So What header image

So What? How to Identify Marketing Opportunities

So What? Marketing Analytics and Insights Live

airs every Thursday at 1 pm EST.

You can watch on YouTube Live. Be sure to subscribe and follow so you never miss an episode!


In this week’s episode of So What? Katie, Chris, and John walk through the best ways to use generative AI to find marketing opportunities in your data. And in this economy, we can all use the extra biz.

Catch the replay here:

So What? How to Identify Marketing Opportunities With Generative AI


In this episode you’ll learn: 

  • How to use the 5P framework to gather requirements
  • How to extract insights from generative AI using your ICP
  • How to identify marketing opportunities from your own data sets

Upcoming Episodes:

  • TBD

Have a question or topic you’d like to see us cover? Reach out here:

Please note the following transcript is AI-generated and may not be entirely accurate:

Katie Robbert 0:27
Well, hey, how are you everyone? Happy Thursday. Welcome to so what their marketing analytics and insights live show once again, I only know it’s Thursday because we’re doing this. Otherwise the days all blur together, how are you guys?

Christopher Penn 0:40
doing good living in the tundra here

John Wall 0:43
which is March. Here we go?

Katie Robbert 0:45
Well, you know, because it was the first day of spring. So now you get snow.

John Wall 0:50
exactly the scary is crying in their beer, unfortunately, oh, well,

Katie Robbert 0:56
let’s do something we’re not going to cry about we are going to talk about how to identify marketing opportunities. So today we’re going to use the five p framework together our requirements, we’re going to extract insights from generative AI using our ideal customer profile. And we’ll go over what that means. And then how to identify marketing opportunities from your data sets. And so this came about because we’ve been doing a webinar series on generative AI for marketers, and you know, we’ve done it a couple of times. And now we want to start addressing other needs in the marketplace from our customers. We asked them a couple weeks ago, what kind of content would you want to see, and they gave us a list of different they voted on different categories, different verticals, so real estate, technology, education, ecommerce, so on so forth. And that’s the data that we’re starting with, that’s our starting place for figuring out what are our marketing opportunities? What do we do next? And so you can see, people want to know about, would they want content about general consulting, they want content about education, and so on, so forth. And so that’s where we’re starting. So Chris, we said, we wanted to do this on the live stream, because we wanted to walk through our process of how we’re going to take this data, and actually turn it into something valuable for our audience.

Christopher Penn 2:23
so the first part we have to look at is, in the data itself, what do we see? Right? So agencies is consulting is by far the leading, followed by sort of creative, then higher ed, and all education, ecommerce, and financial services. So if we’re going to create some content for a vertical, these would be the verticals to serve. Now, here’s the important question this, this is statistically useful information. But does it align with essentially the cost the kinds of customers we want? And that’s a question that, you know, to what you were saying, Katie, when we think about the five P’s, the first one is, all of these companies, which ones are our customers, which ones are best suited for us to serve? Because we could I mean, we certainly could make content for the construction industry. I mean, other than having a hat, I don’t really have any experience with the, with the construction industry. So I don’t know that even a real hat. It is actually, this is an OSHA approved safety hat that has been displaced by the way. OSHA regulations say they must include full helmets no longer just hard hats and worksites. I found that out because it was on sale on Amazon. But we don’t have a lot of background in construction compared to other industries. So the first question is, of the, say, top five, or even the top 10? Which ones as the CEO would you want? So here’s the five Ps, that’s, that’s a big part is that purpose?

