AI Agencies: You'll Never Hit $10k p/m Without This Simple Change

**Exploration Milestone: A Game Changer for AI Agencies**

As an AI agency owner, I've experienced firsthand the frustration of reviewing and revising chatbot systems that are overperforming but not meeting expectations. The process can be tedious and time-consuming, especially when dealing with low-ticket clients who don't understand the complexities of AI development.

I remember sitting on my Saturday afternoon, going through a client's revision notes for a project that was supposed to cost only $3,400. The chatbot system had performed as expected in the initial testing phase, but the client wanted more changes, and I was stuck in review hell. It was painful, and it made me realize that I needed to find a way to avoid this scenario.

That's when we introduced the Exploration Milestone concept, which has been a game changer for our business. The idea is simple: if the system is performing as expected, we're done. We can't keep pouring time and resources into making changes that aren't necessary.

We've taken this concept a step further by implementing a coign similarity check before starting any new project. This ensures that our developers are working on systems that are already optimized for performance, rather than trying to fix issues that don't exist.

The benefits of this approach are numerous. First and foremost, it saves us time. We're no longer spending hours reviewing and revising chatbot systems that aren't causing problems. Instead, we can focus on more complex projects that require our expertise.

Another advantage is that we're able to attract high-paying clients who understand the value of a well-designed AI system. These clients are willing to invest in quality, and they're less likely to waste our time with revisions or changes that aren't necessary.

So how does it work? We've implemented a filter at the early stages of the project development process. This ensures that only high-quality leads make it past the initial screening stage, reducing the risk of wasting time on clients who aren't serious about investing in AI solutions.

For general agencies like ourselves, who offer custom solutions every time, this concept is particularly valuable. We can use it to differentiate ourselves from competitors and attract higher-paying clients.

The Exploration Milestone has also helped us reach new revenue targets. In our case, we were struggling to break the $10,000 per month barrier. By implementing this approach, we've been able to increase our earnings significantly, thanks to more efficient project development and a focus on high-quality leads.

In conclusion, the Exploration Milestone concept has been a game changer for our AI agency. It's helped us save time, attract higher-paying clients, and reach new revenue targets. If you're an AI agency owner looking to optimize your business, I highly recommend exploring this approach further. With the right mindset and tools in place, you can avoid review hell and focus on delivering high-quality AI solutions that drive success for your business.

WEBVTTKind: captionsLanguage: enhey guys Le wle here and in this video I'll be breaking down the simple change that I made in my agency to allow us to scale past 10K per month and how you can do the exact same in yours it's relatively easy it's just a simple tweak within the sales process but it made a complete like a world of difference for us in terms of our closing rates and the amount of effort and and time we're putting into dealing with people who are never going to pay us any money anyway so this is something that I've I talk a lot about in my accelerator but I thought I'd bring it out here and do a sort of a a white boy session on YouTube so that you guys can get some of the some of the source as well cuz it's so and made such a massive difference for us so this is a a graph here showing that you keep getting stuck at the 10K per month Mark um a bunch of reasons why this can happen but uh I want to break things down a little bit more uh granular first so first things first let me illustrate this and how I came across it and how we made this Discovery at my agency with a bit of a story so back in middle of 2023 uh middle of middle of last year when the AI automation agency boom was going crazy I just started talking about it a few months earlier and suddenly everyone was talking about it and it was the hottest thing in online business at the time um and there's so much tension I can probably draw the the graph from the last video If you guys remember that um but the Google trends for AI automation agency had this massive Spike and then it sort of flattened out to where we kind of are now but at this point morning side in my agency we were getting about 20 20 leads per day inbound leads per day via our contact form on our website so we had so much attention and so many people who wanted to work with us um that our sales process and our sales pip plan really started to get choked up and it really pointed out the the main issue that we were facing which is a low close rate despite all of that attention right so we were getting these 20 leads 20 leads per day LPD but when you looked at our actual close rate and the amount of people we and businesses we were converting into paying clients like 20 per day that might result in like one client and now that's a shocking that's a shocking close rate and uh we we realized that pretty quickly we said well this is a really terrible conversion rate for the amount of attention we're getting um but we realized that we were doing a few things wrong in our in our sales process and there was a few things that we tested and and played around with to try to fix that and what I'm about to break down is is the one that we settled on so the the sales process that we had at the time was an email would come in from our uh our contact form submission so the email come in it would have information like um name um budget description of the project and then on that alone my business partner Josh would go in and just evaluate some of those emails and go yes no yes no delete them if they're not interested interesting for us um but then if there were kind of a good Prospect a good lead we would then send them a discovery call link so then that book oh can I do it oh and that book and a discovery call with us and then we'd have a 30 minute Discovery call with them 30 minutes and then from there me I was going to you realize how complex this thing was at the time um and from there then we' need to go to my my CTO Spencer he would have been on most of those Discovery calls and then he would need to analyze how we can how we can build that project right so he needs to go and and do a full analysis of okay this is what the client this is what the lead wants this is how we're going to approach it this this is the the opening stage of the project this is the middle this is the end this is the amount of hours that are going to be for each of them and then we can go back to them um we can do sort of a a calculator here let's do a bit of a calculator we needed to calculate how many hours it was going to take to do that and then we'd create a full proposal I'm changing colors here then we need to do a full Proposal with like all of the the js on it which is okay this is the this is how much this stage of the project is going to cost this is the scope this is the cost Etc this would take I don't know like three 3 plus hours I'm running out a pen here but this whole process here of getting on Discovery call 30 minutes taking that going to Spen and say hey fener can you come up with can you figure out how we'd approach this build um and determine if if how many hours it's going to take then that information would need to go to my business partner Josh and he'd create this from a template albeit but then he'd have to create a proposal saying this is how we're going to do it this is how much it cost and then we need to book in another call I'm getting pretty good at draun these phones now need to book in another call and that's going to be a proposal call where it might be 30 men to 1 hour and now look at all this time we're taking just to get uh get to the point of proposal then on that call they'll either say sort of yes or no or make a verbal agreement and if at the end of the call they say yeah yeah sure it sounds good we found that it was a massive drop off between the people who we actually pres presented proposals to and the number of people who actually ended up signing and and paying us an invoice right so this drop off between proposal and and actually uh how can I draw this make this a we skip a couple steps in here like sending the contract going back and forth on the contract you would go back and forth and back and forth um flopping around and then by the time you gone back and forth with a contract three or four times then they're like suddenly not as interested in the in the project anymore they've lost the buzz uh you really need to get as quickly from this point of saying hey I'm interested I want to work with you and and getting them signed as fast as possible because in here Anything Could Happen they could something in their business could blow up Suddenly they're not interested in AI anymore they read an article and they're scared of it all sorts of crap so the longer this was the the higher the drop off we saw here but really the key drop off point was this proposal here cuz they go from someone who's filled out a form don't have really any idea how much this is going to cost them we go through all this time and we invest all this time here and calculating calling writing the proposal then we get on the call and present it and a lot of the times like this stuff a stuff isn't cheap if you have a development team like me and our developers cost quite a lot of money per hour we have to mark our services up significantly it's it's like very expensive projects a lot of the time and when you then present that to them they are are nowhere near ready for that kind of price point well in in most cases so it's really an issue of lead quality and we weren't checking or or sort of price testing them beforehand and we're wasting all this time in getting to this point