Embracing Generative AI in B2B Marketing with Justin Berger

Erin Pearson: Hey everyone! Welcome to today’s episode of Decisions Now. I am your host, Erin Pearson, and I’m very excited to have an old friend joining me today. As our guest, we have Justin Berger, who is the Chief Operating Officer at 7 Knots Digital. Justin, welcome, and thanks for joining.

Justin Berger: Thank you, Erin. Thanks for having me. It’s good to reconnect.

EP: Yeah. So, Justin, can you start by just giving the audience a little bit of background on what is your role at 7 Knots Digital and the type of clients that you’re working with?

JB: Yeah, I’d be happy to. Well, 7 Knots Digital is a digital marketing agency. Our clients overwhelmingly are in the B2B space, primarily actually in the B2B media space. So, these are clients who serve a specific industry, provide them with their news, their events, their networking, all that sort of stuff. And you can imagine in that specific set of industries, audience is really, really crucial. So having really good insight into those audiences, catering to those audiences in a way that’s really specific, all that sort of stuff is really crucial. So, my role is overseeing the operations of our organization, helping to make sure our clients are successful, including the use of technology to get there, running programs, allocating budgets, all the stuff that goes into rolling up our sleeves and getting work done on behalf of our clients.

EP: Yeah, no, I think that’s great. And it’s such an exciting time to be in this space as well. There’s lots happening, lots changing. And you said it very aptly, especially when you’re talking about really any company, but in particular, somebody in the B2B media space, audience is really key to all of these people. That is your bread and butter.

JB: Yeah, that’s absolutely right. And some of our clients are really narrow, specific industries, so it’s really crucial that we understand nuances of how to subdivide those audiences in ways that a ZoomInfo might not subdivide audiences. We tend to get a lot more narrow.

EP: Yeah, well, one of the, or mainly what we’re going to be talking about today is how people are bringing in generative AI and how you guys are using it as an agency as well. But just to kick that off, as a marketing agency, you’re working with many different clients, you’re hearing a lot of different perspectives. So, what is the overall reception that you’re hearing around AI, and in particular, generative AI as it’s come to fruition? Do you hear a lot of fear? Is it excitement? Is it somewhere in between?

JB: I think it’s both. It’s both at the same time for a lot of our clients. Well, first off, certainly it’s something everybody is talking about a whole lot. I think when you ask if there’s fear, I think there is some trepidation for a lot of people. I think there’s a little bit probably more talk than there is practical use of it, right? I think there’s a lot of analysis, a lot of reading news articles, and a little bit less day-to-day usage than you might expect relative to how much everybody’s talking about AI. So, I think there’s some folks who are starting to find some ways to bake it into their products, to bake it into their day-to-day processes. I think there’s a whole lot of people, probably more than those who are actually using it, who have an interest, who are aware, who wonder how it’s going to impact their role, if the robots are going to take over but haven’t quite taken it and applied it to their day-to-day role just yet.

EP: Do you think that the people who are feeling more fear from it are ones who are not actively using it to quite the same extent?

JB: I do. I think that’s a great question, Erin, because I think there’s definitely a little bit of a learning curve. And I think some of that fear tends to dissipate as you do a little bit more of the work and start to understand some really practical use cases of how AI can help you. So, I do think there is a correlation with the people who are sort of most overwhelmed or concerned or maybe feeling a little bit of that fear, perhaps haven’t gone into the tool and just played around a little bit. It’s fun, right? Like beyond as helpful as it is, it’s just interesting and it has its own quirks, and there’s things that you learn and you can have some fun with it as well. So, I tend to find that most people who have worked in there have a little bit more of a positive attitude.

EP: Yeah, I think the other thing that comes into that is when you’re actually the one using it, you get a sense of ‘When would I actually personally want to associate myself with the answer that it gives?’ versus if I just read about how it’s able to produce answers that people might want to associate with. It’s very different. If I wrote a blog, if I wanted my name to be associated with it, I would still want it to be in my voice. So, I find it’s really good at…I like to call it the blank page effect, but I think it’s really good at eliminating that, and it can help you get started, but it’s still not quite there. So, it eliminates some of those tedious things or some things where you just need some extra brainstorming. As Microsoft puts it, it’s your co-pilot. I think it’s really good at that. But you have to use it to know that it can only get you so far.

