Transcript
Speaker: Welcome to the AI Micro Skills Podcast, where we're empowering marketers with AI tools to overcome the overwhelm. Today, we're diving into a fascinating discussion with Mark Thomas, who is a senior growth marketer at Podia, about AI for customer research. Mark, welcome to the show. All right, nice to be here. And that is one beautiful background you set up. A beautiful foreground too, Dan.
Speaker: Yeah, I appreciate that, man. That's a whole story about how I got into home video studios, but maybe for another time. Sounds spicy. Yeah. Well, I started the vlogging thing and like Casey Neistat and found out the world was poorly lit almost everywhere you went, indoor, outdoor, and the sound was disastrous. So I'm like, I'm just going to set it up once and forget about it. So here we are. Short story.
Speaker: But I'm excited man, cause you posted on LinkedIn a few weeks back and we finally get here to the podcast today to talk about how you're using AI to do customer research. So before we actually dive into how you're automating that and using AI to do it better, faster, kind of bring us back to the very beginning. Like how did it, how were you doing it manually before? What was the problem you were solving? So it's a problem that I solve personally, but also it is so common in
Speaker: B2B marketing, if you are doing any level of customer research beyond sort of, we did it once, which most people will hopefully get beyond that, you're ending up with this sort of, in your business there is so many sources of customer insight, right? And like you're thinking about like surveys, there are phone calls, there's like,
Speaker: you know, in -app surveys, maybe someone did some screen recording analysis, all this kind of stuff. And it's all over the place all the time. Now, the challenge for me, because we have a huge amount of this at Podia and we do a huge amount, right? Like we've got hundreds of thousands of users and stuff is coming in from all the time because they love giving feedback. We're processing huge amounts of data and kind of customer insight there.
Speaker: Now, what we realized was happening was that some of that was in one survey tool account, like in type form we use. Or some of it was in Hotjar because we were doing screen recordings. And we've got iterate, which has our in -app surveys. And then we've got fireflies for recording phone calls. Now, that's valuable information. But it's only valuable as long as you can put it to work. And to put it to work, you have to have it in front of you.
Speaker: What was not happening is we weren't having a cataloging system where this research would come in and then be in one central location where we could maybe tag it, we could summarize it, we could create an archive of it that was in a Google Drive rather than all over the internet.
Speaker: And actually when we started to do that, we were able to make research and customer insight really like come alive in the company. And we were a very customer research focused company anyway, but it took on a new dimension actually at Podia since we built this little system internally, which I've been kind of attending to a course afterwards in a very Podia sort of way. Nice. So yeah.
Speaker: common problem. I mean, it's so many companies I've worked at, we're always arguing about where the single source of truth is, right. And some people are like, it needs to be a sauna, because that's where we're tracking all the client work, if you're a services company, or it needs to be HubSpot, or it needs to be the dashboard thing that we plugged into HubSpot or Salesforce, you know, it's like, where's, where's, where's the data going? Where did you guys end up landing on where to put the data?
Speaker: Well, I think probably like all those teams who choose Asana or HubSpot, where we work is in Google Drive. All of our tools basically end up back in Google Drive or Slack. And so we ended up putting all of this data into Google Docs and Google Sheets, which is just incredibly effective. There are definitely some technical limitations there, specifically around database management.
Speaker: which we ended up solving with some AI stuff. But there are some kind of limitations, but it is good for proving out this is worth investing in, right? And kind of scaling it to the point where it's, I hope we managed to get it to the point where it's almost unusable. It's so effective. That would be the ideal outcome, if I'm honest.
Speaker: Wait, wait, wait. You want to make it unusable. It's so effective. I want to make you contradicting things. I'm like, back up, back up. All right. There's a limitation with Google sheets. Yeah. Uh, which is that you can only have, I think it's like, I think it's like 400 ,000 rows in a Google sheet. I want to reach that point. Right? Like if I reach that point, honestly, job done. I see.
Speaker: We have proven that research, uh, collection and kind of cataloging and management is really effective at Podia. It's a badge of honor to reach the limit of a software program. Sometimes like maxed out Excel, like some kind of like, I think it goes farther at 400 ,000. I think it reaches in the millions or something. I had a friend, a coworker that was doing, I was there actually direct report doing work for me and he broke it. I was like, Hey bro, you know, you're doing some serious stuff when Excel can't accommodate for the data that you're doing.
Speaker: I had one of the most stressful weeks of my life by melting down Excel numbers, all of the spreadsheet tools we tried to put this dataset in for a customer. It did not work. It was the most stressful thing ever.
