How to use AI Tools for Marketing - 3 ways I use them at B2B SaaS Startups

Speaker A: If you're in B2B SaaS marketing or you're doing any kind of B2B SaaS marketing and you're only using AI to develop content or blog posts or just creating social media, you're leaving serious value on the table. You're probably seeing so many other people talking about it and feeling slightly left behind, you're probably wondering how you can start making the most of it. I mean, where do you start? So in this video and to help you get started, I'm going to show you the three ways that I use the AI tools day to day with when I'm building and running marketing for B2B SaaS startups as a growth marketing consultant and how I just generally use it day to day for myself as well. And for those of you who are new, I'm Neil, I was formerly in it and now I'm a B2B SaaS growth marketing consultant. And this background has actually taught me to work in a more systematic way with marketing so I can help those startups build their marketing infrastructure for long term growth that actually is scalable. Most people actually using AI right now to essentially drop draft quick emails and generating these one off social media posts or writing these one off content pieces. But there's a real difference between using AI for the one off shortcut and uh, one off tasks and using IT as a systematic layer across marketing to improve your marketing workflows and operation, making them more efficient. And we're taking this approach in this video today because I kind of want to just show you guys how I will I go through. What I go through can be used day to day and you can try out some of those use cases yourself. So what's interesting is that marketers and founders actually getting results with AI right now aren't using actual AI to replace thinking or people. You know, the things that uh, people have been scared about, it's not just, it's not been about those things. They're actually using it to accelerate every stage of their marketing workflow. So that includes creation, strategy, analysis. The way we enhance the way we work is where a lot of the edges and essentially what we want to be working towards together is using AI to do the things that a machine can do, right, not what humans can do. So the idea is we're not trying to replace humans, we're trying to replace the work that the machine can do, right, and give that to AI. And then we're supposed to be focusing our time on the areas for example deep strategy and new experiments and that sort of thing on the human capability side of things. And that's where we really shine, where it's a lot of critical thinking involved, but also a lot of creativity. So we're going to quickly go into the first section, which is about that content and copy section. Now before we get into the first section, if you're aspiring marketer and you want the exact AI prompts that I use day to day, I will be releasing what I tend to use especially. And the things with the three use cases that we're going to be talking about today will also be included in that file. It's more like a swipe file, but there'll be a link in the description for you to actually sign up to my mailing list, which allows me to then send you that once it's ready. And of course, if you're a SaaS founder, more technical founder, who wants to build the marketing infrastructure, but you're just not capable at the minute of experimenting with these AI tools and you don't know where to start. You know, I'm putting together a marketing infrastructure workshop, so join the wait list if you're interested, which will also be as a link in the description. So the first way I use AI at the minute is for research, content and copy creation, but also more in a repeatable way. Right. And so four of the core content, content and copy types that are actually used with AI across B2B SaaS marketing right now in Flyte is the messaging and positioning copy. So that's refining the value props, uh, through research and validation, through benefit statements, creating those and creating homepage headlines on copy for those sort of assets. Also the next is sales, email sequences. That's another area so you focus on like outbound, um, prospecting to come up with the copy templates for those. But more so maybe some hooks or some benefit statements based on the icp. So how you would follow up with that specific icp, the next is ad copy assets, of course, which a lot of people probably would use this for. So we've done this with LinkedIn ads, Google Ads, Microsoft ads, which allow me to sort of craft those headlines and descriptions by also funneling a lot of the data and performance data from the existing ads we've created over time. And the last is channel content. So we're talking about LinkedIn posts, YouTube scripts, uh, for example, for this, right? Me feeding exactly my knowledge of what I want to talk about, my thoughts and what I want to do, and then putting it into the, automatically into the structure for this particular episode. And newsletter drafts. Right. So using it for my brand, but also B2B startups brand. By the way, you want to personalize this based on what you're using this for. If you're using this for your startup, then you want to feed the AI with the ICP information, the value proposition and the goal. And I also use it for research which also allows you to find the angles and pain points for those specific ICPs. Because you can, you can develop your ICPs and you can feed everything you've created for your startup and if you haven't already, you can actually start to generate these things slowly. So it's almost