Transcription
80% of AI projects fail. According to a recent Mackenzie study, most businesses, they're implementing AI backwards and hemorrhaging money. They see a demo, they get excited, they plug it into their operations and hope that it works. 3 months later, the tools collecting dust, and they're back to manual processes, but there is a 20% that is getting massive returns, 100% plus return on investment, scaling without hiring, eliminating bottlenecks that were bleeding them dry.
What's the difference is they start with diagnosis not technology. They find the expensive problem first. Then they quantify what it is costing them and then they prescribe the solution. And that is what separates successful AI projects from those that are just going to be draining money.
But here's what most business owners do not know about AI implementation. The technology, it's the easy part. Building the automation usually takes weeks, but figuring out where to build it, that is what determines whether you succeed or you fail. As I mentioned earlier, roughly 80% of AI implementations, they fail to deliver any return on investment. And that's not because the technology just does not work. It's because businesses they are solving the wrong problems.
So with dental practice, they buy an AI writer where their actual problem is that leads are sitting unanswered for 4 hours, losing them 30% of potential patients. an HVAC company. They implement a chatbot when their real issue is that after hour calls goes to voicemail costing them $90,000 annually and miss jobs. The tool gets shelved after 2 weeks. The executive team decides AI just doesn't work for us and they go back to bleeding money.
And that's why before we get into long-term engagements with businesses or when a business owner just doesn't know where AI can actually help them, we run through this six-step framework. It forces us to find the pain first and then quantify the cost and prove the return on investment before touching any technology. And this isn't just my process. This is the process that separates the 20% of successful AI implementations from the 80%. And if you are a business owner considering AI, this framework will save you from wasting money on solutions that just will not move the needle. We have implemented dozens of successful AI systems and every single one has started with this exact framework.
So, in this video, I'm going to be walking you through our six-step process that we go through with seven and eight figure businesses to deliver complete end-to-end systems that end up saving them time, increasing their bottom line, and allowing them to truly scale the business. And make sure to just stick around to the end because I will be giving you guys some free resources that you can actually walk away with.
Step number one, define clear outcomes. Now most discovery calls they start with the client saying like we need AI or we want to automate our operations and that's not an outcome. That is a vague wish and if you cannot measure it you cannot improve it and if you can't prove ROI you're not getting paid for the implementation. So here's what clear outcomes actually look like. So saying things like cut lead response time from 3 to four hours to under 5 minutes or increase conversion rate from 20% to 24%. Another one is save property managers 12 hours per week on repetitive tasks or respond to 100% of afterhour calls instead of just 0%. And if you actually notice the pattern, each one has a current state, a target state, and it's measurable.
So when we started a project with a property management team that we have worked with in the past, they said to us, we need better communication with tenants and that obviously is not an outcome. So through further discovery, we actually translated that into reduce tenant question response time by 68% and increased lead retention from 70% to 90%. And now I can actually build a solution and measurable results and prove a return on investment. And the way that you extract these in discovery calls is just by asking what I call outcome forcing questions. So asking things like what is the most expensive problem that you are dealing with right now? Or if we could solve one thing that would move the needle on revenue, what would it be? or where are you losing money that you know about but you have not fixed and you're not asking what they want to automate. You're instead asking where they are bleeding money. And once you know that outcomes they essentially write themselves and every outcome it must have a dollar amount attached. So instead of just the response time saying like it's just faster. We say it is faster which means that we capture 20% more leads which is worth $161,000 annually. That's the shift. Obviously much better.
Now step two is mapping how the business actually works. And this is where most agencies and companies actually fail is they talk to the CEO or the founder and they just get highle overviews and start building. But the CEO they know the destination not the road. And it's the people actually doing the work knowing every pothole every detour every place that things break down. On any one of our client engagements we do not just talk to the executive sponsor. We get on 10 plus calls. across their organization like their department heads, their individual contributors, the people in the trenches, right? And that is how we uncover the real bottlenecks. So the questions that work, they aren't just what do you do? People will obviously just give you the sanitized version of this. You instead have to ask them, okay, well can you walk me through yesterday morning when you got to the office? What did you do first? Then what? Then what after that? And when you ask about yesterday, well, they tell you what actually happens, not what is just supposed to happen.