Speaker: I still get a little bit of PTSD when you just said, have you ever tried maxing out Excel? Yeah. You max it out or you have to upgrade your computers. Your computer is just not powerful enough to run that many computations at a time. You're like, yeah, that's next level. It is a badge of honor. It is terrifying when you run into it, especially if you're on a time crunch and don't have the amount of time to fix the problem you didn't know you were going to run into.
Speaker: for sure. So I'm amazed you picked Google Drive. That was like the last thing I had on my list that you would pick a single source of truth on because honestly, I, I hate Google Drive. I love that everybody uses it and it's easy to collaborate in it. And it's just kind of like the standard, but I'm like, dude, I even like 365 more and I like Google Drive because there's more options. I feel like I feel like Google Drive every time I use it, I'm like, it's just a search doesn't work well, which is surprising. It's hard to find things. It's just docs. I'm like, so I'm,
Speaker: I'm all in, man. I got to figure out what you're doing here. I'm guessing you're using shared folders and you have a good file hierarchy system. We've got to follow the structure. And also the, you know, the kind of part of the system that we ended up building was, is to make it easy to find the files. So the goal is get a quick look at what might be in the file, make it searchable as a kind of a spreadsheet, like so that you can narrow down the thing.
Speaker: And then create a place where you can dive in to the specific response that you want to look into, if that's something that you want to go and do. Now, honestly, like that's just a couple of sort of functions of Google Drive. Like I'm sure it could do a lot more and I'm sure I'm missing out on a lot of stuff, but you know, I basically hate Google products, but this one is really effective. Uh, and because the whole team is using it anyway, and like, that's where we do our work. It just makes sense, you know?
Speaker: I thought about building an encoder, first of all. And I decided after trying to build an MVP on a project, I just didn't like it enough to want to use it every day. And that also the team wasn't using it. So it's kind of pointless. I'd just be trying to change a change of behavior for the sake of some technological stuff, which just didn't feel like a good solution.
Speaker: Yeah, we used Coda at a former, the last company I worked for, they were using it as a place. They were also using Google Docs though, so it became this both hand. Interesting. I found Coda to be a lot like Notion, but with, I guess, more Coder or developer functions in it so you can actually automate more of the database. But it couldn't serve well as an Excel sheet. Maybe it did. Maybe it did do sums and all that kind of stuff in it, but... Yeah, I'm not sure.
Speaker: Like it could work. That's the problem with some of these tech. It seems like it should work, but nobody uses it. I think if you would like, if that's what you would using, you'd find a way to make it work, right? Like that's the, that's the kind of mindset that you have to have to approach all of these challenges and solving them with technology automation in particular is like, Oh, okay. There's lots of things I could do.
Speaker: But what should I actually do? Where do I already work? How can I make this system work with the behavior that I am already doing or my team is already doing? And if you're in a CODA environment, then great. Double down on it and make everything in CODA. But yeah, just not for us, basically. Hey, man. Every single system has to have its strengths and weaknesses, right? You have to kind of pick the one where
Speaker: You can kind of live with the cons of it, right? Yeah, for sure. You can share these ones. Yeah. So you have a, you're using Google Drive, you organize it well. It's probably labeled and everything's, everything's cleaned up. That's right. You start to introduce AI. Tell me about that. All right. So the whole process here was not just about the, the kind of, like we could, we could manually get all of this stuff into a Google Drive and, um,
Speaker: and then kind of create the documents. And in fact, when I first joined Podia, I put into place this kind of, I call it a catalog because it's just got all of the kind of all of the data in it for customer research. And the project that I proved this out on as a concept, we were launching email marketing as a part of our platform.
Speaker: And we were doing a huge amount of beta research or beta research, depending on where you're from. And what I was doing is every time a survey result came in, I piped it in with Zapier. We created this record, so we had this system going. But at the end of every week, I would sit down for probably a day and just read all of the responses and summarize them.
Speaker: and say like, what did I learn from these things? And I was like, wow, that's a really high value activity for sure. That provides the majority of the value of customer research is reading and seeing and saying like, oh, this is what I learned from this. But I thought there are parts of even that small process that could be improved using AI, right?
Speaker: Initially, when I started looking into what that might look like, I think I built a little GPT, actually, just in Chat GPT that basically would just like, I'd feed it a survey response and it would pull out these kind of, it would pull out some kind of different summaries and stuff like that. Because actually the first pass of
Speaker: of any kind of survey response or call transcript or anything is actually not like fully understanding it, but just getting a sense of what might be included and the kind of sentiment and all that sort of stuff. And so just applying a small kind of little process there where you said like, you say like, hey, summarize this customer feedback in like eight words.
