# Vibe Marketing Tutorial: Zapier MCP + Claude Code & Zapier Agents

Channel: The Boring Marketer
Video: https://www.youtube.com/watch?v=INIvPezgdsM
Duration: 38 min
Language: English
Words: 6569
Transcript page: https://viewrankai.com/tools/youtube-transcript/INIvPezgdsM

---

[0:00] One of my big goals lately has been expanding my audience and figuring out how to use AI to help me automate the boring parts [music] of doing that. LinkedIn is a channel that I've been wanting to grow on for a long time. So, I created a Zapier agent that helps me access multiple tools and skills and even drafts content ready to post and publish if I wanted to, all with one easy to use interface. I'm also going to show you for the more advanced users how you can activate MCP servers in the LLM of your choice. I'm going to show claude code for this example to access Google

[0:36] Analytics data or in fact any data you want from over 8,000 different apps that Zapier connects to. The great part about it is that you can authenticate right into those tools and have one code that you can go and install within your LLM. So, let's go ahead and get started. We'll start with the basics and then we'll move to more advanced. We're going to vibe. We're going to show how to create these agents. I'm going to open

[0:58] up my terminal and show you cloud code. So, this is how we actually work and get this type of stuff done. Okay, here I am within Zapier. So, I am in their agents interface. You can find it by going to agents.zapier.com. I have a pro account uh so I can access 400 activities uh here. But um basically what you're seeing is an agent called a content repurposing agent that uh that I created. Uh what you see is a trigger and this is just the button I press to run this complete process. I've got instructions for the AI to follow. And we're going to read through these so you

[1:41] understand exactly what this agent accomplishes and how much time it can save for you. So you're a content repurposing specialist. Before taking any action, you must load and read content repurposing agent.md from knowledge sources. Now, I'm going to go down to the bottom of the page here, and you'll see this little thing called knowledge sources. So, this is documents and information from your company or instructions and skills that you want an AI to use. I use skills that I've expertly crafted based on, you know, my specific experience in marketing. So, this content repurposing agent skill instructs an AI exactly how to take a piece of content and repurpose

[2:25] it or atomize it for the specific platform I want to utilize it on. So, you can do that to repurpose social assets from a blog post. You can repurpose a blog post into a LinkedIn post, an expost. The the sky is really the limit when it comes to repurposing. And the great thing about AI is that you can train it for each individual use case. and create content a lot quicker that way. It's going to load and read this skill, which is going to give it a step-by-step set of instructions on how to produce what we see as good content for LinkedIn. So, uh, we're asking the

[3:01] agent to load and read it, confirm that they have load and read it. Uh, we want to make sure that each of these instructions are very, very clear for the AI. when it's using tools, when it's using things like external documents or knowledge sources, AI can get kind of overloaded. So, my best recommendation is in your instructions, spell it out exactly. You can go over to the LLM of your choice, whether it's chatpt 5.2, claudus 4.5 to make sure that you get these detailed instructions for your Zapier agent. You can also use their co-pilot feature. Something that I like to do whenever I'm working with AI tools

[3:38] is I play a little game called AI pingpong. So I start with one model, I bounce it over to the next one, and then I bounce it over to another one, and then I come back. So maybe you start with the co-pilot, maybe you create it in chatpt, maybe you pass it into claude for the final steps. That typically helps you to develop those detailed instructions that AI needs. So um after we confirm that we've referenced that knowledge, we're going to proceed step by step. So when the user, me, or whoever's using this Zapier agent, whether it's a team member or a client or something like that, when the user

[4:16] asks to repurpose content, we're going to go and get X posts that have performed and we're going to convert them to LinkedIn posts and we're going to make sure that they're optimized for that channel and for that platform. So, we use a little tool called Appify. And when you go down here to the tools that