Now, we were working with a sales team in the past where the vice president just told us like our reps, they're spending their time selling. But when we actually talked to those reps directly and ask them about their actual day, here's what we found out is from 9:00 a.m. to 11:00 a.m. manually building lead lists, checking LinkedIn, cross referencing with Salesforce to avoid duplicates, copying data field by field, two full hours before they even started selling. The vice president, they had no idea. He was just trying to optimize the pitch while his team was, you know, drowning in data entry. But once you actually have these interviews, you just needs to visualize it. So we often just use Excel draw, you know, mermaid diagrams essentially or even just a simple whiteboard, you know, boxes and arrows just showing how work actually flows. So for that sales team, the map look like this. So the SDR they get target profile then they search LinkedIn manually then they find the prospect then they open Salesforce check for duplicates leading to manually entering all the fields adding to sequence and then starting the outreach right and when you see it laid out visually the bottlenecks they kind of just scream at you. So steps two through six they're pure manual waste and that is what we're actually looking for.
So with that being said you just have to document everything. So, time spent, tools used, handoffs between people, where things get stuck, and this is going to become your road map for where AI can actually help. Really quick, if you're finding value in this breakdown, you just want access to our entire client acquisition guide with actionable steps for you to follow, join the school community. It's completely free. Links down below in the description. We share a lot of stuff that we use internally, so go ahead and check that out.
Something like follow up with leads. This sounds like one task. It's actually seven tasks hiding under just one label and only some of them are automatable. So the atomic breakdown looks like this. First up, open the CRM. Number two, filter for leads from the last 24 hours. Check if they match the ideal customer profile. Then read their submission details. Draft personalized response. Send email. Set follow-up reminder. And now you can actually evaluate each step individually. So can AI open a CRM? Yes. Can AI check ICP match against criteria? Yes. Can AI draft a response? Probably. It depends on how structured the inputs actually are. And this is the level of granularity that you actually need.
So in the past when we were working with an air hunting company, client has said to us that their problem was researching deceased property owners. It's a very niche industry uh very morbid. So something like this, it is too broad. Now the atomic breakdown, it revealed the real workflow. Number one, pull the county tax delinquency records. So this was about 300 to 500,000 properties. Number two, filter by property value thresholds. Three, filter by percentage of taxes owed. Filter by individual versus company ownership. Five, for each of the remaining 800 to a,000 leads, search ancestry databases. Six, search obituary records, search public records, cross reference multiple results for common names, verify identity matches, then the document findings, and then move to the next lead. And now we can see that steps 5 to 11 are where they're actually stuck. So 15 minutes per lead, mostly on steps 8 and 9 because, you know, John Adams returns 50 plus results and they have to figure out which one is actually the right one. And that specifically is what lets you identify that steps 5 to 11 can actually be automated with API access to databases that just allow businesses use. While steps one to four stay manual because you know the county systems don't have APIs. And the method for getting these atomic tasks it's pretty simple. On the calls with all of these teams we're just asking then what happens and then what happens. We're just repeatedly doing this until you can't break it down any further. And then we just document time per steps, all the tools required and the decisions being made.
Step number four is identifying what AI can do. Now, as you guys should already know, not every task is AI suitable. You need a filter. And we use four questions for every single atomic task. One is the input structured. So forms, emails, and database records, this equals yes. Vague verbal requests with missing context, this is going to be a no. Number two, is the output predictable? So any standard responses, classifications, data extraction, this equals a yes. If it's creative strategy or just novel solutions, then it's a no. Three, [snorts] are decisions rule-based? If then, logic scoring against criteria, this equals a yes. If any complex judgment calls requiring context, then it's a no. Lastly, is it repeated often enough? So if it's daily or weekly, then yes. If it's once a quarter, then no. Automation is not going to be worth it. And if you do get yes to all four, then you have a prime AI candidate right in front of you.