like building blocks and go and do the research for some of these specific icps that you're targeting and you would be able to research and write the copy specific to that ICP, which is good. And that should get you about 60 to maybe 70% of the way. There's and as you feed it some review data so feedback, those things will improve over time to give you the seven to eight out uh, of tens essentially. And then the next job is you need to then go and take those assets and put it into a workflow where you get your editor to look at it or you look at it or for example getting the team to run an eye over and make sure everything's worded accordingly. Sounds legit, is factual, has some that you want to add like statistics and stuff based on what you have and then you can take that to market. So don't hit publish straight away. Then you could sort of take this and if you're producing something, you give it the pain points and the desired actions that that person would take in their day to day role and then the desired actions that you want for that specific task that you're trying to say that you want it to do, but then what channel it will run on. Because for example, if it's for LinkedIn ads or Google Ads, you want to give it the framework so that it knows in which direction to produce that asset in. And you want to build a picture over time so you can help it understand as much as possible. So as much of your day to day workflow as possible. And then you would say write three headlines and with VAR as variations with under 150 characters let's say for my LinkedIn ad and be sure to produce those three variations. It's important to know that the output isn't always perfect, but it gets you to a strong 60 to 70% like I mentioned and a draft in minutes and not Hours, that's important. So you would be able to be more efficient and come up with those ideas a bit sooner as well. And so in terms of infrastructure, because that's what we always talk about here, right? You want to position this as part of your messaging and your content infrastructure. So it's a system that will actually ensure that every piece of copy stays on. Brand is speaking to the right audiences, is accurate for the way it's worded, the tones, uh, what you, what triggers you're using and how you execute on it normally, which resembles what you've done in the past. And it allows you to do it at speed. Right. So you'll be able to then just go into it and double check that all the sales points have been hit and that's how you create this repeatable infrastructure based process. Way number two is actually campaign strategy and planning. And here we will technically be using AI as a thought partner or a strategic thought partner before building out your paid campaigns. Let's say that's how I've used it specifically for like Google Ads or like Microsoft ads or LinkedIn ads. And then looking at what you can do as part of your strategy to execute on those campaigns for the, for your business. The type of strategic inputs you ask AI for could be any of the following campaign structure and objective setting, which is the first one. And that's like understanding whether you want to raise awareness and how you would do that or whether you're going for a demand capture strategy more so which campaign type to use for the different platforms that you're considering. If you are going to trying to generate awareness or generate demand. Right. Or capture that demand. The next is like your targeting strategy. So I've used it for refining the audiences, the job titles that continue to do so on a daily basis, refining the company sizes, the intent signals on LinkedIn and um, which ones to consider. If I've missed anything out, what other ones can I consider? But also the next is keyword ideation. So let's say for Google and Microsoft ads, you're unaware of any adjacent keywords or relatable keywords to the seed list that you have. You can generate the seed list in the beginning if you have no idea, especially for your startup. And you can identify those intent clusters behind what you're actually selling and the audience and what they're looking for. And this will allow you to produce more of a more robust strategy because you have a lot of information in front of you from the get go rather than trying to figure this out and consume A lot of that information by yourself, and then trying to come up with a strategy so you can use it to gather that information, but you can also use it to organize that information so that it can tell you what type of strategy it would develop in the very beginning for yourselves. And last one, which I've actually done fairly recently again, was event planning analysis. So if you're potentially trying to go for an event, but you want to understand what the audience breakdown is, use AI to understand and review the past performance of a particular event, like looking at who their attendees lists are, who joined, who didn't, what companies came up, do they fit your icp. And you can, because you've already fed AI with a lot of your ICP information, it will be able to pinpoint whether the event actually has the right people. And this will allow you to plan those future activations. For example, what worked, what to replicate from those events, and how you would allocate budget based on those things. So let's say you have a new campaign that you're trying to create. That's a good area to start with. So new campaign creation. Rather than starting a campaign from scratch, like I was talking about, you, you give the AI the brief of the campaign that you're trying to develop. You give it your icp, your goals, your