[4:34] this agent can use, you'll see it here. Go to Appify, check out the capabilities. Essentially, I use it heavily to go and get data. So, I can access data that is otherwise hard to get into an LLM or into an agent from social channels, from ads libraries, from websites, you name it. people can build and deploy actors there and you can authenticate into Appify uh right from Zapier which is one of the reasons it's pretty handy. I don't need to leave and go do separate setup steps and all that. I can just add a tool. I can

[5:09] select the tool uh that I want to use. Um we're using Ampify in this case. I can log in and I can set up the specifications that I need. So you can configure [clears throat] the tools that you use here. You're going to need to select which account um you are pulling data from or leveraging uh the platform with. Uh boring marketer is me. We're going to search from the actors that

[5:34] I've used recently. And uh here we go. Um the uh the the agent can actually uh decide which ones to use based on the steps in the instructions. So we're going to look at some Twitter posts. We're going to go look at some LinkedIn posts. All right. And uh one important step here is that uh we are going to run this synchronously. Uh I won't get into the details of why um but just something that you should set up if you are trying to access a tool and do a big job like go get a bunch of data or something like that. All right. So um I've

[6:07] authenticated into LinkedIn as well. So if I do want to just automatically create content and post uh I can. Um, other ways that people often work with uh, workflows or agents like this is they like to keep a human in the loop. So, something that you can do here as well is you could authenticate into Slack and every time that one of these pieces of content is created, you could get a Slack update with the information, with the research, with the final piece of content that your team or your client or whoever you're working with can edit and make sure that it's approved and

[6:42] ready to go. Something else I might consider is adding my brand voice into a knowledge source. So, a lot of folks when they work with AI, they've got, you know, a big project or a long chat history that they finally gotten to the point where it's started to sound like them. If you're using Claude Desktop, you can say, "Hey, create a skill." Um, and that skill will help instruct an AI model exactly how to use your brand voice, and it's portable, and you can take it anywhere. The beautiful thing is about these skills is that they're simply a set of instructions or a markdown file that you could use within Claude. You

[7:22] could use it within chat GPT models. You could build gems, whatever that looks like. But we're using it here in Zapier as a knowledge source. I can sync to my Google Drive and pull out all these knowledge files and pull them right in here through OOTH and access the information and the instructions for whatever AI jobs I want to do, which I find to be really, really powerful. So, after you build it out, you can preview your agent and uh we're just going to test it and we're going to get a gauge on the outputs here and uh see what happens live as we're recording. So,

[7:58] here we go. test agent. All right. So, you can see that we triggered the agent. The trigger is just the command for this thing to go and get started with its job. There are various types of triggers that you can use. Um, this is when you decide you want to run it. You can come in here and you already have a system. Think about the manual work that it would take to go and do all of these steps on your own. Think about going and finding, you know, posts that have performed in your niche from creators that you like or something like that. Think about going out and doing

[8:30] all that research. And now what we're doing is letting AI go and handle this for us. So, it's asking for a couple inputs. And I'm going to go ahead and give it a couple inputs.

[8:53] All right, I've went and grabbed my LinkedIn profile URL as well. So, we have two sources that we're going to go and run these Appify tools on to arm our AI model with exactly the kind of information that we need. So, we're going and looking at the different themes, the different topics that we've talked about on social media to load this up with the right context. So, I always sort of uh follow the same framework whenever I'm working with uh AI. I go out and I get a bunch of information. I get a bunch of data. I call this context loading. I get that all into the model. So, that's what

[9:31] we're doing here. And Ampify is a great tool to do that. Um, great thing too is that AI is pretty smart. So, it can help to find out when errors are happening and go and solve it for you. So, let's see if uh if this works here on the second try. This is the nature of like VI coding and creating agents and VI marketing is that you know sometimes you're going to run into errors that you have to use AI to help you solve. I always say LLM is your friend. I keep one open. I'm always screenshotting and copying and pasting and figuring out how to, you know, get through those little