Speaker: just to give me a title so that I can look down this list of 1 ,000 things and my eyes will pull out the bit that I want. And then you go one deeper and you say, OK, there might be like 50 to 100 responses that are pretty much saying the same thing. So give me the 30 -word version of this. And then you've got all these responses. And then you say, well, look, I also want to get some tags here because I might want to search for all of
Speaker: the responses or customer feedback insights that have kind of talked about a specific feature or a feature request or a bug report or something like that. And so AI is actually really great at that task already. And the task of effectively saying, I think these are the relevant things. I'm going to label them with this.
Speaker: The first step there was just prove that we could do this in a useful, meaningful way on a one -to -one basis. So I'd take the response, put it into the GPT, and say, OK, here are my questions. Can you give me the answers to these questions? Which seems like a really not useful thing. But once you can prove that you can do it a couple of times on a one -to -one basis, and you've got some
Speaker: You've got a high degree of confidence in the output. You can start to say, well, what would it be like to massively cut the effort and also maybe drop the confidence in the response a little bit? What would that be like if we could do that and we could do it thousands of times a day? And so that was how we got started with that kind of AI stuff, really, with the customer research.
Speaker: So you built a custom GPT to dump, dump some data into and see if it could just do one off stuff. Yeah. Can I manually get this GPT to give me the right. Output essentially.
Speaker: Yeah, I mean, look, we didn't even have to use a GPT, to be honest. That was probably overkill in some ways. We could have literally done this with just kind of, you know, a templated kind of set of prompt or whatever that we had created in advance, but that, you know, it was just a sort of a shortcut because I was able to then feed it some information about the company. I dropped in our, like,
Speaker: I dropped in a PDF of our pricing video, which is about nine minutes long, and it talks about all of the different features of the platform. And I was like, oh, that's like enough information. If you don't know what Podia does,
Speaker: nine -minute video script would be enough for you to be able to pretty confidently say, oh, that's for me or that's not. And it's certainly enough for an AI basically to say, oh, yeah, this is related to this feature that I know this product has or whatever. And so that was the kind of initial thing that we did while trying to work this out.
Speaker: So you did it, you tested it, you liked the output. How'd you scale this thing? How'd you hook it up to your drive? The, um, the fun part cause I'm like, I'm trying to figure out how I can replicate this now. Yeah, for sure. The, um, the next phase of this was really to say, okay, well, what process do we already do? The process that we already do has Zapier.
Speaker: heavily involved in it. So with Zapier, what we're doing is we're saying, OK, let's take type form, which is a great, easy example here. When a new type form response comes in, I want you to do these things with it. And the things that we want it to do, we want it to add a record in the spreadsheet that this response has happened. Then we want to do some analysis on that. So we can talk about that in a second.
Speaker: Then we want it to update all of the spreadsheet with this information, this kind of analysis. And then we want to create a permanent record of that as a document. Then we want to store that in the folder structure. And then we want to finally update the spreadsheet row again to say where that permanent record is.
Speaker: So that whole system is like, when you say it, it sounds quite complex already, actually. But when you look at that, you're not doing a huge amount of stuff with that. Yeah, it's not too bad. But you're doing it thousands of times a week. Now, where we put the AI in is we built steps into Zapier, which use chat GPT directly.
Speaker: And that really, I think, was the big unlock here in our system. Previously, it was good. But when we put the AI steps in and we kind of used those in Zapier, that really is when everything went from good to, this is honestly brilliant, and has made our lives 100 times easier.
Speaker: In practice, what that looked like is breaking down each of the questions that I previously would have asked manually as a step. Because in Zapier, you're effectively creating variables with a script. That's like the software equivalent is like you're saying, here's the raw data, do something, and then store it in a variable so I can use it later in my thing.
Speaker: And so we have a variable for the title. So we do that summary of eight words from the original type form response or whatever. Then we do the same thing with the description. And then we categorize it from a list of
Speaker: a list of categories that we created in our catalog sheet. It's like a separate tab on our spreadsheet, which basically says, here are the kind of categories that we use. Chat GPT, tell me which one you think this is most likely to be from this list.