at the end where we've we've pitched them and then they look at it and they go oh W like I'm no chance no chance I'm paying that and they go that's way too expensive for what I'm looking for thank you have a nice day and so we got this big drop off where you might get I don't know you make six proposals and get like one of these guys coming out the other end and they're actually interested in signing for the for the project so looking at this we realized we are spending so much time on this step here of hopp hopping on calls with basically everyone and then allowing these Tire kickers let me draw a tire tire kickers are all over the AI space they business owners who are interested these early adopters if you remember the tech adoption curve were in this early adopter phase watch that video which I'll put up here if you haven't already this is a very important concept to understand we are getting a lot of early adopters especially earlier on in my journey last year in the first sort of six months of starting my agency we were getting a lot of these early adopters who didn't really care much about the the functionality of it they just wanted to feel like they're building something cool and and being on the front foot with their business so these guys do tend to be tie kickers they because they aren't seeking some really really business changing implementation the the rate of them being TI kickers and just kind of wanting to to hop on calls and learn more and feel like they're in the space is uh is much higher so tie kickers they're just coming in and we're spending all of this time 30 minutes and then Spencer is having to do uh I don't know maybe like one to two hours here and then Josh has to put 3 hours in here and then we have to put Spencer and Josh on this call for a 30 minute to an hour call so much time just to find out are they or are they not a tie kicker and so what we did to change this is I mean it it was kind of like a Hail Mary we were just saying look these close rates are so awful there has something has to change so let's just like it let's just let's just throw a low ticket offer at them to get them started and and see who gets through so what we did is implemented what we now call an exploration Milestone and this this may be standard for uh development companies all over the show but um until about a year ago I wasn't some development company Company CEO as many of you guys won't be so we were definitely learning this as as people have just been online business entrepreneurs and having to pick up how to run these development companies um there's there's Nuance to to the the sales process as you can imagine so we tried this thing we'll call it an exploration Milestone and if I rewrite this this flow uh the email still came in uh we still approved or disproved it and then disapproved it declined uh and if it was good we we probably have a thing here where it's like eh we don't want to work with them we book in the discovery call and that's going to still be 30 minutes but then at the end of the discovery call if we've qualified them and vetted them out properly and they they seem like a good client there's a like when we get we get so many leads coming to us that we can be quite picky about who we work with because there's nothing worse than I mean I built my business to have a an income stream but also to enjoy what I do and if I'm dealing with clients who are at Absolute pain in the ass that is working against what my goal was with my business right I I don't want this to feel like a trap and my business partner in the whole team don't want to work with awful people it's so bad it's it's like multiple levels of of of of Badness that there's multiple levels of of Destruction and damage within the business because not only you are dealing with this this bad client but everyone in the team is it just brings it's like an anchor that that brings down the whole team um so we get very picky here um and at the end of the discovery call if they have a clear solution that we can solve if they are someone who we want to work with and overall seems like a good potential client for us then we will at the end of the call we will transition it transition it that's a transition to at the end of the call I'll go to my business partner Josh so hey Josh yeah everything checks out for me here I think it's good for us to just jump straight into an exploration Milestone uh and kick things off next week boom now we've gone from just talking about hey AI stuff is cool yeah tie kick of this tie kick of that we're starting to put the heat on them and say say hey look we're looking we need some money like you've got your free Discovery call we've had a chat with you and now you got to cough up to keep working with us now this does come down to some degree to the brand that that you've built for me and morning side we have a a considerable brand position as a AI Solutions company because of the stuff that I do here wav my hands around on the camera because it shows that I know what I'm talking about right and you want to work with people who know what they're talking about which is why you should consider making content either YouTube LinkedIn whatever it is I can't even explain I mean we're getting 20 of these a day we still get quite a lot of them and I don't even post nearly as much as I did back in the day so getting a little bit off track but then we say explow exploration Milestone that's what we're looking to do we end the call and straight after the call well we get them to agree to it and we say yeah well if if you're happy we can we can start what we call an exploration Milestone they might ask question say well that's when we just uh go in and we do a very quick sort of prototyping scoping specking determining how much this is going to cost for you we can't give you an accurate quote on the whole project until we've gone in and done some exploration and done some tests because we need to actually cuz the clients will want certain kinds of outputs from the system and because we're going to get into this in a second which is a really cool part of this video so stick around um but clients want specific outputs from their systems as as anyone would and if we aren't able to get those kind of outputs then we shouldn't take on the project right so exploration is kind of saving your ass and also making sure the client and expectations are managed correctly because you get a chance to go okay let's very quickly spread up a prototype for okay pass me your data Mr client and let me give me some examples of input and output peers and and responses that you want from it okay I'm going to take those in give my to my D team and they can who it up quickly sort of determine if we can get kind of close to the outputs they're looking for if we can great yep that system works to some degree now we're good to take on the full project and we can be confident that we can charge you x amount of dollars and it's going to come within this time frame so it's really such a key thing and mass like an absolute game changer for our business um so how do we go from here exploration milestone we send that via email so we'll send a templated contract we'll send an email just saying hey just following up on uh on on the C which is had um the exploration Milestone is going to cost x amount and we started off with ours at uhu $800 and that would be sort of a 3 to 5 day exploration and I usually more it usually ended up sort of closer to five than three um but the $800 exploration was a good price point that we found people would be interested in and and they took the bait they took the offer pretty often so right here you're asking them to cough up and you're saying sorry buddy no more Tire kickers Beyond here you are not going to get a single second more of my time until you've given me money to compensate me and my team which allowed us to then only make presentations to clients who are actually qualified to some degree financially qualified um so I bring in the price uh the time that they had to cough up some money from way down here to up here we just saved ourselves all this time and that that was a absolute game changer for us our business partner Josh instead of working on a being on the sales call hamster wheel and proposals we' go through these cycles of oh yeah cool this is Big C in the pipel and then next week yeah they didn't reply to the proposal so it happens to it happens to me it'll happen to you but you can do things with your sales process to try to split split the W the wheat from the chaff a little bit sooner so you want the you want the good guys to make it past year you want to give them a chance to show that they are serious and this this amount is a is a good way of doing it now for full transparency we now charge most recent one we did was $2,000 um we've actually charged right up to $5,000 now that may sound ludicrous but when we do $100,000 deals a $5,000 exploration uh in order to scope it all out there's a lot of work invol for us but the client would rather pay 5 grand up front and determine hey is this feasible uh what kind of what's the actual budget going to be at the end of the day and and are these kind of systems actually something that are a good match for my use case so you can charge up to $2,000 $5,000 for these Explorations it take these these take sort of 2 to 3 weeks so uh just being fully transparent so it's not like it was a 3day and 3day these are definitely big boys for bigger projects and then exploration Milestone you send that off bya email say Hey look it's going to cost2 Grand or in your case might be $800 to start it's going to cost $800 let me know if that's all good and I can send over my we can send over a templated contract you just change a couple little things uh and once I've signed that you send the send the invoice with the contract as well once it's all paid then you go sweep straight into slack and then you go hey send requirements and that's when you ask them okay in order to do our our exploration actually I missed the best part of this whole thing let me go back a step this is when it gets real Saucy so not only does this exploration give us a chance to weed out the the guys from uh the split the wheat from the CHF we also get