objectives, the budget range, the platforms that you're considering. You ask it to suggest the campaign architecture in the beginning and then the potential ideas. The point is that it won't actually replace your judgment because that's what ultimately you need to take the output of AI and then you still need to judge that. But what it will do is it will accelerate the strategic thinking phase significantly, so you'll get to the end points faster because you're able to gather that information and think about it critically. So another example for this is like for a new client project, I remember I fed all of the handover files from what was done in the previous phase of the project to through research and ICP creation and coming up with the messaging, the website that was created, the landing pages that were created. For example, in this instance it was Claude. And before feeding it what I thought, I kind of wanted to let it review those documents and ideate on its own and provide me with recommendations on how it would look as a campaign strategy. And funnily enough, actually it was pretty aligned with what I was thinking. This is what separates that sort of infrastructure thinking versus the ad hoc tactics, which is great, but the latter, it will only get you so far. And whereas when you Try to implement it as an infrastructure. Then you're actually using AI to build a repeatable, uh, workflow so that you can make your day to day more efficient. So you'll be creating a repeatable campaign planning process here when you do it for these things and you're not just filling in, let's say, a brief template once with the same knowledge you actually had earlier, and you're actually allowing it to give you new ideas as well. Right. So way three, the final way that I am going to talk about and what I've used it for is actually performance analysis and optimization. So over here we're talking about how to use AI to analyze campaign and uh, marketing performance data. This is actually a use case a lot of marketers aren't using and a lot of people actually leaving this piece of value on the table because you can do so much by analyzing data. How many times have you gone into your dashboards, into specific tools, have looked at reports, pulled it into sheets, try to run pivot tables, try to generate charts all yourself? Why not use something that can crunch data, uh, very quickly for you, to your advantage? And this is precisely how I've been using it. It actually has saved me, like I would probably say tens of hours on single workflows or single tasks I was doing before in analyzing those specific sets of data. So what it looks like when we're actually doing it day to day is things like pasting campaign performance Data like your CTR, your CPCs, conversion rates, cost per leads, Interclaud or ChatGPT, if you're using that and asking it to actually identify patterns, identify specific anomalies, uh, throughout your campaigns and what it would suggest based on those things and based on what, what it's understood in terms of the context of your campaign itself. And then next you would actually ask AI to compare the performance across the different campaigns, but also the different audience segments and the ad variations. So you can go from campaign level right down to the ad variation level and you can surface insights that way. And then once it's understood a lot of that data, once it's gone through and crunched all that data and it's gleaned its own insights, you would actually ask it to write up some sort of performance summary for you to understand. So what it would include about the recommendations that it would provide you with what issues it found, and then you can share this with stakeholders and clients. Once you've cleaned things up on the report and got it into a format that you really need, that's something that I've done on weekly sessions before on taking things to marketing syncs. And there's different ways that this could look. For example, I've used it for an event performance analysis where we were feeding it attendee data like I was talking about earlier, but we looked at the engagement metrics as well of the particular sessions that of our past, how much pipeline was generated from the event. And we would look at building a workflow where you would feed AI exactly what you've done at the event, the budgets, that sort of thing. And then you get AI to build that sort of post event review, right, which you would always run so that you can take it to the overall go to market team and that will allow you to plan for next time and reallocate your budgets based on the type of events that you're actually wanting to attend. What I've used it for weekly now is I'd say Here is my LinkedIn Campaign Performance Data from April. Let's say I want to run a monthly analysis, identify which ad sets are actually underperforming and which are performing well. Um, and provide me with your analysis and recommendations based on what you find in the data. And so the output becomes a foundation for a structured optimization cycle. So you'll have a weekly optimization cycle and then you're doing 30 day optimization cycle and so forth for the quarter and then you would optimize your campaigns accordingly. Right. So it's not a one time dump of data and then that's it. And then you walk away with AI giving you the insights, but it's actually a repeatable process that you can use. And so you have your very own assistant that actually works with you in your day to day workflow of hey, it's Monday and I've got this specific prompt by copy paste or and I've already uploaded the information to the