Speaker: And then, you know, there's a tagging functionality in there as well. So like suggest, I don't know, I can't remember how many it is, like suggest eight comma separated hyphenated tags that I can use to summarize this specific thing and put them in a list. And then that's how that all worked on a practical level. And the cool thing about, I think,
Speaker: AI is like, if you think about your actual problems, if you think about the things that you're trying to do, they're probably not that complicated. Most problems in technology are like classification problems or summarization problems.
Speaker: what I think most people, or I won't say most people, that's probably unfair, but a lot of people certainly are doing and seeing AI as and the uses in marketing are like, oh yeah, this is generative. This is creating. I'm creating an image for my social media. But marketing isn't really about social media. There's a lot of
Speaker: There's a lot of, like, organizational, informational processing tasks, and AI is fantastic at that stuff. Even right now.
Speaker: It's just an exciting time to be alive. I've said it so many times, but I'm like, even if AI gets no better from chat GPT four, there will be five years of innovation because it bridges so many gaps we had with just marketing automation. We couldn't do analysis. We couldn't do summaries. We couldn't shorten or lengthen things. We just couldn't do that with.
Speaker: databases and algorithms, not easily, and it wouldn't be good. But now we can do it pretty dang well. Even without any more innovation, I feel like we're only still tapping the surface of what we can get out of AI. Yeah, for sure. One thing you're doing is you're taking all these survey responses and you're running a list of actions for every single survey.
Speaker: and you're creating summaries that you would have done manually before reading every single survey and creating summaries for them. But you said you're getting thousands a week. So that's cool to have analysis on an individual survey times a thousand, but even still are using AI to do analysis across like a weekly cohort. Yeah, it is manual right now after that. Well, it's semi -manual, I'll say, right? So what I actually do is
Speaker: When we're in a research project, so we have two kinds of research at Podia, really. There's like active research, which is like we are researching a specific topic and we are building features or changing messaging or all sorts of different things that we're actually working on at the moment. And then there's like all the time, right? Always on research, which is basically
Speaker: Because we're very kind of customer -focused, whatever that means, we actually end up with this huge amount of just data everywhere. When we're on an active project, which is really where customer research packs a punch at Podia, what I've been doing is creating a report each week. So as you said, the kind of analysis post of those 1 ,000 responses.
Speaker: Because one of the things about this is that we decided to basically put, during these active projects, put the research there in front of everyone's faces. So we have a Slack channel, which gets an update every time one of these pieces comes in. It's very noisy.
Speaker: But it's it's the only thing in that channel, but it means that literally all week I'm seeing small kind of summaries of stuff come in. And so there's like some kind of mental computation there as well. So when I come to do it, my brain has a sense of the data.
Speaker: Oh, that's so helpful though. I mean, a lot of it is, the secret sauce to marketing is really getting live data and information, whether from talking directly to customers or interacting with them on social media, or like this is a great use case of you're just feeding your subconscious what people are freaking caring about. Which informs your gut, helps you make better marketing decisions. I mean, I can imagine that being a superpower because it's going to just make you smarter.
Speaker: I think this is aside from AI, but if you want to level up your marketing career, becoming the person who understands the customer best is certainly the way to do that. And it's one that no one really seems to be that interested in.
Speaker: in many places. So it's like a big opportunity in the kind of like most real sense. But taking that kind of like mental computation and saying like, well, my brain lies to me all the time. I make mistakes on a day -to -day basis. So what if I use AI as like a way to verify what I think? So one way that I've done that
Speaker: in research projects where I'm doing this active research is I will export the database. I'll filter it, and then I'll export the data for a week, for example. And I've got all of these lines of data. Now, let's say I've got 1 ,000 rows. I'm able to upload an Excel spreadsheet, or a comma -separated value spreadsheet.
Speaker: And I can upload that to chatgbt and say, okay, look, here is some data. Firstly, tell me what this data is. Tell me what your kind of interpretation of this data is. What is it?
Speaker: I always ask that step because I think it's a really, really useful qualifying step because if there's a problem and if it doesn't really understand what the heck the data is, my experience is you can end up with subpar results. So I'll say, hey, tell me what the row titles are and what each kind of data in each column is. And then from there, I'm able to say, OK, well, look.
Speaker: I am interested in answering these sorts of questions. This is for a research project that I'm doing. I want to write a report. I want to check that I am understanding the data correctly. So give me please a list of the common user experience friction reports that have come in through this data and where possible site
Speaker: the title, for example, of that report because I've obviously got a title column in my spreadsheet. So that I can go back and check to see whether this is accurate, whether the output of the chat there is actually accurate or whether it's maybe making some stuff up.
