what we're going to talk about today. So, when we're talking about what your team is doing, most of the time, what's happening is you're identifying a place to automate or speed up within your organization, and so you're throwing an AI tool at it, saying, this is going to speed things up. But you haven't actually set the standards for what good looks like and how that's going to be leveraged, where it sits, and how it impacts the rest of the organization. We think that small pilots in small areas are less risk, and it's going to create faster impact and higher adoption, which it does create speed and higher adoption, but it also creates massive impacts across the organization. Who can access data? Who owns what consistency looks like? What your standards are? That's all fuzzy. And so today, we're going to talk about really understanding what we taught in the last episode, which is about diagnosing the problem, and how you can apply this same perspective on AI implementation. I want to talk about an example. We worked with a B2B service firm recently, and they had bought this really amazing AI platform that promised incredible speed on their delivery side. It was going to deliver 40% reduction in delivery time, allowing them to not have to increase head count and really increase capacity and support the growth that they were really pushing for. The problem was six months in, the team was using the AI on the smallest portion of their work, and delivery speed hadn't actually sped up. Small tasks had sped up, but there was gaps in between the tasks where work was compounding, decisions needed to be made, coordination needed to happen, reviews and revisions needed to happen, that was actually taking up more time than it was previous to the AI.
What was actually happening underneath all of this was the team had grown speed and grown capability faster than they could actually build the standards they needed to scale. What I mean by that is each member of the team was doing client work slightly differently. Each client had different expectations that lived in that account manager's head, and the AI was filling the gaps, right? It was doing what it does best, speeding up the inconsistency between the different people, functions, and components of the organization. What we did is kind of contrary to what most would say. We actually paused the AI for 90 days, and our job was to align the leadership team on what good actually looked like. We set standards across the team, clarified ownership within the business, and where information lived so anyone could find what they needed quickly and effectively. And then once we had that solidified, we could bring the AI back in. And when we did, delivery time actually dropped by over half. Because the AI was now reflecting the work, the team was already doing, and the standards the same way. So the AI wasn't broken. It was the foundation underneath and the people within the organization who weren't supported or connected well enough to really leverage the AI for what it was. It's one of the reasons so many companies struggle with giving their team members an LLM to leverage to speed up work. It doesn't connect to the rest of the work of the organization. It speeds up individual tasks, but leaves gaps between them that really existed before, but are now amplified by that speed. And so the gaps are becoming buyers that constantly keep popping up, inconsistent work, and it's causing challenges with adoption. It's causing challenges with standardization and creating those efficiencies that you want to create. Now there's four things that we've identified that you need in order to let your AI truly pay off, and it's much simpler than you'd think. The first is tied to the last episode. We have to diagnose the true problem, not just the symptoms, not just picking a task because it's easy or simple, but really understanding the true problem we're trying to solve, or the opportunity we're trying to tap into with the tool. And that diagnosis is incredibly important because it's very easy to see symptoms and opportunities, but to really understand the leverage, the risks, and the implications, we have to look at the bigger picture of the organization. The second, a little bit harder, but we need clear standards across the organization. What I mean by clear standards is what good looks like. Really written down somewhere so anyone can find it and refer to it, and it's consistent to the AI, and the team members have clarity on that. We need to know who owns what and where it lives. This is really important. Now, the third is the team aligned on how they work together. Speeding up one task in one role in one department is only going to do so much, especially if it's causing a ripple effect across how the functions work together, what that looks like. So everyone needs to know who owns what, how information gets shared between the functions, how decisions are made and who makes them and what type they are, and then how silos are broken down and resolved before the AI gets connected to any of them, because it tends to amplify those silos more than anything. Now, the fourth is consistent operating behavior across the team. Slightly different. So when we're talking about standards, those are shared organizational standards on what good looks like and how we do things here. We're talking about operating behavior. We're talking about the day-to-day rhythms, the process, the steps, the behaviors that the team and the roles have and do every day across the company, because AI is going to accelerate them whether you like it or not. If we have certain individuals who spend more time in revisions and are more thoughtful and really go through information, and then we have some who just hit plug and play and go, that's going to create some differences. If we have some behaviors where communications or information is shared in a chat on Slack and some where it's not, again, the gaps between the two are going to increase because we've got speed now pumping up that information. When those four are in place, AI does what the vendor says it will do. It really does support speed and growth within your organization, but when you don't have those in place, it compounds friction and creates massive gaps within your organization. It also will end up with additional token spend that you weren't planning for, inconsistent adoption rates, inconsistent usage, and really the growth numbers that you were looking for end up flat 12 months in. So here's a quick test that you can do with five questions you can run from your seat in order to identify whether or not AI can support you and what to do in this place. You're going to pick a piece of work that AI is currently doing in your company, or one you're investigating or about to add AI to. And then you're going to ask these five simple questions. You might want to get a pen and paper to write these down. The first, who owns this work today? Could you name a single owner or is ownership distributed across multiple people in an undocumented way? Two, has the company set a clear standard for what good looks like? And could a new team member learn it without sitting with a senior person for days or weeks? Three, where does the information the AI and your people need to work with live? Is it in one connected place where the team can use it or is it scattered across systems that don't talk to each other? Question four, if you asked five different team members how this work should be done, would you get five different answers or one consistent answer? And that's one question that you should be asking across the board frequently within your organization. Now, if you can answer the first four really clearly, and the answer to question five is clear and consistent work, then your work is ready for AI.