JB: Yeah, no, that’s exactly right. And it’s a tool, right? It’s working off what you feed it. Oftentimes I find, while super, super helpful, it might take a little bit of massaging and some adjustment to get it to where I want it to. So those fears, again, of the robots taking over, tend to dissipate quickly, because you realize it’s a tool, it’s an accelerator, but it’s not something that exactly just kind of turnkey spits out an end result just quite that easily.

EP: Yeah, so you said that a lot of people aren’t quite using it yet, but for those that you’re working with that are using it, whether it’s internally at your company or it’s within the clients that you’re working with, what are some of those common use cases that they seem to get started with and why do you think that that’s the area that they’re leaning towards first?

JB: Sure. Well, I think both because these are areas that our organization focuses on so much and because ChatGPT and other AI tools are really predisposed to be helpful here. We do a lot of audience work, a lot of audience research, a lot of account universe creation. We might go into ChatGPT and say, hey, we’re looking for all the best brands in, let’s say, B2B construction, right? Who are the accounts and the companies that are building the most buildings in the United States, right? We could have ChatGPT sort of point us in the right direction, give us the URLs of those organizations, help round out an understanding of what the top brands are in that space. And then it’ll also help us speak to those folks, right? Act like a marketer and talk to the people who are in this really, really specific industry, tell me about the personas that are most common there. So, I think, Erin, it’s both because that’s where we do a lot of our work, and we spend a lot of time and effort there, and we can use some of that accelerated impact there, right? We can take some of that effort that used to be manual and human off our plates, but also because the tool is really good at that, right? I know ChatGPT and some of the other tools may be looking at, I think it’s search results up until 2021, right? But if you think about looking at an industry and looking for the top 50 companies by revenue – tends to be fairly accurate, right? If you don’t have the biggest movers in the last year or two, but you’re starting to create something and you’re starting to get a sense of who’s out there, it gets you pretty, pretty close.

EP: Well, I think at the end of the day, if you’re actually asking it for data, whether or not it’s up to date or from 2021, these tools have a strong rate to hallucinate as well, as they put it. So, you have to validate all the data regardless. So even if it did, it’s not always trustworthy data. I know personally, I’ve put in prompts that I didn’t even ask it to come up with data or anything, but it just decided to spit out statistics, and it put in quotes, and it said who these quotes are from. It made up names, and I have no idea where it got any of this. But it sounds good, you can’t use it, but it sounds fine.

JB: Yeah, well, and I think also that goes back to what is it you want to ask these tools, and where do you want it to take your projects? So, I think where you’re looking for something really exacting, where you’re looking for the most up-to-date data, there’s some caveats of where you’re going to use it. But if you’re looking to build a big directional universe of the types of organizations that are representative of a specific industry, it’s really strong at that. If you want to look at the personas that might make sense within that industry, it’s really strong. And then once again, if you prompt it, ask it for some work on personas, and there’s some things you don’t see, you could sort of guide it along and help get it to where you want it to be, right? I’m noticing that this is not present there, right? And you could sort of talk it into getting you something that’s a little bit more well-rounded, and it becomes a teammate, becomes something that’s supporting you versus something that’s doing all the work on its own.

EP: Yeah, no, I think that’s great. And I think the research use case is oftentimes underused or perhaps just under-talked-about amongst many people because people are really associating it with content writing or kind of more job replacement type skills. Cause that’s where, I think, the fear comes in. And so, there’s more attention on those skillsets, but it’s that research that really makes you do your job a lot better that I think are such critical ones as you’re saying as well. And they can really help somebody just get onboarded at a new company, or if you’ve maybe been there for 10 years, but you guys are going after a new industry and you don’t know that much about it, it can just give you a few talking points. So, when you get on a call, you have a perspective to put in, or at least you have questions to ask. So then, when you’re talking to an expert, you can get better answers from it as well. So, I think that that’s a really cool use case. One of the…

JB: Absolutely.