Katie Robbert 3:58
Right? And that’s exactly it. So the question is, what is the purpose of us doing this exercise at all. And so the purpose of this exercise is for us to reach our different audience segments with valuable information that shows them what’s possible, so that they then in turn, hire us or purchase or subscribe to something from us take an action that leads to some sort of dollar exchange, you know, being fully transparent. It’s all about the money. But it’s all about also making sure that we are reaching the right audiences at the right time for what they want. And so, you know, agencies, agency consulting, like it’s a big topic and so what I would want to do if we were to go that route, is I would try to dig into what is that mean? Because there’s a lot of information there like a lot of different directions whereas you have other CAD glories like consumer electronics, or E commerce online retail or financial services, banking insurance that are a little more specific. And so even these ones are still, you know, their big topics. And so the first thing that we want to do is figure out, you know, how much of our current customer base falls into that category. So we would look in our CRM to say, you know, do we have enough prospects, that we could say, Hey, we’re putting together a webinar series, based on your industry? Would you like to come and invite them in personally, I can pick them. The other thing we could do is say, well, we don’t have a lot of those folks. So let’s, you know, put it out on our newsletter or on LinkedIn and try to capture some of that segment.

Christopher Penn 5:47
Exactly. And for folks who are wondering how do you get this information to begin with, here’s how we did it. So we send out a dedicated email. And you’ll note, on the right hand side is what the customer would see in the inbox. So we sent this obviously, to our entire mailing list. And on the left hand side, you can see the back end code, and the back end code is very simple. It’s just goes to a single thank you page on our website. And what we do is we have each response tag with a an arbitrary, what’s called a query parameter, it’s a text string, that for each of these responses, has pushes that into our web analytics. So this, this poll, when someone taps on any of these responses, it then sends it to our website, that little tracking code goes into Google Analytics. And then what we do is we go into Google Analytics. And we export all the data that has this URL with these tracking codes. And just use a us a little bit of just basic, basic addition to add up all those responses and tabulate them. So if you wanted to do this for yourself, for your company, your mailing list, this is the format to do it super straightforward, very easy. It uses Google Analytics. So you don’t have any additional serving software to set up nothing to buy is very, very helpful. So we’re not going to I’m not going to pull it up in our CRM on a live stream for a variety of reasons. But that is that is would be indeed the next step. And if we look in our CRM software, what we do see is that agencies and consulting does occupy a substantial percentage of what our customer base is, in terms of the companies we we typically work with.

Katie Robbert 7:28
Right? And so that’s sort of what we’re trying to understand first, because we there’s a difference between who’s subscribed to our email newsletter, and who’s actually a sales qualified lead, you know, so they fall into different categories in our funnel. So we want to see first, like, where do we align? Do we have enough prospects or sales qualified leads first, to even, you know, create some of these categories? Exactly.

Christopher Penn 7:56
So if we look in this is an old snapshot, but this is an older snapshot from our Hubspot instance of the companies identified by revenue by revenue earned. This is from 2021. So it’s, I did not have a chance to run this before today’s live stream. But you can definitely see there’s a different set of companies in here computer software, marketing, advertising, public relations, communications, that would all fall into the the agencies is consulting, hospital health care telecom. So there’s, there is a difference between who we’ve earned revenue from versus who is on our list today. So that will be part of the intersection with some of the things that you do see, do you see the agencies of consulting, we’d have earned revenue from companies like that, so that would probably be a good first place to start.

Katie Robbert 8:45
So given that our initial idea around this was to focus on predictive forecasting, I would say we would look at the agencies and consulting as a broad category, and then the different kinds of agencies, agencies and consulting, who’ve been customers in the past or prospects. And the methodology that we were going to do was to take a look then at SEO software, and start to look at you know, what keywords what topics, you know, exist for these verticals?

Christopher Penn 9:20
Yep. Okay. So if we were to go into so this is the H refs of software package, I just typed an AI for and I specify specified matching terms. Let’s see. Let’s do phrase match. To make it a little more specific AI for teachers AI for writing AI Ferrybridge is AI for resume AI for Excel AI for logon for business logo design. So there is there are a variety of different terms and terminology and let’s do we’ll take a look here. Image generator, business. AI I stock trading AI for teachers as the parent category AI and graphic design. So there’s a lot of options here in terms of sort of the general purposes that people looking for, when you might want to potentially narrow down to two other ways people might type this in.

Katie Robbert 10:17
Well, so why are we focused on AI for? So if we’re looking at agencies, and consulting, we’re immediately focused on AI.

Christopher Penn 10:27
Well, so what is what is the content we want to create? Because we want to know what those people are searching, probably searching for, we will try and for at least a broad set of terms. So if we want, do we want to do webinar content about AI for agencies.

Katie Robbert 10:41
so originally, so our purpose was to do predictive fork, yeah, predictive, okay.

Christopher Penn 10:55
Let’s see what we get just, just in general on those terms. Weather forecasts time series forecasting. People would have done some strange things to Google, Japan forecasting, financial forecasting, and so on and so forth.

Katie Robbert 11:10
I was gonna say, so I believe the idea and you know, correct me if I’m wrong, because you know, we are, you know, for anyone watching this is us live, workshopping. You know, what our process is for finding these marketing opportunities. The idea was that we would do a predictive forecast on the keyword terms that that vertical wants to rank for. So not AI for or not predictive analytics, but we would do a predictive forecast for all of the, you know, keywords and topics that they would then want to be ranking for. Gotcha.

Christopher Penn 11:48
So for their agency, so like PR agency, for example, karate. For that, okay, let’s take a look at PR firm. We’ll use the old classics here. Yeah. All right. So we’ve got a relatively small volume. So PR agency, PR firm. Let’s take those to do a quick export here, only selected. So this is going to give us the overall search volumes, we are then going to need to actually extract those terms themselves. So let’s go and put a keywords folder.

Let me grab agents keywords there while I’m doing this. You to talk amongst yourselves.

Katie Robbert 12:45
So John, I mean, you are on the front lines, talking with prospects talking with a lot of audience members, are you hearing that there’s an appetite for predictive forecasting?

John Wall 13:01
Everybody wants predictive forecasting, right, everybody feels that they’re gonna get an edge there. But then the problem you find is they usually don’t have the data, right, you need to have the data ahead of time, and they don’t have the feedback from their customers. Showing off that surveying tool is huge, like anytime you can set up something like that, where you’re continually harvesting info and getting insight from your customers, without having to talk to them one at a time, is just a huge competitive advantage, and where you want to go so. And yeah, you know, right now, in the business cycle, predictive is low on the list. For people, it’s a nice to have, you know, for most companies, they know that it’s a gamble, you know, they’re going to build it and run it. And they’re going to hope that they get some actionable info that delivers a return, but there’s no guarantee on that, you know, you have to really be kind of on the cutting edge and not afraid to take a little bit of risk. But the other thing is, the risk is there. And it’s huge. And so many companies just do this as stream of consciousness, right? Like the CEO, or some other Rando employees on the team, come up with some ideas, and they go to market with that. And we all know that the odds for failure are huge in this. So to be able to actually have some data and you know, if you can double or triple your chance of success, by acting on real data, it can make a huge difference in your ability to survive through these down cycles, or to become, you know, hugely profitable. Because the big thing is, if you can use this to find a niche that everybody else is not in, you know, it’s just like all the other things we see with SEO, where everybody’s just kind of doing the same five things. And it’s the companies that come up with something new and out of the box that they have these huge returns until the rest of the world catches up to it.

Christopher Penn 14:46
Let’s do this, as promised with generative AI. So we’re gonna go log into ChatGPT here, and we’re gonna give ChatGPT So a forecast So in this case, I’m going to take and this, I’ll show you how to do this, if you go to Google Trends, this is, there’s a more efficient way to do this for a lot of keywords. But since we’re just doing a sort of a toy version, type in the trend that you want, in this case, we looked for our keyword tool, we know PR agencies kind of the term, we’re going to look at five years worth of data. So there’s what this search term looks like over the last five years. One of the cautions with the Google Trends data is it is always relative, which means that you’re never getting actual keyword advice, you’re just getting the inferred relative terms, and you have to decide what market you’re what geography you’re in, you might be worldwide. I’m going to use the United States here, because that’s easier. Let’s take this turns. Okay, today, we’re going to do some predictive forecasting, specifically, time series forecasting. Let’s do this. Let’s take out the data file, let’s say, What do you know about best practices for time series forecasting? What we’re doing here is we’re going to we’re priming the model, essentially, we are asking it to build us a really big prompt, in some ways, because one of the things that’s true about transformers, which is what ChatGPT and its brethren are, is that when they predict things, they take into account, what they’ve what has already happened in the conversation, whether you did it as the user, or whether the machine did it, we talked about this on our recent webinar, which you can go over. I think it’s in the in the, in the newsletter this week, if you wanted to catch the replay of the webinar. Because I’m asking it for what it knows about best practices in time series forecasting, I would have to write all this out in a prompt, this would suck, right? Instead, I’ve had it write this out and say, This is great. Now, let me give you a file of a time series with a date and a value. The date is column, DS, and the value is column. Y, we want to forecast ahead using the most statistically valid method probably something like ARIMA the next year. Now we can attach that file. And this is right out of Google Trends. And it’s going to do some thinking it’s going to take a look at the file, analyze it and then start writing the Python code that’s necessary to do the time series forecast. This may or may not blow up because ChatGPT had a rough morning, I was talking to this one he had had a rough morning. I don’t think had enough coffee. But this is what you would do. If you don’t have a because one of the questions that you’ve always had Katie in the past is, what does a person do? Who does not have a data scientist on staff?

Katie Robbert 18:09
Right? Or, or a developer? Exactly. Now?

Christopher Penn 18:13
ChatGPT, you’ve got one on staff.

Katie Robbert 18:20
But it’s still just giving me code? I still have to do something with the code, right?

Christopher Penn 18:26
No, this oh, this is using the advanced data analysis module.

Katie Robbert 18:29
Oh okay.

Christopher Penn 18:32
running. Yeah. So it’s going to write the code and run it. It says this is just the time to use a stationary winded up the difference that before fitting the ARIMA model, that’s still good. Would the average person have known to do a stationarity test on this data? Probably not. No, because and if you just did this, if you just loaded this in as is and said, forecast this, it wouldn’t have done it either. The difference is when we primed it with the with its best practices. Now, it took that into account all those best practices, it took into account as starts writing its code. So it it by having that initial write your own prompt question is going to generate a much more valid result.

Katie Robbert 19:22
All right, so is it. Okay, it’s still thinking. So in terms of us finding marketing opportunities, you know, the purpose of today’s episode was for us to take a look at the data that we’ve collected, and figure out what can we do with it? How can we turn it into something valuable? And maybe predictive forecasting isn’t going to be the thing, you know, so we could do our own, you know, internal, you know, SEO research, keyword research, forecasting, and then just start creating content. But what we wanted our purpose was something a little bit more interactive than style. At a content that we write, and then put out on our blog or on social, we wanted to really engage people. So I think that’s part of, you know, our purpose statement is we want to create content that is engaging. And for us that felt like we could do a webinar to introduce, you know, our predictive forecasting services, our subscription. For those who are watching, you’re hearing it here, we actually did roll out our predictive forecasting subscription, the day before, everything shut down for COVID. So timing wise, it was poor, but we did not know. Ironically, we could not predict that that was going to happen, we just didn’t know nobody knew. That was what they would call a black swan event. But now, four years later, we want to reroll it out. And so the idea was that we would start with a specific industry a vertical, whether it’s agency owners, or other consultants, and walk them through, here’s the power of using a predictive forecast. You know, here’s what you get, here’s how you use it, here’s the results. And now you can subscribe to your very own forecast, that’s updated once a quarter, if you’re interested in learning more about that. Please feel free to contact us Trust Insights at AI slash contact. You know, so that’s something that we’ll be rolling out soon. What we’ll be doing is everything that Chris is doing now is what we would be doing behind the scenes, but we felt like it was a good idea just to show what’s in the box, what it is that we’re actually doing.

Christopher Penn 21:41
And what’s in the box from what ChatGPT ChatGPT just did was, it didn’t do a great job. And it didn’t do a great job. Because the initial data itself isn’t robust enough. There’s there’s very little variance in this data, to the point where, yeah, it’s not great. It also crashed as it was. On and this is the cautionary tale about using generative AI. This specific task, believe it or not, is a poor task progenitor of AI. The reason it is, is that language models are good at language, language models are not good at math, this is a math task. This is straight up statistics. And there are limitations even with the best tools like this, that no Matt, no amount of prompting is going to fix because they just don’t have those capacities. The way that if I was starting from scratch today, I would do something similar to this. But I haven’t actually be writing the Python code, because that’s the most reliable way to get working, working data out of these tools. That doesn’t require you to basically hit regenerate over and over again, in the hopes that ChatGPT gets it correct this time around. And this is true for ChatGPT clawed Gemini, there is no one tool on the market that does this amazingly well. They all do cogeneration reasonably well. But data analysis is not their thing. So the next step from for, you know if this was your company, and you were looking for Okay, well, when is PR agency as a term going to trend? Is it trends, typically, at least based on this data? Believe it or not in the next four weeks, that is when it’s it, you see the highest bias spike in this this very dense data set. And if that was the case, then you as a marketer would be like, Okay, well, we should get some kind of campaigns in the air sooner rather than later. Because this is this is going to happen very, very quickly, we need to be in the market. If we’re not, then the results we get from the rest of the year won’t be as robust.

Katie Robbert 23:58
Well, and one of the one of the ways that you can use generative AI is if you have that forecast, you can use generative AI to help come up with ideas for what those more robust topics would be what those campaigns would be. And that’s where the ideal customer profile fits in. So a couple of weeks ago, and you can watch the episode on our YouTube channel, trust We did a live stream about putting together your ideal customer profile using your data and generative AI. And so once you have your forecast, you can say, okay, great. These are the key words that I want to be ranking for. Now, what the heck do I do with this thing? So you have your customer profile? And you can start to use generative AI to say, hey, generative AI, this is my customer profile. What do they care about? What do they want to know? How do they want to be reached? What kind of campaigns, should I be creating around this set of keywords for this profile? And that’s where that’s sort of the advanced version of using these predictive forecasts. So Chris, are we able to do like a quick demo of what that could look like?

Christopher Penn 25:14
Yeah, so let’s do, let’s feed in an ideal customer profile, we’re gonna say today, we’re going to design a digital marketing campaign. What channel do you want to do it on?

Katie Robbert 25:33
Let’s do LinkedIn

Christopher Penn 25:34
That’s a good choice, but unlinked in to appeal to our ideal customer profile. Let me give you this ICP to begin with. So you understand our company and what our ideal customer wants. Now add in our separators, put in that entire gigantic, huge profile. And let’s see what what first it will it will summarize. While it’s doing I’m gonna say next. Well, let’s generate five campaign ideas for our LinkedIn account in in bullet point format, that would appeal to our ICP. So campaign goals, awareness, lead generation engagement, pillars, content, targeting content, strategy, thought leadership, and so on, so forth. Here’s your link, overall LinkedIn strategy, key messaging, again, this is all from the ICP. So let’s see what it has to say. I should probably clarify. While it’s so it’s going to come up with a list of ideas, I can guarantee the ideas are going to be boring. The digital transformation readiness assessment, solve a common pain point series, mini case study showcase. Did you know industry insights, and the future of whatever webinar was the first five ideas?

Katie Robbert 27:09
They’re not bad. I mean, yes, fairly generic. But if you asked me for campaigns, I would not have come up with those necessarily, I would have said, