075. AI-Powered Customer Research & Message Strategy with Chris Silvestri

Foreign.

Hello. Welcome back to another episode of boundify's Demand Gen Studio. As always, my name is Miranda Jenkins, your host and I'm joined by Emilio Garcia from boundify. And today we are also joined by Chris Silvestri. We have another guest episode for you. How are you guys doing this morning?

How are you? Hey Miranda, Emilio, thank you so much for having me. It's a blast.

Yeah. Good to have you. Awesome. Chris is the founder of Conversion Alchemy and specializes in messaging strategy and copywriting. So Chris, why don't you kick us off and just kind of share your path in marketing and what led you to focus on messaging strategy and copywriting?

Yeah, so my path actually starts way outside the boundaries of marketing. I was a software engineer in industrial automation and I did that for 10 years back in Italy where I'm from, before moving to the uk. And the funny thing, I stumbled on marketing because I wanted, I don't know, like a more flexible job, something to do online because I was playing in a band, drums and so I wanted more time to tour with my band. And so I stumbled on lots of online types of businesses, tried different things before actually stumbling on copywriting. And it was instantly hooked by the psychology behind it, understanding decision making processes. And so yeah, I started studying that on the side, weekends, evenings, started doing a couple of projects on upwork back then and I went full time after a couple of years again tried and experimented different things inside the the industry, copywriting for different niches, different types of businesses. Then I stumbled on SaaS. Then I worked at an agency as a UX lead just because my past programming experience was very, very much like type of logical programming, very much also designing human machine interfaces. So there was some user experience involved. And after my experience at the agency then I decided, okay, let's go back on my own. I still want to do copywriting. I'm still passionate about writing to persuade and to have people take. But I saw something in the user experience work that I was doing, especially watching hundreds of usability test videos and truly putting myself in other people's shoes and understanding how they act on websites, how they make decisions. I wanted to try and combine a lot of the research that we used to do at the agency with the copywriting work because I saw a lot of similarities and that's basically what brought me to build Conversion Alchemy, my current business in 2021. And I decided to focus solely on B2B SaaS companies. And yeah, combining this passion for copywriting, the psychology and the user experience side of things and which basically brings me to today.

Yeah, that's fascinating. That's such a like analytical field. Going from such an analytical field to then a more like creative field. Do you find yourself like bringing some of that structured like analytical thinking into copywriting?

Yeah, totally. And especially nowadays where you basically have the basic writing basis covered with AI. I think one of like the underrated skill or. Yeah, under discover skill is more of the system thinking behind architecting the messaging strategy and being able to build the systems and strategies that actually scale your messaging throughout the organization. Not just writing a single webpage or a single email sequences is actually building the strategy behind it and then operationalizing it inside the company throughout different channels.

Yeah, that's interesting.

I gotta add to that that you know how popular is nowadays that you have like first party data or research on your own for to share with clients. One that we should do, Miranda is asking all the markers that we interview and engage with, what is the original feel? Because you know, you are, you're part of the trend, Chris. Right. All of us is like where you. Do you actually study marketing to begin with? No, I'm coming from other thing or feel and it's just to stumble upon this and I think that's very interesting because it's an industry that because of the diversity of all the background that we all have, we bring something to the table that is unique. Right. It's coming from our field and you bring it in and build it. At the same time though, I think we are reinventing the wheel all the time. Right. Every new practitioner is kind of like not aware of some of the tested principles of marketing and we rediscover them as we implement as we practice.

Yeah, exactly. And like the, the way that I started out learning this stuff was like right from the source, from direct response copywriting, the 70s, the old ads. I used to copy by hand sales pages for days until my hand hurts. And yeah, I think like the basis, the foundation are super important still. And that's probably where, where someone should start. If someone wants to get. Wants to get into copywriting or messaging, those are still the place to actually start from.

Yeah, yeah.

So tell us a little bit about conversion alchemy and kind of what problems that you set out to solve with the company.

Yeah, so we work with mostly, As I said, B2B SaaS. Companies typically service A to C sometimes enterprises, but not that frequently. We work a lot with private equity firms that acquire tech companies as well. So what we do is most of the times the Clients come to us saying, oh, we need copy for these new product pages or we need copy for these emails. What we find most of the times is that these companies don't really have a copy problem. They actually need to fix some of the layers before the copy at the source, which for us is first, even before starting any strategy work is the research. So what do they know about their customers, their market? What do they know about their point of view, their, their product internally? We need to clarify that. Then we jump into the positioning strategy. So are they clear on exactly what they do, who they do it for and how they do it uniquely better or differently then the messaging strategy? How do they say that? Through different lenses, which could be different customer Personas or different points of view. What's their strategic narrative that stems from their unique point of view? All of that. And finally, once we have all of that clear, then we can jump into the copy and writing, I don't know, individual pages or email sequences or sales collaterals. And so this is the typical flow of a process for us. And as I said, it's quite tricky at the beginning just because you need to kind of diagnose where everything is breaking for them. And so that's part of our process as well.

Yeah, love that you're saying that because it's so similar to our case.

Right.

It's like we as a demand generation firm, our clients will come with a symptom of a problem on advertising. Right. Either they have failed in the past or they're doing right now. They're happy and you stumble upon the same situation. Right. As well. First we need to understand who are your clients. Right. Or the unique pain points you solve. And definitely usually is that the things that you're saying in your advertising, the channels you have picked, the audiences that you have picked, do not reflect either who your clients are or where they spend time and, or as you said, right. Your, your positioning, it's not appropriate. And typically, even if it is your copy, your copy do not reflect it. Right. It's like the advertising copy do not reflect your positioning and they do not cater to the problems of your clients. So it is, it is very interesting, Chris, that you're, you know, just this analogy that obviously most prospects will reach out or potential clients will reach out for the thing we do. And in their minds, you know, all this background, all these backend work that has to happen might not be seen as important, but it's cruel and I.

Think it's fair because I don't know, in my case my clients are CMOs, heads of growth, product marketing people, sometimes the founder, but it's a bit rare. But if you imagine all the work that they need to do, being able to pinpoint exactly what's not working, it can often be hard, especially when they have to bring together all of different departments. There's a lot of alignment work to be done and a lot of times when we come in, my clients appreciate the fact that a third party like myself was able to kind of validate some of the assumptions that they had. And so that's an additional point of proof that they can bring in to get buy in on the projects.

Yeah. Now what really got me excited about our conversation today, Chris, as we were talking about this, this episode is as you know, right. And everybody knows our clients and everybody, every marketer and demand gen expert out there is asking how can I leverage AI to do faster, better work for, you know, whatever field I am. And I was really intrigued by your point of view and experience on that foundational work. Right. The two things that really I'm interested on is if how right can you leverage AI? For one, as you were saying, is to understanding your clients, I think there are opportunities there. But second, I think the one that is really interesting as well is to define your positioning right to. So I think I'm super, super interested. We talked about that synthetic research, kind of like I put a hole there on their topic because understanding your clients or doing customer research is one of the most manual things that you can do. Right. And there's not a ton of ways that you can automate or simplify that work. The closest what we have had so far is to record those conversations and hopefully have a tool that kind of gives you a transcript and you, you know, read through all of those comments and you have a structured way of survey. But I'm really interested on your point of view on this, you know, synthetic customer research. What is it that he does and how helpful or can be a decent replacement to the more manual work. So let us dive in into that, you know.