EP: Yeah, one of the things that you had said that you were doing is you’ve been trying it out. You had a persona-based campaign that you were running, and you did an A-B test on it. Can you tell us a little bit about how you came up with that and how it worked?

JB: Yeah, well first off when you just mentioned content writing, I think one of the things that we’ve found really useful and really impactful is to create more variations than humans might be able to create, right? So, can it write things from scratch and replace the writing needs of an organization? Sure, but I think what’s really impactful is: can it do things that perhaps you would never have the scale to be able to get done? And that’s a lot of what we did when we’ve been running some of these A-B tests. So, what we’ve done, Erin, is we’ve taken audiences. We’re primarily using email as a channel to engage audiences for a specific set of campaigns that we’re doing. We’re really big on rich audience data, right? Enriching with everything we can to know the most about these folks that we’re sharing content with and certainly when we can do that and target properly, we’re a lot more efficient, right? We have to, there’s a lot less effort that needs to go in and we’re able to, you know, get the right folks into our funnel the most quickly. So, part of what we did is we built out these really, really robust personas where we might talk about, you know, on a spreadsheet, every single person who we’ll be emailing. What’s their persona or their main job function? What’s their level of seniority? What’s their industry? We actually used a plugin. It was called Numerous.ai, and we had Numerous.ai take a look at all of those data fields we have on an individual. We give it a starting point, right? We’re not starting from scratch, but rather we’re feeding the artificial intelligence something that it can work with. And then we’re saying, ‘Take everybody on our file here and produce a version of this, invite them to this webinar, see if they’re interested in an event, or engage them in a specific piece of content,’ whatever that call to action might be, but ‘make it really specific based on what we know about them, the combination of their seniority, their job function, their industry.’ And what we did is we did an A-B test and always want to use data to validate and make sure the efforts we’re putting in there are fruitful. And what we found was we sent half the audience a generic message that was sort of similar, the call to action was the same. A lot of what we put in there was really similar. And then we sent half the audience the version fed by Numerous.ai and really sort of specific to what we know about that audience. And we saw, I think it was a 63% increase in engagement from the folks who had the more robust version that was really personalized. And I think what’s really cool about it when you think about that personalization aspect is we’ve all been recipients of those messages that are sort of the tokenized version of personalization, right? Insert job title. Hi Justin, I see you’re enjoying your time at 7 Knots Digital, blah blah blah, right? But by using this secondary tool, what we’re doing is actually creating messages that sound a lot more human. It’s not a puzzle where those pieces are dynamically inserted, but rather it’s a separate message and a separate tone of voice that makes perfect sense to each individual based on their combination of values.

EP: So, just a question on that, if you had had multiple companies in there, but let’s just say the persona was the same, let’s just say it was VP of Marketing that you had in there, would it have written the same message for both of them, or would it have adapted and created a different tone for each person?

JB: I think the combination was unique for each and every individual. There were some recurring themes, right? So, it might think about a VP of Marketing and, you know, digital analytics being an area of focus, right? So, it’ll sort of touch on those same areas that the personas are interested in, but you might have two different VPs of Marketing who have the same areas of focus but different industries, right? So, it’s all of these sort of infinite different combinations. But there are certainly some recurring themes, because it is zeroing in on what we know each of those personas cares about most.

EP: Yeah, no, I think that’s really cool. And if I’m not mistaken, Numerous.ai also works in a spreadsheet format. So, it actually makes it quite simple to, whether you’re exporting the data…like if you’re just using ChatGPT directly as an example, you would have to put in each input, I guess individually, or as its own line item. But it really does. Essentially, it’s just an Excel formula, effectively. And it just spits out all the information.

JB: That’s exactly right. That’s exactly right. We’re using Google Docs, and we’re saying take the persona and sell B2 and the company and sell C2 and make that up for it. So, it’s a lot more scalable. I think one of the limitations, especially as you become more dependent on ChatGPT is that you see it can be really, really good, really useful, but sometimes it’ll limit the amount of output it’ll give you in one go around. You commonly will see it cut off after 50 rows or something like that. So, a tool like Numerous.ai allows you to put it into a spreadsheet and you might have thousands of records and be able to enrich them that way.