[10:09] road bumps along the way. But, you know, it's always worth it when you get something running and it ends up saving you hours and hours of time and uh performs well uh with whatever marketing channel or or thing that you're trying to attack. Okay, it looks like that action is complete. So, we've gathered information from 42 different posts and now we're also going to go look at some other posts uh to make sure that we have a full set of information available. Okay, so it looks like we have both data sets here within our agent. And look at it work. We've got a full phase one

[10:48] research complete. Uh we've got a bunch of information on different themes that perform around the niches and the topics that I always talk about. cloud code, AI, AI agents, automation, content creation, marketing positioning, and marketing predictions. So, this is really interesting. We've also p pulled it from the channel that we're creating content for, and we have identified the best themes for us to focus on. So, uh it even scored and ranked the potential for each of these topics to go and perform on LinkedIn, which I really

[11:22] like. And uh it helps us to prioritize. Again, whenever I'm working with AI, I cast a wide net. I context load and then I use the model to sift through all the data and help me to exactly prioritize what I want to focus on. So when you're working with things like this, prompts that can be helpful are, hey, analyze all this data and rank the opportunities

[11:44] by loweffort, high potential impact. Typically AI does a good job at looking at these large data sets much faster than we can do manually to help us find out where to focus. Now given you can go and test and create so quickly with a tool like this, you can go and have multiple variations and you can learn and see what works and uh figure out what kind of fits your style or fits your vibe. All right, we have a checkpoint and uh checkpoints are really valuable with AI workflows or agents I find. um they kind of let us pause and keep us in the loop. Now, one of the

[12:18] very important aspects of working with AI to get good results, you also need to have a little bit of taste. Uh know what good looks like and these steps are very valuable if if you're a marketer or a founder and uh you want to provide some feedback along the way. So, it's given some top recommendations. I'm going to

[12:38] go ahead and uh go with number one. And this is just a simple, you know, chatbased interface where I can throw in the uh selections that I make. This is much uh more direct, I feel, creating these agents with the knowledge sources and the tools and the connections than those super crazy complicated node-based workflows that started popping off online uh about 8 months ago. Um, so this is really kind of how things are evolving into more of an agentled world where you're leaning on the LLM and you're arming it with the things that it can go and select and pull from. Now we are defining some steps in this workflow

[13:23] which is or agent uh which is still valuable but uh you know the future is really like leaning on the power of the LLM and giving it what it needs and in my view like that's one of the values of a platform like Zapier is getting what the agent needs is as simple as you know clicking a button and logging into the tool that you're already using and it's contained uh with this environment. So um you know if you're just getting started building agents or if you're within a company that uses something like Zapier or you want to build them for clients I would definitely recommend

[13:56] uh you know using something uh like Zapier. All right so feedback so based on the analysis um the link then audience responds well to educational frameworks business and ROI focus step-by-step breakdowns and personal experience. So, um, let's see. Says it has strong potential for adaption. Let's go.

[14:24] All right. Perfect. So, it's just moving on to the next step here. and it's identified the gaps in this content that it should close while it's creating the content optimized for LinkedIn which I find to be uh really cool. So, you know, this information is actually very valuable. Uh what kind of business context or professional ROI focus is there? Uh we need some longer form content, social proof, professional hook, call to action, visual elements and a lot of these really help um you know content perform well. So, uh, here we go. Here's an optimized LinkedIn postrap. I turned Claude Code into a $443,000 marketing team. Here are the 10

[15:07] skills that replace an entire agency. After building eight months of workflows, I've cracked the code on using Cloud Code for marketing that actually sells. The result, X amount in digital product revenue with a oneperson operation. So, then we talk about the different skills that are valuable and that work. It's got a lot of detailed information in here uh that really adds value, highlights the result, and then has kind of an open loop or a call to action at the end that can help to drive engagement on the platform. Again, these are patterns that are happening out there from top performing posts and we're able to work those here right into

[15:43] our agent output. So, checkpoint draft approval. Um, you know, this LinkedIn post transforms your ex content for a professional audience. It expands each skill with business context. It adds social proof. It includes an engaging hook and a CTA. It follows winning format patterns and it maintains my authentic voice because we went and found content that I've actually created uh in order to uh add that that relevant context. So if I'm ready to publish this, I can say go ahead and publish it and it will go and publish this on my LinkedIn account. So uh I don't even have to leave. I can also kind of vibe

[16:21] on this and I can say hey like yeah adjust the hook. Uh make this sound better. Include this call to action. Maybe I have a website or a free resource that I want to give away to an audience or the audience that I'm trying to build. That's a great opportunity to uh to go and edit and uh make things even stronger. So, taking a step back just to review here, uh, you can create agents pretty quickly, use LLMs or the co-pilot feature. All right, so we've

[16:50] moved past this gap check analysis. Super impressed with what it was able to find and how it's going to optimize the content for LinkedIn. So, let's take a look at what we have. We have a fully optimized draft that is ready to publish if we want to. Now, again, I like to review. I like to make it my own. I like to be the human in the loop. I consider most agents like AI assisted rather than, you know, AI dependent. So rarely if ever am I just like oneshotting something and publishing it onto a platform. Always recommend that you read it and dig in and check to make sure it

[17:25] fits your brand, it's accurate, etc. But look at this. This is really good stuff. Super strong hook. I turned Claude Code into a $443,000 marketing team. I would probably click or engage on that. That's interesting. Um, it's got the social proof element. Here are 10 skills that replace an entire agency. It's going through the frameworks. This is all real valuable information from what it was able to go out and and gather through that uh ampify step. Um, so my takeaway here is that, you know, Zapier agents uh are quite easy to set up and good at leveraging these knowledge sources and these additional tools. I think one of

[18:06] the main values again is just being able to easily and quickly log in and integrate those knowledge sources and those tools right into this environment. So, if the terminal's scary, if if you're just getting started or wherever you are in your journey, this is a much easier way to go and ship an agent and start seeing results than all the other scary stuff that that you might see and that we're about to look at actually uh when I open up my terminal and show you

[18:33] the MCP capabilities here. All right. So, uh that's the agent. We performed content or we created content. If I wanted to publish this again, I could just say publish, but uh I'm going to take some time to edit this and make sure uh it it is exactly uh fitting to my audience and what I want to say out there online. All right, so if you uh if you poke around in Zapier a little bit, you will also find Zapier MCP. Okay, part two, and this is for my more advanced AI users. We're going to talk about Zapier MCP. So, if you log into a Zapier account, you'll see a button that

[19:15] says explore apps. You scroll down a little bit and you will see the MCP option. You click on it and you come to this screen that allows you to create an MCP server. So, I'm going to show you how that works. But first, what is an MCP? So, I'm not a developer. I'm a non-technical person. I've been living in these tools for about eight months now. So, I'm going to give you my best understanding of what an MCP is. An API is uh basically a programming tool or interface that allows you to access data and perform actions in third-party tools. An MCP wraps up that API

[19:55] functionality into a way that you can interact with it in natural language. So it essentially takes an API and allows you to put it into an AI model or an LLM that you can chat with directly and then get the data from it and perform different functions that that model has uh to do things with it. So to create your own MCP server uh you can click new MCP. You select a client and in this case I used Claude code that is my main LLM where I get all my work done. Obviously there's other options that you could uh create. You could use uh cloud desktop. Uh you could

[20:34] use cursor etc depending on what you want to do. But I selected cloud code. Now again we come back to this sort of like ability to ooth into your tools. And what I did was I added Google Analytics 4. No one likes digging through data. No one likes trying to sift through and and find the trends and build charts and graphs and and sheets and things like that. I want to get my

[21:00] Google Analytics data into an AI model. I want to chat with it to understand it and I want to do things with it and I want to do it quickly. So I clicked add tool and again you can create an MCP server and get any data from these tools LLM ready and then activate that MCP within the tool that you're using cloud code in this case. And I mean really like the the opportunities are simply like limited by your imagination. You can search apps, you can think through use cases. Hey, where would uh where would AI make my life easier when working with tools my business is