Yeah, that's something that I'm pretty passionate about. Lately I've been diving into, I've been having calls with different founders of different platforms as well, or researchers. And so there are a couple of different areas in synthetic research. So there's synthetic data, which is basically generating fake or AI generated data that resembles data that you might need to conduct your research. Then there are synthetic research platforms or synthetic conversational tools which are basically chats or they call it digital twins, which is basically copies of your customers that you can have chats with. And so these are the general areas in the synthetic research field, but also inside the synthetic research kind of industry, there are different ways of doing synthetic research. And that's typically the main objections that I get when I start talking about synthetic research. People say, oh, you can't replace humans just with synthetic data. That's not accurate. AIs tend to make you happy as much as possible. That's their sequencing effect, or they hallucinate and all of those things. But there are different levels of synthetic research. So there's fully synthetic data in this case, which is again, as I mentioned, completely made up by the AI, basically following specific guidelines, but it's completely AI generated data. Then there's partial synthetic data, which is basically AI generated data that closely mirrors your real data, but basically excludes any sensitive information like names or specific information that you want to exclude. Then there's the most important, I think most undervalued one, which is augmented synthetic research, which is conduct your research, your human research first, or at least the basis for it, and then expand it using. Using AI. So you basically get the same kind of accuracy, the same directional power and effectiveness, but at scale. Because let's say that I can interview five people with synthetic AI data. I can basically turn those five interviews into as if I did 20 interviews. Because AI is able to spot patterns, expand them, replicate them, multiply them. And so you get a lot more data that can be actually useful and accurate at the same time. So these are the types of synthetic data and the way that I approach it, I use it in most of my work. So from the research to the strategy to the copy, I use AI in all the stages of my work. But yeah, I'll stop here in case this is a bit too much and you want to ask questions.

No, no, no. Plenty are coming. And also for our audience, I want to highlight and correct on this briefcase because probably you have done way more than we have done on this. Let me know if I'm mistaken, right? I will start saying why customer research, from my point of view, is so difficult. And even manual, regular customer research is not deceiving, but it will not show you all the truth because there's a lot of inferences, right? It's like I am very passionate about this idea that when you ask someone about something, the answers do not necessarily match the behavior. Right? The example that I use all the time is your Netflix or book wish list, right? You look at your wish list of whatever you want to do. And then you look at the actual movies or books that you have read and they don't match. Right. The categories, the intentions do not match. Because I feel wish lists reflect our future self. And our future self is always better. It's a better person than our actual self. Our actual self is not as great. Right. So I think when we dive into just regular customer research and we are backtracking a customer, asking them why did they buy or didn't buy for one brand, they will come up with an explanation. And more often than not, that explanation is just a rationalization rather than the actual reason. So there's always that big gap. The other thing that I feel that on regular customer research is really important is that that when you're asking about a product or service, you're making it about the company. Right. Because the buying journey is the same for, you know, let's say if you, if you pick project management software, right. There's a bunch of companies that will pretty much ask similar questions about the buyer journey for a customer, but they will tailor it to their product. Because you're trying to find both on winnings and losers. What is it that, you know, make your company different? Exactly. Right. Where were the things that, you know, made someone decide to purchase or not from you? And so I think because of those things, when you move that into AI, you will get a bunch of potential pitfalls. The biggest one, and correct me if I'm wrong, is that the biggest issue, the biggest question that you have to ask about these models that are doing synthetic research is what is the training data? The training data set. Right. I believe, and maybe you, you tell me if I'm wrong and I will stop there is that, for example, models like ChatGPT Quad and all that, there's. There's no assumption that any data about customer research has been used to train them. Right. At best, there's some information out there, but they was, they weren't built with that purpose in mind. So probably from what you've seen, seen models or companies that have developed a specific training sets for customers will do a better job. That is to just jump into ChatGPT and say, pretend you are this, this and this and you know, kind of like answer these questions. I don't know what you think about that.