So, let me just show you guys how this actually played out on one of our past projects working with a property management company. So, the test was responding to tenant questions via email and text. So, obviously, is the infrastructure? Yes, tenants were sending specific questions through set channels. Next up, is the output predictable? So, what we had done is we had analyzed their email history. So 78% of questions, they fell into 12 different categories with standard answers. Rent payment procedures, maintenance requests, lease renewal, policy questions, this equals yes. Three, was it rule based? How do I pay rent? Well, this always just gets the same answer. When is my lease up? Well, this just pulls from the database, so this is inevitably going to be a yes. And lastly, is it repeated enough? So the property managers, they were all spending about 15 hours per week each on this. Absolutely, this is a yes. So this had passed all four of these filters. So we built an AI system that handles the 78% of routine questions all instantaneously escalating the 22% of complex issues to all the humans. And the result that we got from this 15 hours per week dropped to 3 hours per week. The property managers now were only handling issues requiring judgment like the lease violations, disputes, non-standard requests. And here's what we found that AI just can't do well is the conflict resolution between tenants or any strategic business decisions, highly unstructured requests where you're missing half of the context or tasks that happen so rarely that building automations just literally makes zero economic sense. And here's a good rule of thumb for this is AI is going to handle 80% of routine structured role-based work. Humans, they just handle 20% of edge cases just requiring any judgment. So, if you try to automate everything, you know, you will probably run into some issues.
Really quick, if you're a business owner looking for help implementing any of these frameworks and the AI solutions to increase your bottom line and grow your company, you can book in a free discovery call. Link is down below in the description. Let's get back into it.
Step number five, prioritizing for impact. So, by this point, you have about 15 to 20 automation opportunities identified. You can't build them all at once, and you really shouldn't. I plot everything on what I call the reprise impact matrix. And this is just a 2 x two grid. So the x- axis this is implementation difficulty. On the y-axis this is the business value. So what this does is it creates four different quadrants. We have the quick wins and these are just low difficulty highv value projects. And this is where you start. These prove return on investment fast and they fund the bigger projects. Big swings. These are high difficulty, high value. So any phase two or phase three after you've built trust. Low priority. These are low difficulty, low value. I probably would never build these. And then the money pits. These are just high difficulty, low value. I would avoid completely.
So, we're going to use an example of a property management project that we had actually finished up some months back. So, our quick win was lead auto collection. So, the problem, we're just going to use random names here. Marcus, he was spending 45 minutes every morning manually checking six lead sources. The solution, Zap, you're connecting all sources to a central air table. The difficulty, it was low. There was basic integrations, zero AI needed. The value though, $6,656 per year saved plus faster response time. The timeline, it took two weeks. Another quick win. Tenant FAQ automation. So, the problem was 30 hours per week across two property managers on just repetitive questions. The solution was an AI assistant that was trained on 12 common categories. The difficulty, it was, I would say, medium. It requires training on their specific policies, but the value was $42,000 saved per year. The timeline 3 weeks, very short, a quick win. We then had a big swing and this was predictive maintenance scheduling. So the problem here, reactive maintenance, inefficient vendor coordination. The solution we identified and what we built out was an AI analyzing the maintenance patterns, automatically scheduling based on vendor availability and their performance rating. The difficulty it was high so it did need historical data, complex integrations, vendor database, and of course some experience. The value $50,000 plus impact from fewer emergency repairs in optimized scheduling. And this is very conservative. But the timeline, this was 8 to 12 weeks. So notice that I'm not building the big swing first, obviously, even though that it has the highest total value. We're instead just building quick wins that prove ROI in weeks, not months. And once they actually see results, the big project, it literally sells themselves. And this is how you structure your projects. So phase one, quick wins. Phase two and three, these are going to be bigger projects. It makes the transformation just feel manageable instead of overwhelming.
Now, step number six, calculating real ROI. Now, this is what makes executives actually cut the check. You need to put dollar amounts on everything. So, the formula that we use, it's simple. It is just time wasted per person times number of people times days per year times loaded hourly cost and this equates to the annual cost of inefficiency.
Now let me show you exactly how this works on a project that we have done in the past. So we were working with a B2B software company and their STRs they were spending about 2 hours per day just building lead list manually. Eight SDRs total, $260 working days per year, $40 per hour loaded cost when you factor in the salary and then the benefits and the overhead. So 2 hours time 8 SDRs time 260 days * 40, this was $166,400 per year wasted on a task that should have been automated. Now suddenly a $60,000 automation, this does not sound expensive. It pays for itself in 5 months and it saves them $100,000 plus every single year after that.