Speaker: And once you're able to get this sense of, does chat GPT understand what I'm trying to do here? You could step back a little bit and say, OK, here are the five questions I want to answer in this report. Give me some sense of your interpretation of this data to say whether I'm right or not.
Speaker: And then you can kind of argue with it. You can say, well, here's how I have interpreted this data, like compare and contrast our outputs to see who's right. And it's almost like you've got a coworker there who you're having a debate with. And at the end of the day, you can choose what to put in and stuff. But actually, the process of going through that sort of almost argument with a computer, effectively,
Speaker: is really good for finding out what the themes are in reality. Absolutely. I mean, that's a great use case for AI that I've used quite a few times where I'm sending, maybe I'm writing a proposal for a new client and I'm writing it up and I just copy and paste it and be like,
Speaker: Does this even look good? It's something you would have asked a coworker before in a remote situation, or in my case, being in a solo marketing role. It's just nice to be able to bounce your ideas or your work off of somebody and get some feedback. But in your case, using it as a customer research tool and still having checking the hypothesis and you pushing back on it and it pushing back on you, it probably gets better results.
Speaker: then maybe even some people, because it can do things that some people can't, right? And that's kind of the interesting thing about AI is it's starting to creep up on people in some ways, not in most ways, but in a lot of ways it still can. What are some use cases you're getting out of it now? So you talked about like using it to find common friction points. What are some other things? You're building all this data. You have this Google drive full of amazing, amazing data now. How are you querying it? Yeah.
Speaker: Well, that really depends on the kind of project. When it comes to these active things that I'm doing, it's like a lot of the time it's kind of around features and what features people want. So we might ask, what are the most common feature requests grouped by category or something? So like we might say, we might be looking for like email marketing is a good example.
Speaker: People want more design options, they want more ability to manage segmentation in the platform, or they want stuff around automation.
Speaker: when chatgbt comes back and says like i've got all of these things and you say well okay give me some like very specific examples of things that people have said about uh design options like what do people want to do and also because i don't just want to know like what individual users are like don't tell me about anything that was only mentioned once
Speaker: and try to group together things in a topical way. So don't tell me people want purple buttons and green buttons. Tell me they want more color variations in buttons or something like that. That's almost a silly example. But basically, discovering feature requests is a common one. The other one that I often do is, how do people talk about these
Speaker: How do people talk about these features or how do people talk about Podia? When they mention competitors like why do they mention those competitors? Now there's like a limitation to that in that right now Because of the way that we structure our data like we don't put the whole raw response into a cell so we are having to rely on the summary of the AI, you know from the AI into
Speaker: to kind of tell us a little bit about that and find some information. But we can go off and then pull, because we have a record of it in one place, we can pull the original quotes and stuff like that, put them all into one big doc and paste them back in and stuff like that. So I think there's probably ways that we'll improve that over time.
Speaker: I don't know whether this is technically possible yet or mostly because I haven't investigated it, but I would love to be able to automatically feed that data back into a GPT somehow so that we effectively build up this kind of corpus basically of customer insight. I think there are some challenges around that with how things are currently structured, but you might be able to tell me different.
Speaker: Yeah, you'll have to go through the API. Everything in custom GPTs is manually oriented. You have to copy paste or get it started somehow, or ask it to hit the plugin that goes and then checks appier, then checks, whatever, you know? Yeah. But I mean, I almost think.
Speaker: That's one of the problems with Zapier is it's not like something in HubSpot or Salesforce or some other marketing automation tool where you can then set a delay timer or some kind of trigger. You're like, hey, once a month, go back through all those things and do X. That's one of the nice things about marketing automation features is you can set delay timers that hit it every once in a while. Yeah, for sure. One of the missing things in Zapier.
Speaker: Yeah, I mean, there's a workaround for that, which is to create a spreadsheet of times. And I guess you could use like a timestamp spreadsheet, basically. But I think that's probably stretching it a little too far. You should probably just buy HubSpot at that point.
Speaker: or an AI tool that's doing this. I remember a few episodes ago, I interviewed a CMO who's like tech startup called known well, and this is their thing. They're literally trying to capture all your customer data, pull it in one place and then just deliver the insights and do the summary, the mass summary. And then that just delivers the insights. That's like the problem they've chose to pick on and work on because it is a thing that most people don't do. Like that thing that you were doing and summarizing them all.