[21:37] leveraging or relying on or whatever? Like let's say you're using a sauna and you have a big team and you're managing tasks and you want to understand different things people are working on just by chatting away. You can do that. So there were a couple settings that I had to go through here. Um, basically select your account uh and the property and that's pretty much it. Uh, and then

[22:01] just run a report for this property. This is sort of just telling um telling the model uh what I want it to do. So I wanted to go and run reports and access this Google Analytics functionality. So after I can make my configuration, I could add multiple tools if I want, but we're just going to use Google Analytics 4 today. It's a timesaver and a lot of people can relate. We're going to hit

[22:25] connect. And uh it gives you the steps here in order to do this. Um I have Claude Code installed. Some of you may not. They have Cloud Code installation instructions. But I'm just going to copy a command and then I'm going to open up my terminal which is where I work with Cloud Code. Okay. And uh this [clears throat] is where I work with Cloud Code. This is VS Code. If you've heard of cursor, if you've heard of anti-gravity, uh those are based off of VS Code. Essentially, this is where developers get their work done. My view is that the barriers between being a developer, being a

[23:06] business owner, being a marketer are blending and the best marketers and the best builders are using tools like Cloud Code in a uh in a tool that looks kind of like this. So, don't be scared uh of the terminal aspect here. I've got Cloud Code open. I did that by just typing Claude uh into here. I had already

[23:26] installed Claude Code into my terminal. I authenticated in so I'm logged in to Claude Code here. I copy and pasted uh that command uh from from the Zapier screen uh that you saw. And when Cloud Code installs a new MCP server, you need to quit and restart Claude Code to activate it. So I quit the Cloud Code session. I typed slash commandexit and then I typed claude into the terminal again. It reopened and what I did was I hit slash MCP and I looked for the Zapier server and I found it and uh I had to authenticate. So, there was a little button right here where it says

[24:08] Zapier connected and it gave me a link. I clicked the link. It opened it up in my browser. I hit allow and I was authenticated. Now, Cloud Code has access to that MCP server and it can get the data for us. So, let me give you a little demo. Um, use the You know what? Let's do something fun. Let's use Whisper Flow. Let's talk to AI uh since that's kind of the future versus uh the

[24:33] stone age typing that takes too long. Access the Zapier MCP server and pull my Google Analytics data for the last 30 days.

[24:52] All right. So you can see here that our AI model is wanting to use a tool and MCP is just a tool. It's a container on top of this Zapier tool which is connected to Google Analytics 4. So you can see here it says Zapier Google Analytics 4 run a report. It's got all these instructions and I'm just going to

[25:14] say go ahead and proceed. Okay, Cloud Code just did its thing with the Zapier MCP server which is connected to my Google Analytics 4 account and we have real data, real metrics right here that we can learn from, get insights from, and do interesting things with. Right here, I've got an overview for the last 30 days. Uh we've got our monthly totals. So, I did close to 30,000 sessions over the last 30 days, which is pretty good for us. It gives me the page views, the average engagement rate, and then it gives us some like interesting notable patterns here like some spikes, and you can see here on January 4th, it

[25:54] identified that we had a 16x spike in our traffic to our website. That's the day that I think I uh wrote a a viral thread on X about uh skills in in Claude and in Claude code and launched a new product. So that tracks and uh I really like how it kind of like identified why that happened. It it identified a holiday dip, a peak prejan. It identified another spike here. It identified like the viral uh carryover and the settling from that major uh launch and then kind of the return to baseline in the most recent week. So just getting to those patterns and insights quickly is super helpful. I

[26:34] don't have to know how to use Google Analytics either, which is the beautiful thing. So AI can help you de democratize what you can pull and get uh from these tools. And when you can connect them to LLMs easily, that really decreases the barriers for roles and people of all stages, which I think is just a powerful thing that is totally changing the way that people are going to be doing knowledge work with AI. Notice we're not like coding up or vibe coding something crazy here. We're just doing knowledge work, but we're arming a smart agent with the right tools and the right skills to add value and add leverage to