an opportunity to set expectations with our clients and AI systems are inherently random and hard to predict they are non-deterministic by nature which means that the same input can create different outputs and this is due to the sort of randomization of the next token generation we are trying to give in in many cases reliable and predictable outcomes with technology that on the most fun most fundamental like foundational level is random and and tries to be random so in order to solve in order to get client satisfaction and set expectations right we have also had to deal with let me let me give you an example actually this is a bit way to do it this may sound familiar so client uh client uh say You're Building some kind of chatbase system for them client uh doesn't like output say that you're building a system a chat bot for them and they say oh I I don't like the way that it says that and you say okay well they provide ideal output and you go okay well I'll I'll tune the system to try get it to outut output that kind of output and then it outputs the correct one maybe four out of five times client gets mad because it's not always consistent now this this four out of five times the other time the one out of five times that it doesn't say it it might just slightly reward it but this client is looking for exact outputs and you're trying to he wants exact outputs for you to get the money from them and finish the contract you need to give them this five out of five perfect exact same outputs every time when in reality the system is putting out one and five times something that is a slight modification of it's still saying the same thing it still means the same but it's a slightly different wording and this client goes eh client gets mad because it's not consistent um demands refund because I don't like like the way it talks so here you have a feeling I don't like the the way that the client feels about the outputs is now determining whether or not you get your money which is a a bad place to be so we have tried to solve this problem in a number of ways and eventually in combination with the exploration Milestone we came up with a method that solves all of this while also allowing us to De exploration and solving Downstream issues with client satisfaction as well so I'll now get into that cuz I know you came for the source but I'll uh I'll give you double serving cuz you know that's how we work on this channel so the issue here is as I just pointed out llms are you're dealing with two problems one llms llm uh non deterministic they are by Nature random and they they split out different outputs from the same inputs and then two you're also fighting against client feelings and subjective opinions now these two when you're trying to build a business selling AI Solutions these two go together like Mentos and Coca-Cola um they don't they don't get along uh and this is a an awful mixture of things to build a business on and we at morning side realize this pretty quick that okay well clients are going to have feelings towards if we're building any kind of chat based or or even just it's it's a Content generator how they feel about like h i don't it's not really what I like there's something about this and some sometimes would get to a point where a client is yeah this system's cool yeah yeah it's good it's good it's good it's good good and then the next week they'll say oh no we need a big change I I just suddenly don't like this anymore and you're dealing with emotions uh emotions and opinions and you're trying to make money and have some kind of consistent and predictable business uh but you're also now now at at the at the whm of these client emotions and also these inherently random systems so these two problems we realize pretty quick that if we don't solve these at morning side we're going to be we're going to be in the and we're not going to have much of a business so how did we solve this well my uh my absolute wiiz and Genius of a CTO uh we sort of tasked them with this and said hey we need a solution to this and he came up with a now widely used system um can I rub this off no I still I still have a bit more to go uh that we call it's not we didn't it's been a thing but the way that we apply it I believe is a sort of a industry first it's called coign similarity similarity so what cosine similarity does is it's a test that we can run that allows us to determine the similarity this the semantic similarity of two sets or or any number of set of of uh actually I believe it is just for two uh the the the similarity of two pieces of text so whether that's a word just apples and oranges whether that's phrases whether that's whole bodies of paragraphs whether it's a whole book uh you can use coine similarity to get a a t tangible metric is really the key thing here a tangible metric that you can point to and say yes these two things are similar and for us with these exploration milestones we started integrating coine similarity testing as part of the exploration Milestone determine if we can get the the the right mathematically most similar types of answers that are going to be satisfactory to the client so we can kind of make sure that we know what we're doing ahead of time and be like yep we can get a we can get pretty close to this uh but more importantly it gives us a a metric that we can use in our contracts to to stop that subjectivity and oh I like the I like it but like it could say this and this or hey it's giving me the same answer four times but there's one that's different it's like well mate at the end of the day our contract says it's the semantic meaning of it and the similarity of the text if there's if it's one word changed this thing protects us so let me let me break down how this works so coine similarity uh this is going to get a little bit technical but I do quite like uh getting in this and it's important for you guys to know this to some degree so Cent similarity this is going to really challenge my uh my art skills or um down okay so that's a three-dimensional space and we're going to have so it's kind of going backwards by the way it's not facing us so we have one point over here we have another point over here and we have another point uh up over here and so what these are is coordinates in a multi-dimensional space in this case we'll just do three dimensions so that it's easy to understand but in reality this is happening in in I think 8,000 what was it 8,000 Dimensions it's impossible for the human brain to comprehend but uh this will this this will suffice um so we have this 3D space we have .1 2.3 and what coine similarity allows us to do is determine the similarity of these three points so this might be a word of of beans this might be P's and this might be car and so what the co similarity test does is that if we go to our client in this exploration phase or in the the actual contract we will ask them for hey can you give me uh 5 to 10 examples of inputs and outputs that you would like out of the system and when they give us the inputs and output we then have something to grade our AI outputs of the system against so say they have a chat bot and it needs to be outputting a certain type of answer we can run their 11 inputs against our system so we have our system here we take the 10 inputs we run them each through our system and it outputs 1 2 3 Etc and then we have the original 10 inputs and the well this is the system that they used to create which is their own brain of okay what what should this thing be answering so this is a human this is an AI and then it's going to Output the well it's not really outputting the brain has outputed it but we have 10 and 10 and the 10 that we create from our system we can then use coine similarity to get the average or we can we can compare the similarity of this one to this one and determine if yes our AI system is creating outputs that are similar enough to this we add Aver is M and then we can get an overall similarity score so coming back to the graph here so we have this one here which is is one input might be uh how do I change my password and then this one will be the AI output one and this one will be the human one they might say in order to change your password you need to go to XYZ website and then based off the AI system that we built it might say in order to change your password please navigate to the S so they are fairly similar and then when we run them through the coine similarity test it's conver ing these words and it runs it through an embedding model and the words get converted to a a set of coordinates that might be like 0.1 0.5 all 0.9 uh and it goes on and on and on and on and on but in this case we've just got three coordinates because we've got a threedimensional graph so what the cosine system does is it converts words into their numerical equivalence within a vector space and by having these coordinates it can plot them in the vector Space by their similarity so in this case you see beans and peas are are close that's because the embedding model that we use and run the beans through it's going to give out a coordinate and then we run P's through it it's going to go H okay that fits there and they end up being close to each other within the vector space and so how does what does coine similar similarity actually mean well it's the measurement of the angle between these two bad boys here so the smaller the angle the more similar they are the larger the angle see if we go from P's to car or from Beans to car the angle here is much larger oh I'm getting back into my my uh what was that calculous in trigonometry and whatnot um but yeah beans the car is a much larger angle therefore the coign similarity is going to be lower than B's 2 PS that makes sense so this is taking the subjective opinions and emotions of people who go oh but I like the outputs I don't really like the outputs I don't care you have told us these in the in the exploration Milestone we won't necessarily hold them to that in the contract of the exploration Milestone it would just be hey look give us some outputs and uh we'll compare them to it and internally we can check and use that in the end of the report so at the end of the exploration Milestone you'll give them a report and a proposal which we'll go into in a second but the numbers that we get back from our testing if we we say okay look by the way cosine similarity goes from0 to 1 um so high a good range uh we go for in our expirations we might be looking for 0.85 or or uh or higher but for our final builds we'll be going for anywhere from 0.9 to 0.95 in most cases so that means it's it's highly similar and when we do our expiration Milestone we'll be looking to see if we can get this sort of number and say okay give me those 10 out 10 outputs and 10 inputs we're going to run it through the system see what we get okay yep we're around 8. 