specific folder that I have, so the CSV folder. I've got my audience breakdown, I've got my ad, uh, creative performance breakdown, and I've got my campaign performance breakdown. Uh, could you analyze in this order and then could you give me an output? And this is specifically how you can make the most out of creating a repeatable process with AI to help you do these things much faster. That sort of stuff takes you hours and hours to sometimes crunch through, especially when there's a lot of different insights and you've got to be willing to critically think at the same time, which is a lot to ask from just one person. So imagine you can process that information and analyze it and Then you can be critically thinking about the key recommendations that it's provided or the key insights that it's actually taken out from all of that data. Another example I actually use it for is PPC search ads and understanding what keywords are uh, underperforming what the search term reports are showing, which ones are showing less relevant terms versus highly relevant. And it tells me exactly what keywords I need to kill. Uh, I had to validate the keywords, for example, um, that I was going to kill and I wanted to make sure I was correct. So I wanted to go and find out exactly whether the search intent behind those terms were actually right and what the definition it found was through some research. So I'd asked Claude to actually help me go through that information and then find out exactly what these keywords mean. And then I went and cross referenced that with some searches on Google, let's say, and on Microsoft to see if the search results and intent type was different. And then based on that I got TikTok recommendations from AI and I was able to make a decision on killing those specific keywords much faster, but much more critically because I actually had some more information armed, um, with my thought, into my thought process to help me make that decision. And it allowed me to actually explore other angles too so that I can critically think about these campaigns. This gave me so much of my hours back because you don't have to crunch through the data yourself and you can just start and execute much faster in getting the answer that you need. But I was able to actually improve the PPC search ad over time and it actually has helped me cut tons of budget down to the point where we were spending 2 to 3k less on those specific campaigns simply because we're able to find out those things and make a decision much faster. And from the hours save point of view, you'd be also saving a lot of money in terms of resources as well because you're going from analysis to execution within a couple of days instead of the one to two week period where you would have put things together and got sign off and that sort of thing. So you have to come to the table with loads of insights much faster and come up with a decision sooner. And so the reason why I wanted to talk about this third way was of actually performance analysis and optimization is because from an infrastructure standpoint, which is where we always go back to, this closes the loop on your marketing infrastructure, right? Or it keeps your flywheel going, essentially you create, you plan, you execute, you analyze, you optimize after review. And AI is essentially embedded at every stage of this journey, so and not just the beginning. And this is how you would actually accelerate your management of that operational workflow that you actually implemented beforehand with just humans. But now you've deployed AI in those very areas that AI a machine can handle much more thoroughly, much more quickly compared to a human. And then you can focus on the human task. So I've got three practical tips for you to action this week. I'm going to go through each of these now. Tip one is to actually start with the use case document closest to your immediate need. So what that means is if you have a campaign launching, start with a campaign strategy, right? If you need content, start there, right? If don't try to implement all three that I've given you at once. The next tip on actually ironing these out and making use of AI more effectively this week is to actually give AI more context because the quality of your output through feedback is improved and it's directly proportional to the quality output of the of your brief, let's say. And this means you need to include things like your icp, your goal, the platforms you're using and any constraints every time you are actually using AI. And the last tip I can actually give you is to get access to that swipe file I was talking about by signing up to a newsletter and getting on the mailing list so that when it drops you have access to it. And this will have sort of exact prompt structures that I've messed with already and I've documented. But I will be also updating this over time. But also you'll get information to the AI workflows that are being set up in the day to day environment from a growth marketing perspective and you don't have to start from scratch. Also, another use case is using AI in advertising. But there can be a lot of pitfalls, especially if you're not using AI to the maximum. So I had a very interesting chat with Brooke shepherd where who's a marketing agency owner and they're doing some very interesting stuff there. So you can find out exactly what you should be avoiding and how to use AI more effectively more efficiently by watching this chat here, which you might find interesting. So otherwise, thank you guys for watching. Um, we'll see you guys in the next episode. Take it easy and see you soon.