[27:12] our day-to-day tasks and things that we would otherwise do manually. So, we've got our top traffic sources. I get uh most of traffic from from X. Uh almost 14,000 sessions in the last 40 days from X. Super cool. Uh, we're getting some organic Google traffic, some YouTube traffic from, uh, my appearances on Greg Eisenberg's podcast. Shout out Startup ideas podcast. Uh, and then we've got some newsletter traffic, LinkedIn, and hey, chat GPT, here we go. Looks like I need to optimize my GEO a little bit, but always cool to see those LLMs

[27:48] driving some traffic into my business. All right, [snorts] top pages. Clearly, skills was the big winner over the past 30 days. Okay, so summarizes the key insights. Now, what would I do with this information? Um, well, I could go and find out what patterns exist in this data so that I could optimize my website, dig into key traffic channels so that I can improve on them or double down and generate more traffic. I could say, "Hey, Claude Code, um, how do I drive more traffic to my workshops page, for example?" And then I could have it do some research on ways that folks are driving organic traffic to their

[28:28] website. And I could consider, you know, an SEO strategy or an audience growth strategy. And Cloud Code is smart enough to help you with all of that uh stuff. That's VI marketing in my view. And you can create those assets and deploy those assets here just like we did with the agent that we built in Zapier. Okay. So, uh, let's do something fun here and let's use a skill on top of this data. Again, I think skills and MCPs are the most powerful stack in AI and uh, I'm going to show you uh, what I would do with something like this. Uh,

[29:05] let's see. I am a digital marketing consultant and this website is my client. My client is the boring marketer James. I want to create a beautiful report based on as much data that you can pull over the last 30 days that actually shows impactful patterns and insights that I can put into action for my client to deliver them great results. I want you to use the front-end design skill to make this report look stunning, highquality agency output, make me feel

[29:40] like the McKenzie of digital marketers. I really need to wow uh his team and uh you know just make it locally uh so that it's an HTML file that I can open and check out.

[29:58] All right. Now, Cloud Code is going to do its thing again using the data that we pulled into our environment from the Zapier MCP. Big unlock makes it easy to get that information into the models that you're using. It looks like it's going and uh doing some more queries uh within uh our Google an analytics data to really make sure that it has more details versus that you know top level summary. So it's getting all kinds of different data from the last 30 days, sessions, users, page views, engagement rates, uh etc. Let's see what it comes

[30:32] up with. Excellent data. I have device, geographic, landing page, source and media breakdowns. Let me track this project and then invoke the front-end design skill to create a stunning report. This is awesome. Uh back when I was running agencies prior to getting into AI, doing something like this would have taken me forever. Uh, I probably would have tried to export the data from Google Analytics, get it into a spreadsheet, slice and dice the numbers and the metrics, have to create a bunch of custom calculations and things like that. And then my blocker was then how

[31:11] do I make this look good for my client? Um, I'm not a very good designer or anything like that. I'm not very good at PowerPoint. So, uh, you know, creating those client ready deliverables was always like a major pain point for me. And uh if I was still running an agency, I would be using the front-end design skill to wow clients all the time. You can get data so quickly and make beautiful assets that really differentiate you from like other service providers. You could also use this at your job. Like let's say that you work at a startup that's fast growing and you're the founder of the

[31:44] company's like, "Hey, you know, I need a uh I need to know exactly what's happening with traffic this month. You know, create a report that I can share with our board or something like that." You know, typically you'd be stressed. You'd be up all night doing something like that, making sure it's good. You're

[31:59] sending it. You're going back and forth. You're getting feedback. You're trying to make sure the deck is awesome. But, you know, with this, like, you can have all the knowledge and tools and skills uh from the web to pull into this environment to create something truly worldclass in a fraction of the time. Okay, this is pretty cool. I'll create a stunning Mckenzie quality analytics dashboard for the Vibe Marketer. Let me design something that feels like premium strategy consulting output. It's got a a design direction here. Uh editorial lux data storytelling. Think Bloomberg terminal meets Bane presentation. Um I like this. This is cool. If you've ever