0.85 or higher we're confident that we can build the system for you so it's as much of client satisfaction as it is for our sanity we don't want to be taken on trying to take down a mammoth with a slingshot you know like we need to be taking on clients that we know that we can deliver for because I tell you what These Guys these these early adopters have huge expectations um it's actually it's more the people who are the sort of early majority who have the bigger expectations cuz they like oh well I've heard all the stuff about AI let me give it a try and they expect that to just completely change your business so you need to make sure that they're not expecting too much and given the data and the kind of inputs and outputs they have provided we are able to get close enough to it that we can feel confident in going for a a complete contract a full contract on it so that's C on similarity you can convert the 10 inputs or five to 10 inputs that the client gives you into uh or the output sorry into a set of coordinates or or numbers that a computer can understand and when we convert these over we can plot them um automatically we have a tool for this that we give to our accelerator members so that they can do this testing um for their clients we have an internal one as well that we use at morning side and you just run these run these outputs through the co similarity system it's going to determine the similarity of them and then we'll average it out and say okay what was the average number so that's cosine similarity might sound a bit complex but we are dealing with complex issues and complex technology so uh and complex emotions of clients too H and so how does this tie back into the exploration system and how it changed our our sales process and sort of revolutionize our business and how can how can you you use it too really so now they've paid us we've done the exploration Milestone and the uh we then need to present the the findings to them so once you've done all this work and you've you've created your uh you've given this to your divs they've gone and that's a that's a that's a the coding symbol uh they've gone and done the work they've done the exploration we've got the cosine coine similarity scores out pretty confident we've done enough of the exploration that we can get to the point where we're confident in pitching uh for the full project now uh this comes out and we now have a proposal that we put together which is let's just go half and half here the first half is going to be the findings the findings of the exploration which is hey look we did these tests and these are the kind of outputs we're really happy with this we're kind of concerned about this so this first half is the findings and then in the second half you're then going to be pitching your proposal and in this proposal we need to have that's when you get your your developer either you do it yourself or or the the lead developer on your team then you need to get them to create that proposal which is what Spencer was doing earlier in the in the older system which is okay what are the sections how many hours three four Etc how many hours is each of these things going to take and that's going to determine how we can price it within this proposal so you just we typically just mark up we get the our estimates from our CTO Spencer and then we just mark up based off our kind of inflating markup rate that we apply to our our services um so whether that's a few hundred now whether that's you're get to thousands of dollars uh you can just kind of Mark that up and and apply that overall and that proposal you will also include gosh makes me realize how much how much there is to cover in this stuff uh I thought this was going to be a quick video but inside the proposal then you have the cosine similarity score that you agree on so this will become part of the contract and that is what's going to save you from the subjective opinions of clients who are unhappy with h i don't really like how it sounds that kind of rubbish first half the findings of the discovery and here as we get to there's so much for me to talk about on this I probably can't squeeze all into this video but there's a number of different factors as to why this works and why it's so important uh one of them is that you get a chance to build trust you get to show them how you operate and this right here is the most important part because they've just paid you $800 it's basically like a like they're just sort of having a dig and saying look these guys seem professional let's see if they really know this stuff and they've given you a small amount of money and you've gone a way that should pay your developers and hopefully give you a bit of extra so youve again you're getting cash flow here which is great even if you don't end up taking the project on you can just get paid to do Explorations for these people which cash flow early on in your agency that that kept us alive I sh you not but then you come down here and you get a chance to show them who you are show them how you operate show them your level of professionalism and the findings and the way that you do this presentation is going to be key so this is actually done on a call oh let me see if I can get this again this is done on on what we call our proposal call so we'll get them on a call 30 minutes to half an hour we'll go through the findings of it very professional hey this is what we did we're concerned about this this is the kind outcomes we've got um this are the co similarity scores and we you can kind of introduce this as the concept on these Calles we can say hey look given the the inputs and outputs that you gave us we were able to get to a 0.85 similarity which means that it's very good um and and with a bit more bit more funding and a bit more effort we can get this to a point where it's it's far more similar it's up sort of the 0.9 0.95 range um so you get to give them the findings and then you give them the proposal based off the estimations that your your lead developer or yourself have come up with and that goes in the second half which is a proposal breaking down each of the parts of the full build and how much that's going to cost them on the call you need to have the contract ready and in the contract um if they're all good to go and there's no complaints then you send them over the contract get it signed on the call if possible because as soon as that call ends your chance of closing them just just goes through the floor um try to get them to to sign that contract on the call because as soon as they signed it doesn't matter about the invoice really if they've signed that they've made a a commitment to paying that invoice as well so as long as you can get it signed walk into the contract handle objections whatever you need to do point out that the cosine similarity metric is in there and say hey look because these are kind of subjective output uh this is something that we do to both Safeguard ourselves and to make sure that you are protected and that we have a level of of standard uh a bar to work towards and the quality of outputs that this system generates so having this C similarity pitch it as a benefit for you and them which it is and say hey look this is the this is the agreed one we're going to be aiming for 0.9 anything more than that might be overfitting CU you can make the system too good at giving certain answers um you want to keep it around that 0.9 to 0.95 range um and that's what we'll put in our contract and that is the way that we can stop this if you've ever done these kind of projects before where it's Custom Custom builds uh you'll know that it can just kind of go on and on and on and it's like review revision hell we call it review hell and it just goes on and on and on and it's so painful when you have a chatbot system that you've created and then you send it off to them and they send back like a 30 40 minute Loom and you just especially when it's a low ticket client as well if you've been paid three or $4,000 and you're sitting there on your Saturday afternoon going through this guy's revision notes and then having it go back and fix the system up and man it is not pretty and it's not the kind of lifestyle that I want for my business and so we are try to avoid as much of this crap as possible by saying no we have the coign similarity this thing is performing as expected we're done send me the money send me the money so yeah hope I uh got all that across but the exploration Milestone has been a complete game changer for us and if you want to incorporate something similar into your business while well you may not do the exact same thing as a as an AI agency owner what you do need to think about is this line and how do I spend less time talking to tie kickers who are going to waste my time and how can I push quickly to start filtering out cuz you guys may do some Mund Le generation you may do a LinkedIn whatever and you may be having conversations but you'll pretty quickly realize that a lot of these are just conversations and you need to start finding the the guys with money to spend these are the guys you want to be talking to and the longer you give them like there's billions of people on the Earth and don't assume that all of them are like you and I and now rational and and don't want to waste time some of them are deluded some of them have mental issues and they will happily sit on these calls and waste your time and that's not going to help your business you're going to end up spending so much time talking to people are never going to be clients while it may be good sales practice if you need that but talking to a looney is maybe not the best