Yeah, no, that's totally, it's totally true. And one trend that I'm seeing, I was listening to a podcast as well this morning and they were talking about how they started doing this. There's also the trend of go to Market engineering now and a lot of these go to market engineers. They build these custom GPTs where they basically map out your Personas. I don't know, they have different GPTs from different customer Personas and you can ask them questions. But the fundamental issue I think with using platforms like ChatGPT or even Cloud or Gemini for this type of research is that again these models, as you said, they haven't been built with this purpose in mind. So they have all their internal pre training algorithms or system instructions that basically have specific personalities and you can actually see from chatting to different of these models and seeing the different personalities and so those are embedded in it and you can't really get rid of them. That's why when I use, when I want to write or test or do some small scale tests on my own before using some of those purpose made platforms I typically like rely on third party applications, something like Typing Mind or Team GPT are two different platforms that allow you to plug into the API of these models. So you can plug into the anthropic API or OpenAI API and you basically remove all that pre training that's inside the commercial chat models. Right. So you have at least you have cleaner sources to work with. And obviously it also depends on my data because I always like to start with real research and real human data they have. So the more of that I have, the better is going to perform. But again at that stage I'm happy with a 70% accuracy as long as it's like at least directional to tell me okay, this is a good direction to go in, then I can go forward and then at the final stages I can move on to those purpose made platforms.

Yeah. Which trigger a couple of questions. Right. So I agree with you. The first one is the other thing that I wanted to say is that the training set matters a ton. And I will ask you here if you have some recommendations of dedicated models for customer research. But I think the second hurdle is the context window. Right. I went to Inbound last week and I listened to these wonderful conversations, wonderful presentation from Dharmesh, the founder, one of the co founders and, and he's really revealing that the other big thing is the data you do bring into the model. Right. You're as you're saying, you write that your view of the world of how something should be done, that's super, super important because increases the quality of the questions that you might ask. And I tell prospects and clients that if there's one thing that I'm really liking about regular software implementing AI is that it's forcing the practitioner to answer some basic questions. Right. Like when you're asking about icps and positioning and values and those are kind of like the things that you have to write there in order for the AI to actually help you I think are super powerful because he's teaching a generation of marketers the things that you need to know before actually ask. Start asking questions. Like you said, you are bringing your own data here and then combining that with the training set and coming up with something great. So that's, that's kind of like before asking you another question, the question that I have for you, do you know or can you share some like mod, you know, purpose built models that you have seen for customer research that work better than the, you know, vanilla ones? I guess.

Yeah, I mean the ones that I tested personally and that I've also spoke with the founders as well. I'm not sponsored or anything, but I tried them personally, I use them regularly. So the one is called Synthetic Users and from what I've seen and what I tried, they target a bit more enterprise clients and they allow you to conduct more structured studies. So let's say you have a set of questions, then you can, can import them again. You can describe your Persona, you can add your context to it and they literally prepare the results of like an actual test study. Different questions, different Personas, reply to all the questions in a structured case studies. We take away and that's one Synthetic Users. The other one is Ask Rally. Rally is the platform and the website askrally.com and they, they're target a bit more small medium companies or even solo founders, which I think is pretty cool. And they allow you to have chats. It's more like a chat interface again. But the nice thing is that again you describe your Persona and they can take that Persona. Let's say that I have an ICP profile that's describes different ranges between 35 to 45 year old, different types of professions. They take that profile and can replicate and expand it into I don't know, 50, 100 individual specific Personas that you can test that you can chat and you can have chats with and they give you like a, like a condensed response to the chat message and then you can dig into the individual responses as well. Kind of like if it was like a survey with the final response you can have, you can test different, like multiple options as well. If you have a couple of different headlines or pieces of copy and they can tell you, okay, I prefer this one, you can share images or I think also video they started doing, which is pretty cool. And yeah, these, these two platforms I tried and they are pretty useful especially when I need more data and scale for clients.

Yeah.

Or otherwise I would just do my internal testing first at least. But it's pretty cool because even for example, Rally also launched their API. And one very cool thing that you can do with their API is that you could basically create, I don't know, workflow with something like N8 Automation Platform. And even before launching ad variants, you could test your ad variants automatically with their API. It could tell you, okay, this is the best variant for this specific Persona. And then you would basically launch only that variant that's been validated. Right. And you can do that at scale. No need to wait weeks to get enough data. And so I think that's pretty, pretty useful. There are a couple of other ones. There's another one called Yuri I think and they do, I think it's basically a mix of these two. I think they're positioned kind of in between. So there's a mix of structured studies and chats. But everyone's kind of taking their different, their different path and takes. In general, if you're looking into synthetic research platforms, there are, there's two categories that you need to differentiate between. There's the product development type of research and there's the marketing messaging research. One is typically these platforms, these platforms, the product development side, they need to test the usability, the performance. So they work with this, it's called agent experience rather than user experience. They basically have agents that go through websites, go through apps, test for different flaws, different leaks or bugs. And so this is the product development and then you have these platforms like the two that I mentioned that are more marketing. Yeah user research, marketing, messaging related, that you can basically ask questions and, and get feedback from like real users.

We will put those in the, in the notes of the show. That will be fantastic. One closing thought that I have about this and I would like to, to know what you think before jumping into how to use AI for positioning, which I think is also a great, great conversation to have is that I'm sure these tools that you use are sophisticated. Maybe price points, especially for the version that is enterprise, are quite high or at least not for everybody. I believe that for the rest of us, right. For someone that is not necessarily looking for a very sophisticated tool, at least as you were saying, something that will be super useful nowadays is get your conversation with customers right. You're 5, 10, 20 and as you said record them all and store them in a place where they can be part of a context. Because I agree with you. I remember this talk that Steve Jobs have many, many years ago where he was theorizing, right. Wouldn't it be wonderful if we can record everything that Plato ever written and then ask him questions right. In real time rather than just read the books? Well, AI is a little bit like that, right? As you said, if you get enough conversations or enough information about someone will not be perfect, but AI will pick up the pattern and then you can make inference about it based on what they have said. So I think, yeah, maybe you cannot say that out of 10 people, you can map a market, but you have the possibility of ask maybe questions that you didn't come up at the time and based on the training set, get a good guess of what will be the answer. If you were to ask that person again, something different. I think that's very useful because that way you can recycle, reuse some of that minor work that you did in the beginning and use it as a source of knowledge for your company. I don't know what you think.

Yeah, yeah. Well, one thing that you said at the beginning as well, that I think once you're able to build the right context with AI, I think AI really, if you use it well, can push you to ask better questions rather than just a lot of people say. There's been studies that say, like, AI, it's kind of weakening our intelligence and kind of making us more like relying on AI systems. But if you actually put in the work of. I like to compare the building of context to. I like to say, I'm not really asking AI to write for me anymore. I build it a world and then I have a conversation with AI. It's basically the world that I have in my head of context. I have to rebuild it inside the AI so it has the same world, same sources, same memories, same references. And then I just have a conversation and I ask it for results.

Yeah, I have a question kind of off of that. So I'm curious of any of your clients if there's like a level of kind of like explanation or like buy in that you have to get from your clients around this kind of new system of using AI and maybe even in the context too, of like marketers listening to the podcast of then starting to incorporate AI into their workflows and what your response is to that.

Yeah, so far, honestly, I have to say there hasn't been really any pushback. Actually, I was speaking With a client last month we are still like the proposal stage and I wanted to propose them to do a round of synthetic research before launching the copy to just to validate the final copy. And they were, they were basically like, oh, our internal team is kind of already doing synthetic research. So like it's something that we can handle internally. And another other companies. Actually two of the last ones that I finished with in our case study call, they mentioned how they were able after our project was completed to take our outputs, output, documents, the messaging, positioning, a value proposition, store them in ChatGPT and create these custom GPTs. So I think a lot of teams are kind of, they're still not clear maybe on a specific strategy that works well for everyone, but everyone is at the same time trying their own thing, kind of tinkering with it. And so everyone's pretty much open to anything that works and scales now we're still kind of very much exploration phase I think.

Yeah. Awesome. So let us jump then into you know, positioning and again. Right. Making the assumption here that, you know, most of us have done some work position based on customer research, competitive research. And you kind of like how do you are leveraging AI right now to help you with that part of the work before and obviously probably with copy itself. And you were already referencing so very curious on what is your workflow right now.