EP: Yeah, no, that’s really personalization at scale. I think what a lot of people have been talking about is ChatGPT will also get you to a certain extent, but then it’s these plugins or these third-party integrations with it that really can take you that extra mile. So, it’s not just now about using ChatGPT, it’s, are you using the Chrome extensions or the different plugins for it? One example is I found one that will automatically transcribe and summarize YouTube videos as well. So, if you’re trying to go through a video really quickly or see, does it have the answer that you really want in it? I mean, the ad is still playing on this YouTube video before the video even gets there, and it’s already transcribed it and summarized it. So, you can go through that significantly faster than you can click through or listen to the video even on 2x speed. So, I find that that’s also quite helpful.

JB: Really cool. Sure. I love that. It’s almost like Fireflies is recording your own internal meetings, but you could use this tool to look at YouTube and dive in faster than you’d be able to by just watching.

EP: Yeah, I think another common one is just note-taking as well. So, you can have AI integrated into, well, we could have it, we don’t, but we could have it integrated into this podcast right now to have it automatically transcribe and summarize it for us also. But Justin, you actually wrote a really nice article on using ChatGPT in B2B marketing. So, what are some of the tactical advantages that it’s really helping people with and some of the areas that you spoke about in that article?

JB: So yeah, there’s a lot of ways I think that folks can get started with using AI. And I think, like we were talking about before, there’s that fear out there. Not everybody, myself included, is somebody who’s writing code or super, super technical. So, I think it’s really interesting to focus in on what are some of those starting areas where people can get to use ChatGPT without necessarily being an expert or having a ton of experience. We spoke a little bit about building out account-based universe, looking at the top organizations in an industry. I find that that’s a really useful exercise to run through ChatGPT. Another one, when it comes to content, and you were just talking about content and maybe capturing notes from a call, maybe taking a presentation you’ve done for one thing. I think ChatGPT, more than writing something or creating something from scratch, is really good at repositioning something. Or if you’re like me, I love really bite-sized, short, snappy content. Doesn’t usually hit the page that way. So, can you write everything out and then say, hey, ChatGPT, distill this down to the most important aspects of it? Help me shorten it and make it for somebody who has a short attention span, right? Or help make it a little more controversial than it is, or a little more engaging in this way or that way, or appeal to a certain audience. But also, can you take one thing that was created for one specific use case, a specific client presentation, feed it into ChatGPT, and say, hey, actually, I want to take this and make it an article, right? And I think reformatting and taking content and getting more out of it is another use case that I’ve seen that’s really, really interesting.

EP: One of the things that I think that you touched on a bit that is quite good is the whole repurposing different content. Or another one is also for it to summarize what you’ve also written. And I think that especially when you’re launching a new product or you’re trying to put something out to the market, if you can feed it into ChatGPT, whether or not generative AI wrote it, doesn’t matter. It’s just if it summarizes it for you, does it have the takeaways that you’re also intending for it to have? And what’s quite smart is you can also ask it why or why not. Why is it pulling out these as the key points? So, then you can also improve on what you’re doing. Or I think you said it nicely, it can create catchy headlines. So, if you’re writing an article and you want something maybe controversial for Twitter, or I guess X now, and then you want something more professional for LinkedIn, it can help you repurpose your content. So, depending on the audience you have on different channels, you can use the same thing, but you can put it just in different ways and it can really help you do that. Or if one of these is not your strength, maybe some people are really good at more technical writing, whereas other people are quite good at creative writing. So, it can also help you bridge the gap if you don’t have the full crossover of a skill set.