kind of sales practice you you should be after but yeah take this concept whether you apply it bar for bar or whether you just take the kind of distill it down and say hey look I need to put some kind of filter earlier on whether that's uh it's early on in my when I was doing calls for my Consulting calls on my channel I would have a $300 or $500 Consulting call and that those are the people that would then take on as Dev clients so if they came through Consulting calls they've already kind of pre-qualified or pre-selected themselves saying hey look I'm willing to spend $500 on this so that was kind of our earlier exploration Milestone but we started getting leads directly through our website not just through my uh through my channel and through the through the Consulting call link so yeah take it apply it and this is particularly good for for General agencies like myself who are doing custom Solutions every time if you are a niche agency um and you have the same thing to sell every single time may be a little bit different but I think this is Handy to have in your back pocket either way and it will help you get past $10,000 a month which was a a major issue that we had and the rest is history so hope this been helpful guys um I certainly enjoyed making it and yeah exploration milestones and Co similarity I hope you got it all uh and I will see you in the next one keep going uh you can do it this opportunity is not going anywhere but you need to make back you need to make moves on it now um if you want to get there so I'm rooting for you and keep going ciaohey guys Le wle here and in this video I'll be breaking down the simple change that I made in my agency to allow us to scale past 10K per month and how you can do the exact same in yours it's relatively easy it's just a simple tweak within the sales process but it made a complete like a world of difference for us in terms of our closing rates and the amount of effort and and time we're putting into dealing with people who are never going to pay us any money anyway so this is something that I've I talk a lot about in my accelerator but I thought I'd bring it out here and do a sort of a a white boy session on YouTube so that you guys can get some of the some of the source as well cuz it's so and made such a massive difference for us so this is a a graph here showing that you keep getting stuck at the 10K per month Mark um a bunch of reasons why this can happen but uh I want to break things down a little bit more uh granular first so first things first let me illustrate this and how I came across it and how we made this Discovery at my agency with a bit of a story so back in middle of 2023 uh middle of middle of last year when the AI automation agency boom was going crazy I just started talking about it a few months earlier and suddenly everyone was talking about it and it was the hottest thing in online business at the time um and there's so much tension I can probably draw the the graph from the last video If you guys remember that um but the Google trends for AI automation agency had this massive Spike and then it sort of flattened out to where we kind of are now but at this point morning side in my agency we were getting about 20 20 leads per day inbound leads per day via our contact form on our website so we had so much attention and so many people who wanted to work with us um that our sales process and our sales pip plan really started to get choked up and it really pointed out the the main issue that we were facing which is a low close rate despite all of that attention right so we were getting these 20 leads 20 leads per day LPD but when you looked at our actual close rate and the amount of people we and businesses we were converting into paying clients like 20 per day that might result in like one client and now that's a shocking that's a shocking close rate and uh we we realized that pretty quickly we said well this is a really terrible conversion rate for the amount of attention we're getting um but we realized that we were doing a few things wrong in our in our sales process and there was a few things that we tested and and played around with to try to fix that and what I'm about to break down is is the one that we settled on so the the sales process that we had at the time was an email would come in from our uh our contact form submission so the email come in it would have information like um name um budget description of the project and then on that alone my business partner Josh would go in and just evaluate some of those emails and go yes no yes no delete them if they're not interested interesting for us um but then if there were kind of a good Prospect a good lead we would then send them a discovery call link so then that book oh can I do it oh and that book and a discovery call with us and then we'd have a 30 minute Discovery call with them 30 minutes and then from there me I was going to you realize how complex this thing was at the time um and from there then we' need to go to my my CTO Spencer he would have been on most of those Discovery calls and then he would need to analyze how we can how we can build that project right so he needs to go and and do a full analysis of okay this is what the client this is what the lead wants this is how we're going to approach it this this is the the opening stage of the project this is the middle this is the end this is the amount of hours that are going to be for each of them and then we can go back to them um we can do sort of a a calculator here let's do a bit of a calculator we needed to calculate how many hours it was going to take to do that and then we'd create a full proposal I'm changing colors here then we need to do a full Proposal with like all of the the js on it which is okay this is the this is how much this stage of the project is going to cost this is the scope this is the cost Etc this would take I don't know like three 3 plus hours I'm running out a pen here but this whole process here of getting on Discovery call 30 minutes taking that going to Spen and say hey fener can you come up with can you figure out how we'd approach this build um and determine if if how many hours it's going to take then that information would need to go to my business partner Josh and he'd create this from a template albeit but then he'd have to create a proposal saying this is how we're going to do it this is how much it cost and then we need to book in another call I'm getting pretty good at draun these phones now need to book in another call and that's going to be a proposal call where it might be 30 men to 1 hour and now look at all this time we're taking just to get uh get to the point of proposal then on that call they'll either say sort of yes or no or make a verbal agreement and if at the end of the call they say yeah yeah sure it sounds good we found that it was a massive drop off between the people who we actually pres presented proposals to and the number of people who actually ended up signing and and paying us an invoice right so this drop off between proposal and and actually uh how can I draw this make this a we skip a couple steps in here like sending the contract going back and forth on the contract you would go back and forth and back and forth um flopping around and then by the time you gone back and forth with a contract three or four times then they're like suddenly not as interested in the in the project anymore they've lost the buzz uh you really need to get as quickly from this point of saying hey I'm interested I want to work with you and and getting them signed as fast as possible because in here Anything Could Happen they could something in their business could blow up Suddenly they're not interested in AI anymore they read an article and they're scared of it all sorts of crap so the longer this was the the higher the drop off we saw here but really the key drop off point was this proposal here cuz they go from someone who's filled out a form don't have really any idea how much this is going to cost them we go through all this time and we invest all this time here and calculating calling writing the proposal then we get on the call and present it and a lot of the times like this stuff a stuff isn't cheap if you have a development team like me and our developers cost quite a lot of money per hour we have to mark our services up significantly it's it's like very expensive projects a lot of the time and when you then present that to them they are are nowhere near ready for that kind of price point well in in most cases so it's really an issue of lead quality and we weren't checking or or sort of price testing them beforehand and we're wasting all this time in getting to this point at the end where we've we've pitched them and then they look at it and they go oh W like I'm no chance no chance I'm paying that and they go that's way too expensive for what I'm looking for thank you have a nice day and so we got this big drop off where you might get I don't know you make six proposals and get like one of these guys coming out the other end and they're actually interested in signing for the for the project so looking at this we realized we are spending so much time on this step here of hopp hopping on calls with basically everyone and then allowing these Tire kickers let me draw a tire tire kickers are all over the AI space they business owners who are interested these early adopters if you remember the tech adoption curve were in this early adopter phase watch that video which I'll put up here if you haven't already this is a very important concept to understand we are getting a lot of early adopters especially earlier on in my journey last year in the first sort of six months of starting my agency we were getting a lot of these early adopters who didn't really care much about the the functionality of it they just wanted to feel like they're building something cool and and being on the front foot with their business so these guys do tend to be tie kickers they because they aren't seeking some really really business changing implementation the the rate of them being TI kickers and just kind of wanting to to hop on calls and learn more and feel like they're in the space is uh is much higher so tie kickers they're just coming in and we're