Yeah, so the positioning, which is the first stage that a lot of that comes down to how I structure the strategy. So everything that I do with AI comes down to what I do before that with, with the actual work. So for positioning is the research. So how do I structure research? I think it's important to also like for myself for the teams to give a structure to research both so that they can actually use the research to take action and create strategies, but also for the AI after. Just because you again you give it a specific world structure, you do world building in a way. And so the way that I structure research is I see first into three main areas that you need to cover every time or at least cover as much as possible, which is internal research, which is typically interviewing teams looking at sales calls, transcript, everything that comes from like internally and the internal knowledge that we have at the moment. For positioning also means having a collaborative workshop. So that's. We typically have a 90 minute workshop with all the stakeholders, product sales, marketing, founder. And we try to jot down, okay, what are our competitors, what do they do that's unique, what do they claim on their website and also what are other competitive alternatives that maybe are not direct competitors, but still other solutions. And then we try to identify the three to four differentiated value themes that the company offers that together form their unique value, their unique positioning. And so that's super important because first you have alignment internally when you do the workshop, which a lot of times doesn't happen instantly. There's some, a lot of conversation back and forth and agreement that needs to happen. So that's super important as well for when you feed that to the AI. And, and finally it's also identifying the market category and the debt positioning, clear, succinct positioning statement to do that. Again, the research goes down to the internal area, the external. It's where we try to understand the size of prospect so people who are not customers yet and then buyers. And then we try to map that transformation that needs to happen in between. And finally market research or studying competitors, reviews anything that we can find in the industry specifically. So once you have those three areas structured pretty well with the right documents, then you can start feeding those three AI and get to the positioning in a pretty good state. But I also like to look at, again, at the usability aspect because as you said, a lot of times people say something but then they don't do it or they do other things. And so I always like to run usability tests as much as possible or look at heat maps, user recordings, how people use websites, or watch those user recordings, typically 50 user recordings at least, to see the patterns and extract the insights on that behavior that I see, which tells me a lot of data on what types of decision makers, customers or prospects are, whether they are more emotional, logical, and how they make decisions on the website. And then that's again, stuff that I can feed to the AI that can increase that surface that the AI can cover with its expanded insights. And so we basically get to the positioning where we have a clear idea of where ICPs are, whether we have a check signer, a manager, an influencer, or a daily user. And we try to map all of those different Personas out. Then we create a landscape of the competition to give to the AI. And finally some important data on the specific product features that we need to speak because the AI again, doesn't know the technicalities of how the product works. So you kind of need to tell it as much as possible. Yeah, this brings us to a very good place when it comes to the positioning workshop. And then we flesh it out and obviously manually, with our intuition, with what we feel is right. Again with, together with the other teams, we finalize our positioning Next step is the messaging. So again, here we use AI to kind of take everything that we've discussed. So sharing all the transcripts of our conversations, all the research that we have, and then we basically iterate through it, which is basically writing down the strategic narrative first by hand with what we found and using AI to say, okay, does this actually match what we've discussed or are we diverging? Is this completely different? Does it make sense? And probably one of the most underrated aspects of AI that I know from first doing all this, these months of research myself, as you probably guys know are aware of as well, is that you get to a point when you're so immersed in the research that either you would continue on indefinitely because you don't know where it ends, if you have data that you can collect or you kind of get overwhelmed. And the AI is great because at some point it's really good at like spotting those patterns and kind of making clear, okay, you have enough data, this is like clear patterns. We can move forward and kind of crystallize these insights. Once we have the messaging strategy, then the copy is basically almost writes itself. Just because once you are able to feed the AI the research report, positioning Canvas, the messaging frameworks with all our value propositions, then the AI can basically take all of that little like, it's kind of like Lego blocks. The AI based on your prompt can take those Lego blocks, compose their its own and kind of write anything that you need based on your goal. Obviously the more specific with your prompt you are, the better. But it all depends on the context and that world that you built for the AI.

This is great. I so many things that I have to unpack here. The first one I love that you mentioned about getting members of the different departments of the company together and having you know, like a back and forth on what do they think is, you know, their position in the marketplace, competition, value proposition. That's fantastic because I'm sure every single time, especially the, you know, the people in charge of positioning will think so everybody knows what we stand out for, right. And what are our strengths and weaknesses. And I'm sure they're like amazed of oh, I had no idea that you thought that way. Right. That someone will see some another company as the true competitor versus another department and for whatever reason, right. Some feature stands out as more valuable than others. So I think that's always helpful. Right. It's independent of AI that when you're doing these things, especially company wise, when you're saying this is what makes us different Better. There's a third word that you use there, a Unique.

Better. Different.

Yes. That there's alignment among the organization. Right. That you not not only that you source it, but you also when you say it and everybody agrees on. Yeah, you know what? After all I agree that should be our strength or. And exhausted. I think that's super, super valuable. Yeah, yeah.

The second one is, I was saying another. Another important aspect that I see typically in these positioning workshop workshops that it's kind of like a big aha moment is for example when we talk about what customers, what the best fit customers value in their product, whether it's a feature or product capability. Some people in the team have different opinions on those as well. But the problem is a lot of them haven't spoke with customers before like before. And so when we jump into the positioning workshop, having done those customer interviews, then we reveal, we share these insights again. It's another haha. Mind blowing moment.

Yeah, yeah. I believe another thing that get lost in translation when you have a solution is especially if you have not seen it on customers is that the same feature capability will feel very differently to different kinds of customers. Right. I remember that I use always an example, this solution related to some OCR technology, right? Scanning, kind of like extracting text out of images that it was super valuable to a specific segment of the patent industry. Medical, patent industry, because they have many versions, they have to be documented in writing, on paper. But obviously at the end of the day they want to have a digital backlog. So this particular technology was so helpful to this specific segment. And the only way to know that is that you have conversations about what is the problem or job that this technology or feature is doing for you. And once they discover that they went, you know, 100% into that segment and kill it. Right? Because they had the right product which wasn't as valuable for other types of customers. So I think that's, that's super helpful. And the other thing that I really like what you said is about all these, all these preparation set you up for success where you know the part where AI helps becomes supernatural afloat that. I don't know if you agree with this, but I never write a book and I think it will be super difficult for me. But I, I have the intuition that part of the reason is difficult for writers is because they probably had their preparation in their mind, right? Their own experiences, what they have read, all the things that they have learned about writing a book and the topics that they pick. But the issue with the inspiration is when you're trying to organize all those ideas in your mind and put on the paper, then you get stuck. And I think what AI can enable to someone writing copy is that if you have the right preparation, you have the right context, and you are stuck where to start, at least AI will be, as you were saying, a companion that you can have a conversation that can propose to you a set of drafts or proposals to start with. And then your expertise come in and say, I like this and this. And the other thing that I wasn't thinking about that I found really helpful is that you flip it. You say, well, what if I come up with some ideas for copy or messaging for this company and I just contrast against what I have defined is the value of our position. What I have defined is our competitive advantage. I want to have this soundboard that will tell me, you know what, this is not aligned with what you said you are. I haven't thought of that. So how.

Yeah.

Can you detail any more how you have used that?

Yeah, yeah, it's a good point also, because one thing that I still like to do as the message strategist and copywriter, it's still immerse myself in the research. So I conduct the customer interviews. Me or my junior copywriter conduct the interviews. I still write and prepare the surveys and analyze the responses. I still go and mine the reviews, whether it's 100, 200 reviews and look at all of them on Reddit or Captero G2. And so as long as I'm still immersing myself in the research, I can develop that kind of intuitive sense, the taste as well for what copy works or at least what angles work. So that when I start writing drafts for the AI, I know that these are two good directions that could work. Maybe the AI takes those directions, goes into a completely different tangent, or uses completely different language or brand voice and I can recognize it because I've done the work beforehand. Right. Otherwise, if someone just jump into the project and asks chatgpt, hey, write me a landing page for this. Even, even if you have those documents, there's still the risk sometimes, at least for now, that AI goes off in completely different, different tangents or explores new ways that maybe you haven't validated yet. So it still, I think, important as the marketer to immerse yourself in the market.