Speaker: Like you're one of the few companies I've ever heard actually taking the time to do it. Everybody knows they need to do it. Hardly anybody does it. The ones who do it are spending, know it's valuable. So spend, spend the precious man hours of your time, you know, on doing it and get the rewards of it. But if it were more efficient, more people would do it and it would be, it would be more competitive. They would be, they would be better off.
Speaker: There's actually a product out there called Dovetail, which I'm not sure you've come across yet, but Dovetail is a custom. It is a platform built for what I'm describing. It is a lot more expensive than a $50, you know, tech stack though. Yeah, I'm sure. That you wire together.
Speaker: But one of the things that they do is they're able to identify themes and then build reports from your existing data, which you've already put into the platform. And I think that's a really cool set of features. Honestly, I think if you were investing a huge amount into customer research, a platform like Dovetail would be a great option. I'll have to check it out. Yeah, yeah. But there you go.
Speaker: Cool, man. Is there anything else you've done with AI around this that others should know about?
Speaker: around customer research particularly. I think the other stuff here is that you... I'll give you a tip for call transcripts. There are tools out there like Fireflies that you can use to do Zoom call transcription or whatever platform you end up using.
Speaker: And that forms, for a lot of people, a key part of customer insight. Now, again, because of the way that AI works right now, the token limits are often hit by the transcripts. And so because of the length of the call,
Speaker: So you end up hitting the token limit. And one of the things that I've done at Podia for our system is basically I created a version of this Zap that breaks up all call transcripts into lines.
Speaker: And what we do is we basically summarize small chunks of text in the exact same process that we would summarize a whole survey response. And then we store those all in an array of responses. And so you have multiple versions and loops through this response. And then we use effectively a parent's app to basically summarize the summaries.
Speaker: The reason we do that is like even when you could get, let's say you could get an AI to do like a transcription in like, I don't know, two chunks of text, you are gonna lose some of the interesting nuance of human conversation to a summary which is like broad chunks.
Speaker: But by splitting it into tiny chunks, you actually get a huge number of chances to summarize. And I know like quantitative evidence to back this up, but I believe that the quality of the output from summarizing lines of a transcription is far significant to summarizing the chunks of the transcription.
Speaker: I believe that also the way that this works for a lot of kind of automation and stuff is that they summarize the chunks rather than the lines. And so if you are thinking about doing something like this, try to break those
Speaker: try to break those conversations down line by line or sentence or paragraph or something by paragraph. It's going to cost you a little bit more, but the end result will be better by a lot more than the margin of cost.
Speaker: Your experience is something I've, I've run into before. I use cast magic to summarize our, my podcast and they have a, I find that the, the longer the episode goes, the harder it is for it to create a summary, just the more like, I don't know. It's almost like it has sweet spots. Like for some reason, like if I want to blog posts, I know giving it a 10 minute video, it'll best summarize, right? Versus giving it a 50 minute video and trying to get 5 ,000 words at it. It's better to do it in 10 minute chunks. Don't know why it's just the way the AI is working right now.
Speaker: Yeah, even in cast magic, you can combine multiple episodes and then query against those. Oh, interesting. It doesn't work good. It just, it's just too much data because I'm like running into it. I'm like, huh, I should go back to the guys at cast magic that I just had on this show and be like, Hey, like what if you started summarizing it because they'd probably get better outputs too, because I think you're onto something there because I've run into the limitations of just feeding it too many lines.
Speaker: But if it had shorter and more concise summaries to work off of, it would probably be able to feedback better data, better content, better whatever you needed it to do. So that's really interesting. I'm going to have to start fiddling with that now and seeing if I can just manually do it to prove to them and others that that's the thing and start playing with that. So that's a good tip. Yeah, for sure.
Speaker: So Eric, you made a whole course on this. Where can people find it if they want to, if they need, if they need someone to like walk them through how to set up the Zapier and how to connect it to the whole thing and how you're going through the whole process. Cause I know there's like, there's like a hundred small baby steps that we skipped over that you had to set up in order to do this. Where can they learn more, connect with you and learn more about this course that you have.
Speaker: Sure. Well, you can go to my website, positivehuman .co. That's where you can find out about the course. But if you want to find out all of this stuff, you know, as I learn it, as I go, I share a lot. I'm on LinkedIn. So it's Mark with a C and then Thomas. And you'll find me there. Fantastic. Thanks for joining me today. Thanks, man. It's very, very nice to be here.
Speaker: And for those listening, remember AI won't replace marketers anytime soon. It will just give them superpowers to do more with less like Mark here. So let's level up together one skill at a time.