spending all of this time 30 minutes and then Spencer is having to do uh I don't know maybe like one to two hours here and then Josh has to put 3 hours in here and then we have to put Spencer and Josh on this call for a 30 minute to an hour call so much time just to find out are they or are they not a tie kicker and so what we did to change this is I mean it it was kind of like a Hail Mary we were just saying look these close rates are so awful there has something has to change so let's just like it let's just let's just throw a low ticket offer at them to get them started and and see who gets through so what we did is implemented what we now call an exploration Milestone and this this may be standard for uh development companies all over the show but um until about a year ago I wasn't some development company Company CEO as many of you guys won't be so we were definitely learning this as as people have just been online business entrepreneurs and having to pick up how to run these development companies um there's there's Nuance to to the the sales process as you can imagine so we tried this thing we'll call it an exploration Milestone and if I rewrite this this flow uh the email still came in uh we still approved or disproved it and then disapproved it declined uh and if it was good we we probably have a thing here where it's like eh we don't want to work with them we book in the discovery call and that's going to still be 30 minutes but then at the end of the discovery call if we've qualified them and vetted them out properly and they they seem like a good client there's a like when we get we get so many leads coming to us that we can be quite picky about who we work with because there's nothing worse than I mean I built my business to have a an income stream but also to enjoy what I do and if I'm dealing with clients who are at Absolute pain in the ass that is working against what my goal was with my business right I I don't want this to feel like a trap and my business partner in the whole team don't want to work with awful people it's so bad it's it's like multiple levels of of of of Badness that there's multiple levels of of Destruction and damage within the business because not only you are dealing with this this bad client but everyone in the team is it just brings it's like an anchor that that brings down the whole team um so we get very picky here um and at the end of the discovery call if they have a clear solution that we can solve if they are someone who we want to work with and overall seems like a good potential client for us then we will at the end of the call we will transition it transition it that's a transition to at the end of the call I'll go to my business partner Josh so hey Josh yeah everything checks out for me here I think it's good for us to just jump straight into an exploration Milestone uh and kick things off next week boom now we've gone from just talking about hey AI stuff is cool yeah tie kick of this tie kick of that we're starting to put the heat on them and say say hey look we're looking we need some money like you've got your free Discovery call we've had a chat with you and now you got to cough up to keep working with us now this does come down to some degree to the brand that that you've built for me and morning side we have a a considerable brand position as a AI Solutions company because of the stuff that I do here wav my hands around on the camera because it shows that I know what I'm talking about right and you want to work with people who know what they're talking about which is why you should consider making content either YouTube LinkedIn whatever it is I can't even explain I mean we're getting 20 of these a day we still get quite a lot of them and I don't even post nearly as much as I did back in the day so getting a little bit off track but then we say explow exploration Milestone that's what we're looking to do we end the call and straight after the call well we get them to agree to it and we say yeah well if if you're happy we can we can start what we call an exploration Milestone they might ask question say well that's when we just uh go in and we do a very quick sort of prototyping scoping specking determining how much this is going to cost for you we can't give you an accurate quote on the whole project until we've gone in and done some exploration and done some tests because we need to actually cuz the clients will want certain kinds of outputs from the system and because we're going to get into this in a second which is a really cool part of this video so stick around um but clients want specific outputs from their systems as as anyone would and if we aren't able to get those kind of outputs then we shouldn't take on the project right so exploration is kind of saving your ass and also making sure the client and expectations are managed correctly because you get a chance to go okay let's very quickly spread up a prototype for okay pass me your data Mr client and let me give me some examples of input and output peers and and responses that you want from it okay I'm going to take those in give my to my D team and they can who it up quickly sort of determine if we can get kind of close to the outputs they're looking for if we can great yep that system works to some degree now we're good to take on the full project and we can be confident that we can charge you x amount of dollars and it's going to come within this time frame so it's really such a key thing and mass like an absolute game changer for our business um so how do we go from here exploration milestone we send that via email so we'll send a templated contract we'll send an email just saying hey just following up on uh on on the C which is had um the exploration Milestone is going to cost x amount and we started off with ours at uhu $800 and that would be sort of a 3 to 5 day exploration and I usually more it usually ended up sort of closer to five than three um but the $800 exploration was a good price point that we found people would be interested in and and they took the bait they took the offer pretty often so right here you're asking them to cough up and you're saying sorry buddy no more Tire kickers Beyond here you are not going to get a single second more of my time until you've given me money to compensate me and my team which allowed us to then only make presentations to clients who are actually qualified to some degree financially qualified um so I bring in the price uh the time that they had to cough up some money from way down here to up here we just saved ourselves all this time and that that was a absolute game changer for us our business partner Josh instead of working on a being on the sales call hamster wheel and proposals we' go through these cycles of oh yeah cool this is Big C in the pipel and then next week yeah they didn't reply to the proposal so it happens to it happens to me it'll happen to you but you can do things with your sales process to try to split split the W the wheat from the chaff a little bit sooner so you want the you want the good guys to make it past year you want to give them a chance to show that they are serious and this this amount is a is a good way of doing it now for full transparency we now charge most recent one we did was $2,000 um we've actually charged right up to $5,000 now that may sound ludicrous but when we do $100,000 deals a $5,000 exploration uh in order to scope it all out there's a lot of work invol for us but the client would rather pay 5 grand up front and determine hey is this feasible uh what kind of what's the actual budget going to be at the end of the day and and are these kind of systems actually something that are a good match for my use case so you can charge up to $2,000 $5,000 for these Explorations it take these these take sort of 2 to 3 weeks so uh just being fully transparent so it's not like it was a 3day and 3day these are definitely big boys for bigger projects and then exploration Milestone you send that off bya email say Hey look it's going to cost2 Grand or in your case might be $800 to start it's going to cost $800 let me know if that's all good and I can send over my we can send over a templated contract you just change a couple little things uh and once I've signed that you send the send the invoice with the contract as well once it's all paid then you go sweep straight into slack and then you go hey send requirements and that's when you ask them okay in order to do our our exploration actually I missed the best part of this whole thing let me go back a step this is when it gets real Saucy so not only does this exploration give us a chance to weed out the the guys from uh the split the wheat from the CHF we also get an opportunity to set expectations with our clients and AI systems are inherently random and hard to predict they are non-deterministic by nature which means that the same input can create different outputs and this is due to the sort of randomization of the next token generation we are trying to give in in many cases reliable and predictable outcomes with technology that on the most fun most fundamental like foundational level is random and and tries to be random so in order to solve in order to get client satisfaction and set expectations right we have also had to deal with let me let me give you an example actually this is a bit way to do it this may sound familiar so client uh client uh say You're Building some kind of chatbase system for them client uh doesn't like output say that you're building a system a chat bot for them and they say oh I I don't like the way that it says that and you say okay well they provide ideal output and you go okay well I'll I'll tune the system to try get it to outut output that kind of output and then it outputs the correct one maybe four out of five times client gets mad because it's not always consistent now this this four out of five times the other time the one out of five times that it doesn't say it it might just slightly reward it but this client is looking for exact outputs and you're trying to he wants exact outputs for you to get the money from them and finish the contract you need to give them this five out of five perfect exact same outputs every time when in reality the system is putting out one and five