JB: Yeah, that’s really cool. And I like what you’re saying there, right? In that use case, you’re not talking about AI creating the content. It’s almost being an editor or almost grading your content to see if it’s doing what you want it to. And yeah, when you talk about a catchy headline, when you talk about tone of voice or, you know, taking things and adjusting it. I almost use ChatGPT frequently like a thesaurus. Where you’d use a thesaurus on an individual word basis, I might take a subject line or a headline of an article and say, what are some alternate versions of this you might suggest? And what’s interesting is I find, more often than not, there’ll be little nuggets that I like from more than one, right? I’ll say, give me 10 suggestions. I tend to take it not wholesale, but rather take piecemeal, little elements of each of them and plug them together. Again, a really good validation to those who are fearful of the robots taking over, that you’ve got an opportunity and it’s helping get those creative juices flowing. It’s helping to give you options, but it doesn’t need to be a hands-off exercise where it becomes no longer human.

EP: Yeah, no, I agree with you and actually just earlier today I was using it as a thesaurus to just, we were struggling to come up with different words, so I kept trying to type in prompts to say help us think of a word to fill in the blank here. So, as you’re really talking about this with the different companies and as they’re bringing it in, how are you working with people on change management to help them better adopt it within their organizations?

JB: Yeah, I think there’s a few things. One is finding those small wins rather than thinking about it as how are we going to incorporate AI into our business and turn everything on its head and create 200% efficiency. Are there little tasks here and there that maybe you’ve got people doing, maybe you’ve got vendors doing, and you’re associating dollars to it that are possible to do in this way? And I think the more you can get in there and find some small wins, the more your gears start turning and you’re finding ways to use it more. I think another one is working, and this is something we do a whole lot at 7 Knots, is we get together and share our success stories and save our prompts and share those prompts so that others in our organization might be able to replicate them. Especially for somebody starting out new, you might not know exactly how to use the tool. But by having somebody share something and say, you know, Erin, I know you’re creating a lot of headlines for your podcast. This is a use case that I’ve used before. Have you tried asking ChatGPT to provide you alternate versions? Hey, here’s the way I did that another time, and here’s the results I got. So really by sharing those wins and sharing the prompts specifically, it’s a good way to help scale the successes.

EP: No, I like that, especially trying to drill into use cases that people might just have in autopilot for themselves, that they’ve just been, maybe they’ve been doing it for years, and they haven’t even thought, ‘Oh yeah, I can ask ChatGPT or a generative AI tool for help with that,’ because it’s just so ingrained in their own personal workflow. So, I think that’s great. And even if they don’t use it, to the subject line example you were giving earlier, it can help inspire them to come up with a new idea. I’ve done it plenty of times where I’ve asked it to help me write a post on something. I use none of the posts that it wrote, but all of a sudden, it’s inspired something in me and now I can write my own post a lot faster as well.

JB: Yeah. Well, and seeing what you don’t like about it is interesting too, right? Here’s something ChatGPT is spitting out that doesn’t feel right to me. So, I’m going to avoid that tone of voice, and I know what not to do as well. I love that.

EP: Yeah, yeah, that’s great. So, Justin, my last question for you is, we talked a lot about how to use it. So, for somebody who hasn’t used it yet, what is the best way that somebody can actually just learn how to use it?

JB: Well, I think you can go to the tool that you mentioned, Erin, have YouTube videos summarized for you…No, but I think leaning on folks in your own networks, in your specific industries, or your own organization to understand how they’ve seen success is really helpful. If you’re a B2B marketer, you might hear some of the things that I’m talking about in a use case. If you’re working in another field or have another type of role, you might be better to talk to others who’ve gotten practical use cases. But I think a lot of it is boiling what feels like this really big sea change in technology down to something that’s really practical and has a couple of use cases, and I think getting those small wins, once again, is really important. So, encourage people to play around and find some areas…and almost to a one, the folks I know who’ve gone in there actually end up enjoying it and seeing it sort of like a game and getting better and better at it over time as well.

EP: Oh, that’s great, and I think it is a lot of fun to just play around with. I think the first prompt I ever used was asking it to write me a song about something related to work but in the theme of Taylor Swift, and you just have a good time playing around with it.

JB: I love that. We could all use a little melodrama in our work life. I love it.                

EP: Absolutely, yes. Well, Justin, it was great talking to you today. Really enjoyed the different ideas and inputs that you had for all of this.

JB: Yeah, likewise, Erin. Thank you so much for having me.

EP: Thanks!