But the cost savings is only half the equation. So you also need to quantify the lost revenue. Now for the property management company, here's how we broke it down. So the manual lead collection waste. So Marcus, he was spending 45 minutes per day. We times this by $30 per hour times 260 days. It's about $5,850 per year. Next up was another process. So they were averaging about three to four hours to response to leads. Now the industry data it was showing for every hour of delay you lose roughly 10% of leads. And at three and a half hours average they were losing approximately 30% of their leads. So 800 leads per month times 30% lost times 25% close rate that they had times $1,500 in average annual value. This was $90,000 per year in loss revenue. Another process was two property managers time 15 hours per week time $40 per hour time 52 weeks. This is about 62k per year. And now lastly, the manual maintenance coordination. So Tommy, another random name, he was spending about 45 minutes per request calling vendors, checking availability. 80 requests per month times 75 hours time $30 per hour time 12 months. This is about 21K per year. Now, if we add all this up, this is about $179,850 in combined waste in lost revenue annually.
Now, our solution, this had only costed $81,000 to implement. And 8 months later, here were the actual results. So, lead response time, it went from 3 to 4 hours to 18 seconds average. Conversion rate, this increased from 20 to 24%. This equates to about $161,000 in additional annual revenue that they were not capturing before. The tenant communication time reduced by 68% and this equated to 42k per year saved. And then the maintenance coordination this reduced by 58% also equating to $12,500 per year saved. So the total annual impact $215,500 in combined new revenue and savings. The return on investment $166% in year 1. The payback period it was 9.2 months. And this is the exact math that actually closes deals. So when you can show an executive that they are leaving $180,000 on the table annually and your solution is paying for itself in under a year, the decision, it is very obvious to them. They would have to be dumb to say no to you.
Now, if you're trying to do this for the first time, do not start with a 4-w week, $60,000 audit for a 30 person company. Well, I mean, it's highly unlikely of you even ever going to land that being in the position you're in right now. So, don't it doesn't even matter anyways. Now, if you're trying to do this for the first time, do not start with a 4-week super expensive audit for a 20 plus person company. Well, I mean, it's highly unlikely that you would even be able to land that with being in the position you're in, assuming you're a beginner. So, start with what I call the twoe opportunity assessment for smaller businesses, where week one you're doing discovery. So just a few stakeholder interviews, three to five people, 30 to 45 minutes each. Focus on pain points, repetitive tasks, decision bottlenecks. You're not trying to map the entire organization, just identifying the top problems. And from there, you can create a simple process flow diagram showing where work gets stuck. And you can document your preliminary findings, where is the waste, where is the lost revenue. Week two is solution design. So take the bottlenecks that you identified and plot them on your impact matrix. So what are the quick wins? Just validate this with the stakeholders. So something like we're thinking this automation here would solve X problem. Would that actually work given your current setup? Just get their buy in before you're finalizing anything. And then from there create your top three recommendations with a 90-day implementation road map. And then the pricing. So you can charge 5 to 10K for businesses with about 10 to 50 employees. And this is how you actually build your case studies and refine your process. You do not start at the top. It doesn't work that way. You start small, you prove the model, and you scale up as you get better at delivering the value.
And also, this is the difference between small businesses and mid-market companies. So, small businesses, you know, 10 to 50 employees or even anything below that. They want quick wins and immediate ROI. They want you to tell them, "Use this tool, implement this automation, you'll save 10 hours a week starting next month." mid-market companies, anything from like hundreds to 300 plus people. They want strategy, they want education, they want to understand the AI landscape as a whole. And they want to get their team aligned. Plan culture transformation as well. And the implementation, it isn't even their main concern yet. They just wants to know where the hell they are and where they should be going. So adjust your approach based on who you're selling to. It's very likely that you will have to focus on the former, at least for quite a while.
So that is our entire AI audit process. I mentioned earlier that if you stuck around, you will get access to some free goodies. So what that is going to be is our ROI calculator. So what this is, it's just a super brief form asking for some information about your company. And as a result, it estimates the gaps and the bottlenecks within your business. So the link is going to be down below in the description if you guys want to check that out. It is completely free. I think it'll help you guys a lot. And again, if you want hands-on help implementing these frameworks into your own business, whether you're building an AI agency or just implementing automation into your company or AI into your company in whatever sense, we offer one-on-one help where we walk you through everything. So, if you're also interested in that, link is going to be down below in the description. But with that being said, I hope you guys got some value from this. And if you did, please hit the subscribe button down below. Drop a comment. Let me know what you think. Also, like the video, really helps out the channel. and I'll see you guys in the next.