Yeah, I agree. And I will say as last talking point, Chris, for our listeners, hopefully they are the managing practitioners that have a talented copywriter like, like you and that can do the work for them or that Part can help them out on that part for their app, you know, advertising and content creation efforts. But more often than not they will either not have an access to someone like you or you know, for whatever reason they don't have access to dedicated copywriter. What would be your advice right to someone that has to do some topic for landing pages, email, content, how they can leverage some of these tools in an easy way with what they have access to typically which will be ChatGPT, Cloud, whatever of those Gemini, how they can start and do some work like that?

Yeah, I would say yeah ChatGPT is good, but probably I would keep it more for because of the memory function that it has because you can, you can reference basically all the conversations that you had in the past. I would reserve ChatGPT more for like brainstorming ideation. Then when it actually comes time to strategize or analyzing research, I would use Gemini, probably Gemini 2.5 Pro as well. First because it's included in your workspace account. A lot of these people probably have Google workspace already and so. And second is because it's got the biggest context window with I think 2 million token contexts now, which is the biggest one across all models. And the great thing is that it's really good at absorbing a lot of context, whether it's interview transcripts, survey data or anything else and kind of referencing if you ask questions on the specific context. Even if you don't create projects like you could do in Claude and then the knowledge is basically stored in the chat. As long as you use the same chat for that specific topic, it's really good at retrieving data and referencing it. And the other one that I recently discovered is that Gemini is probably the best at pushing back on some of your ideas. When stuff chats like Claude or ChatGPT they tend again tend to make you happy and tend to confirm your biases as well. Gemini, once you have all the right context is probably the best one at still pushing back and justifying its decision. Once you are pretty happy with your strategy, you have your positioning down, you have your messaging strategy, maybe you're in a couple of pieces or draft copy, then I would use something like Claude just is just for the copy. Just because it's the best at writing copy. I don't know why it's maybe there the way that they built it, but it's more. Feels a bit more human once you share those strategy documents in a project for example. And so yeah, those like the three platforms, I think each one has kind of their own purpose. Obviously if you need to limit, you could do everything with Gemini pretty easily. But if you want to optimize and you already have these accounts, I would use these three with different purposes. And if you want to scale again just because cloud, sometimes if you have the basic plan, you have pretty strict limit as far as the responses that you can get a day and then you need to wait 2, 3 hours to start again. I will use tools like Team GPT or again Typing Mind, which are pretty affordable and you can plug them into the APIs and you can basically bypass all those limits because you basically pay per response, which is nothing still. But you don't have those daily limits. So you can. And that's what I use as well just because for my clients I have to write a lot of copy which the normal tools don't work for me. And so with that I kind of bypass that and I still get the quality that I need because I use the anthropic, the Cloud API. So it works pretty well.

Fantastic. Very valuable, James. I think it will be everyone that is listening that is trying to leverage what they have will get a great value from this. So Chris, any final thoughts? And if not, you know, how can you know if anyone is interested in following this conversation with you, where, where they can find you better or what is the best channel to reach out to you?

Yeah, yeah, I'm always happy to, to chat about all this. You can find me on my website, conversionalchemy.net and I have a newsletter you can sign up for where I talk about some of these topics or I'm typically on LinkedIn so you can find me there and connect and we can have a chat.

That's our community. Fantastic, man, that's been great. Thank you for sharing all these valuable insights with us and yeah, appreciate it.

Thanks so much, Amelia. Hi Miranda.

Bye bye.