times something that is a slight modification of it's still saying the same thing it still means the same but it's a slightly different wording and this client goes eh client gets mad because it's not consistent um demands refund because I don't like like the way it talks so here you have a feeling I don't like the the way that the client feels about the outputs is now determining whether or not you get your money which is a a bad place to be so we have tried to solve this problem in a number of ways and eventually in combination with the exploration Milestone we came up with a method that solves all of this while also allowing us to De exploration and solving Downstream issues with client satisfaction as well so I'll now get into that cuz I know you came for the source but I'll uh I'll give you double serving cuz you know that's how we work on this channel so the issue here is as I just pointed out llms are you're dealing with two problems one llms llm uh non deterministic they are by Nature random and they they split out different outputs from the same inputs and then two you're also fighting against client feelings and subjective opinions now these two when you're trying to build a business selling AI Solutions these two go together like Mentos and Coca-Cola um they don't they don't get along uh and this is a an awful mixture of things to build a business on and we at morning side realize this pretty quick that okay well clients are going to have feelings towards if we're building any kind of chat based or or even just it's it's a Content generator how they feel about like h i don't it's not really what I like there's something about this and some sometimes would get to a point where a client is yeah this system's cool yeah yeah it's good it's good it's good it's good good and then the next week they'll say oh no we need a big change I I just suddenly don't like this anymore and you're dealing with emotions uh emotions and opinions and you're trying to make money and have some kind of consistent and predictable business uh but you're also now now at at the at the whm of these client emotions and also these inherently random systems so these two problems we realize pretty quick that if we don't solve these at morning side we're going to be we're going to be in the and we're not going to have much of a business so how did we solve this well my uh my absolute wiiz and Genius of a CTO uh we sort of tasked them with this and said hey we need a solution to this and he came up with a now widely used system um can I rub this off no I still I still have a bit more to go uh that we call it's not we didn't it's been a thing but the way that we apply it I believe is a sort of a industry first it's called coign similarity similarity so what cosine similarity does is it's a test that we can run that allows us to determine the similarity this the semantic similarity of two sets or or any number of set of of uh actually I believe it is just for two uh the the the similarity of two pieces of text so whether that's a word just apples and oranges whether that's phrases whether that's whole bodies of paragraphs whether it's a whole book uh you can use coine similarity to get a a t tangible metric is really the key thing here a tangible metric that you can point to and say yes these two things are similar and for us with these exploration milestones we started integrating coine similarity testing as part of the exploration Milestone determine if we can get the the the right mathematically most similar types of answers that are going to be satisfactory to the client so we can kind of make sure that we know what we're doing ahead of time and be like yep we can get a we can get pretty close to this uh but more importantly it gives us a a metric that we can use in our contracts to to stop that subjectivity and oh I like the I like it but like it could say this and this or hey it's giving me the same answer four times but there's one that's different it's like well mate at the end of the day our contract says it's the semantic meaning of it and the similarity of the text if there's if it's one word changed this thing protects us so let me let me break down how this works so coine similarity uh this is going to get a little bit technical but I do quite like uh getting in this and it's important for you guys to know this to some degree so Cent similarity this is going to really challenge my uh my art skills or um down okay so that's a three-dimensional space and we're going to have so it's kind of going backwards by the way it's not facing us so we have one point over here we have another point over here and we have another point uh up over here and so what these are is coordinates in a multi-dimensional space in this case we'll just do three dimensions so that it's easy to understand but in reality this is happening in in I think 8,000 what was it 8,000 Dimensions it's impossible for the human brain to comprehend but uh this will this this will suffice um so we have this 3D space we have .1 2.3 and what coine similarity allows us to do is determine the similarity of these three points so this might be a word of of beans this might be P's and this might be car and so what the co similarity test does is that if we go to our client in this exploration phase or in the the actual contract we will ask them for hey can you give me uh 5 to 10 examples of inputs and outputs that you would like out of the system and when they give us the inputs and output we then have something to grade our AI outputs of the system against so say they have a chat bot and it needs to be outputting a certain type of answer we can run their 11 inputs against our system so we have our system here we take the 10 inputs we run them each through our system and it outputs 1 2 3 Etc and then we have the original 10 inputs and the well this is the system that they used to create which is their own brain of okay what what should this thing be answering so this is a human this is an AI and then it's going to Output the well it's not really outputting the brain has outputed it but we have 10 and 10 and the 10 that we create from our system we can then use coine similarity to get the average or we can we can compare the similarity of this one to this one and determine if yes our AI system is creating outputs that are similar enough to this we add Aver is M and then we can get an overall similarity score so coming back to the graph here so we have this one here which is is one input might be uh how do I change my password and then this one will be the AI output one and this one will be the human one they might say in order to change your password you need to go to XYZ website and then based off the AI system that we built it might say in order to change your password please navigate to the S so they are fairly similar and then when we run them through the coine similarity test it's conver ing these words and it runs it through an embedding model and the words get converted to a a set of coordinates that might be like 0.1 0.5 all 0.9 uh and it goes on and on and on and on and on but in this case we've just got three coordinates because we've got a threedimensional graph so what the cosine system does is it converts words into their numerical equivalence within a vector space and by having these coordinates it can plot them in the vector Space by their similarity so in this case you see beans and peas are are close that's because the embedding model that we use and run the beans through it's going to give out a coordinate and then we run P's through it it's going to go H okay that fits there and they end up being close to each other within the vector space and so how does what does coine similar similarity actually mean well it's the measurement of the angle between these two bad boys here so the smaller the angle the more similar they are the larger the angle see if we go from P's to car or from Beans to car the angle here is much larger oh I'm getting back into my my uh what was that calculous in trigonometry and whatnot um but yeah beans the car is a much larger angle therefore the coign similarity is going to be lower than B's 2 PS that makes sense so this is taking the subjective opinions and emotions of people who go oh but I like the outputs I don't really like the outputs I don't care you have told us these in the in the exploration Milestone we won't necessarily hold them to that in the contract of the exploration Milestone it would just be hey look give us some outputs and uh we'll compare them to it and internally we can check and use that in the end of the report so at the end of the exploration Milestone you'll give them a report and a proposal which we'll go into in a second but the numbers that we get back from our testing if we we say okay look by the way cosine similarity goes from0 to 1 um so high a good range uh we go for in our expirations we might be looking for 0.85 or or uh or higher but for our final builds we'll be going for anywhere from 0.9 to 0.95 in most cases so that means it's it's highly similar and when we do our expiration Milestone we'll be looking to see if we can get this sort of number and say okay give me those 10 out 10 outputs and 10 inputs we're going to run it through the system see what we get okay yep we're around 8. 0.85 or higher we're confident that we can build the system for you so it's as much of client satisfaction as it is for our sanity we don't want to be taken on trying to take down a mammoth with a slingshot you know like we need to be taking on clients that we know that we can deliver for because I tell you what These Guys these these early adopters have huge expectations um it's actually it's more the people who are the sort of early majority who have the bigger expectations cuz they like oh well I've heard all the stuff about AI let me give it a try and they expect that to just completely change your business so you need to make sure that they're not expecting too much and given the data and the kind of inputs and outputs they have provided we are able to get close enough to it that we can feel confident in going for a a complete contract a full contract on it so that's C on similarity you can convert the 10 inputs or five to 10 inputs that the client gives you into uh or the output sorry into a set of coordinates or or numbers that a computer can understand and when we convert these over we can plot them um automatically we have a tool for this that we give to our accelerator members so that they can do this testing um for their clients we have an internal one as well that we use at morning side and you just run these run these outputs through the co similarity system it's going to determine the similarity of them and then we'll average it out and say okay what was the average number so that's cosine similarity might sound a bit complex but we are dealing with complex issues and complex technology so uh and complex emotions of clients too H and so how does this tie back into the exploration system and how it changed our our sales process and sort of revolutionize our business and how can how can you you use it too really so now they've paid us we've done the exploration Milestone and the uh we then need to present the the findings to them so once you've done all this work and you've you've created your uh you've given this to your divs they've gone and that's a that's a that's a the coding symbol uh they've gone and done the work they've done the exploration we've got the cosine coine similarity scores out pretty confident we've done enough of the exploration that we can get to the point where we're confident in pitching uh for the full project now uh this comes out and we now have a proposal that we put together which is let's just go half and half here the first half is going to be the findings the findings of the exploration which is hey look we did these tests and these are the kind of outputs we're really happy with this we're kind of concerned about this so this first half is the findings and then in the second half you're then going to be pitching your proposal and in this proposal we need to have that's when you get your your developer either you do it yourself or or the the lead developer on your team then you need to get them to create that proposal which is what Spencer was doing earlier in the in the older system which is okay what are the sections how many hours three four Etc how many hours is each of these things going to take and that's going to determine how we can price it within this proposal so you just we typically just mark up we get the our estimates from our CTO Spencer and then we just mark up based off our kind of inflating markup rate that we apply to our our services um so whether that's a few hundred now whether that's you're get to thousands of dollars uh you can just kind of Mark that up and and apply that overall and that proposal you will also include gosh makes me realize how much how much there is to cover in this stuff uh I thought this was going to be a quick video but inside the proposal then you have the cosine similarity score that you agree on so this will become part of the contract and that is what's going to save you from the subjective opinions of clients who are unhappy with h i don't really like how it sounds that kind of rubbish first half the findings of the discovery and here as we get to there's so much for me to talk about on this I probably can't squeeze all into this video but there's a number of different factors as to why this works and why it's so important uh one of them is that you get a chance to build trust you get to show them how you operate and this right here is the most important part because they've just paid you $800 it's basically like a like they're just sort of having a dig and saying look these guys seem professional let's see if they really know this stuff and they've given you a small amount of money and you've gone a way that should pay your developers and hopefully give you a bit of extra so youve again you're getting cash flow here which is great even if you don't end up taking the project on you can just get paid to do Explorations for these people which cash flow early on in your agency that that kept us alive I sh you not but then you come down here and you get a chance to show them who you are show them how you operate show them your level of professionalism and the findings and the way that you do this presentation is going to be key so this is actually done on a call oh let me see if I can get this again this is done on on what we call our proposal call so we'll get them on a call 30 minutes to half an hour we'll go through the findings of it very professional hey this is what we did we're concerned about this this is the kind outcomes we've got um this are the co similarity scores and we you can kind of introduce this as the concept on these Calles we can say hey look given the the inputs and outputs that you gave us we were able to get to a 0.85 similarity which means that it's very good um and and with a bit more bit more funding and a bit more effort we can get this to a point where it's it's far more similar it's up sort of the 0.9 0.95 range um so you get to give them the findings and then you give them the proposal based off the estimations that your your lead developer or yourself have come up with and that goes in the second half which is a proposal breaking down each of the parts of the full build and how much that's going to cost them on the call you need to have the contract ready and in the contract um if they're all good to go and there's no complaints then you send them over the contract get it signed on the call if possible because as soon as that call ends your chance of closing them just just goes through the floor um try to get them to to sign that contract on the call because as soon as they signed it doesn't matter about the invoice really if they've signed that they've made a a commitment to paying that invoice as well so as long as you can get it signed walk into the contract handle objections whatever you need to do point out that the cosine similarity metric is in there and say hey look because these are kind of subjective output uh this is something that we do to both Safeguard ourselves and to make sure that you are protected and that we have a level of of standard uh a bar to work towards and the quality of outputs that this system generates so having this C similarity pitch it as a benefit for you and them which it is and say hey look this is the this is the agreed one we're going to be aiming for 0.9 anything more than that might be overfitting CU you can make the system too good at giving certain answers um you want to keep it around that 0.9 to 0.95 range um and that's what we'll put in our contract and that is the way that we can stop this if you've ever done these kind of projects before where it's Custom Custom builds uh you'll know that it can just kind of go on and on and on and it's like review revision hell we call it review hell and it just goes on and on and on and it's so painful when you have a chatbot system that you've created and then you send it off to them and they send back like a 30 40 minute Loom and you just especially when it's a low ticket client as well if you've been paid three or $4,000 and you're sitting there on your Saturday afternoon going through this guy's revision notes and then having it go back and fix the system up and man it is not pretty and it's not the kind of lifestyle that I want for my business and so we are try to avoid as much of this crap as possible by saying no we have the coign similarity this thing is performing as expected we're done send me the money send me the money so yeah hope I uh got all that across but the exploration Milestone has been a complete game changer for us and if you want to incorporate something similar into your business while well you may not do the exact same thing as a as an AI agency owner what you do need to think about is this line and how do I spend less time talking to tie kickers who are going to waste my time and how can I push quickly to start filtering out cuz you guys may do some Mund Le generation you may do a LinkedIn whatever and you may be having conversations but you'll pretty quickly realize that a lot of these are just conversations and you need to start finding the the guys with money to spend these are the guys you want to be talking to and the longer you give them like there's billions of people on the Earth and don't assume that all of them are like you and I and now rational and and don't want to waste time some of them are deluded some of them have mental issues and they will happily sit on these calls and waste your time and that's not going to help your business you're going to end up spending so much time talking to people are never going to be clients while it may be good sales practice if you need that but talking to a looney is maybe not the best kind of sales practice you you should be after but yeah take this concept whether you apply it bar for bar or whether you just take the kind of distill it down and say hey look I need to put some kind of filter earlier on whether that's uh it's early on in my when I was doing calls for my Consulting calls on my channel I would have a $300 or $500 Consulting call and that those are the people that would then take on as Dev clients so if they came through Consulting calls they've already kind of pre-qualified or pre-selected themselves saying hey look I'm willing to spend $500 on this so that was kind of our earlier exploration Milestone but we started getting leads directly through our website not just through my uh through my channel and through the through the Consulting call link so yeah take it apply it and this is particularly good for for General agencies like myself who are doing custom Solutions every time if you are a niche agency um and you have the same thing to sell every single time may be a little bit different but I think this is Handy to have in your back pocket either way and it will help you get past $10,000 a month which was a a major issue that we had and the rest is history so hope this been helpful guys um I certainly enjoyed making it and yeah exploration milestones and Co similarity I hope you got it all uh and I will see you in the next one keep going uh you can do it this opportunity is not going anywhere but you need to make back you need to make moves on it now um if you want to get there so I'm rooting for you and keep going ciao