[32:37] tried to like vibe code something or, you know, create an output or an artifact with AI, uh maybe you've seen those patterns like purple gradients and emojis and rounded edges and overused fonts and stuff like that that you know make people really feel like, oh, this is definitely like an AI output. Uh this isn't going to work or whatever. Well, that's why I use this uh front-end design skill. This is created by Anthropic. It explicitly instructs an AI model how to create world-class design outputs. And again, you don't have to uh be creating a website. You don't have to be creating a landing page. You can use

[33:14] it to create things that you do day-to-day uh like these reports or, you know, like strategy decks or fundraising decks or, you know, whatever that might be in in your day-to-day role. For my uh non- tech folks out there, we're just starting a local server so that we can pull this report up and take a look at it in our browser without having to deploy it anywhere or do anything like that. So, let's just give it a second to get this thing spun up and then we will see how Claude code did with the MCP data and the front-end design skill. All right, I see it in my browser. I am

[33:53] going to share my screen over there. So for my non- tech folks out there, I am just uh deploying this right now onto a web browser uh so that we can check it out. Uh let's go ahead and open it up. And based on my first impressions, this is definitely uh a wow level output that any client or uh boss or co-orker would be pretty uh blown away blown away by. So, we've got like the title analytics report 30-day performance deep dive. We've got a comprehensive analysis of traffic patterns, user behavior, and growth

[34:33] opportunities for the boring marketer. So, we've got our top level metrics here. Total sessions, page views, unique users, and average engagement. I love how it highlights this opportunity right here in the data. Mobile gap opportunities. So, it looks like our engagement is a little lower on mobile devices. So, opportunity for us to improve our user experience on that website for those visitors. Beautiful chart here. Highlights the viral spike that we saw uh with that 6,500 sessions in one day. uh when it uses uh the front-end design skill, it's designing these to look good. It's also using things that uh can help it make charts and data visualizations uh easily. So,

[35:15] we've got our breakdown of our traffic sources and devices. We've got all of the geographies, so a geographic distribution. Where is our traffic coming from? It even has those little uh flag emojis for each and every country, which is really cool. Nice little touch. Our top performing landing pages uh with engagement, average session duration, users, quality of the pages, which is kind of interesting. I think it gauges that based off of the engagement rate uh per page. So, this might be a little flag like, hey, I've got a page that has a lower engagement rate. How can I go and improve that? And now that you have

[35:55] the data in cloud code, you could go back and ask those questions on top of it. uh thanks to that MCP server key insights all here and then strategic recommendations like this is really good stuff mobile experience optimization double down on Twitter optimize skills conversion rate and uh AI discovery strategy so optimize for those AI search engines like we said it also picked up that our chat GP chat GPT traffic was a little low as we uh as we rightly found as well. So uh that's how you can take data from the tools that you already use, put it into action, find strategic insights by using

[36:36] MCP servers uh in whatever LLM uh that you might be working in. So to wrap this up here, first I want to thank Zapier for sponsoring this video. This is a tool that I like that I use with my team. I wouldn't do this video unless I believed in the company and believed in the things that they're building. So, thank you for sponsoring this. Check out Zapier agents. If you're getting started and you want to get powerful agents that can access knowledge and tools and things like that in one environment, if you're a more advanced user, you got to try to create a Zap year MCP server, get

[37:12] some data in there, and start playing around. I think that some of the most powerful unlocks in AI right now are when you can stack the data from MCPS and use custom skills to design quality outputs. Hope you like this video. Hope you picked up a few useful tips and tricks. We're going to include some uh some downloadable files with this video um so that you have some instructions for how to get going so that you can set this up on your own and uh start doing some buy marketing. Uh best of luck and

[37:42] we'll see you next time.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt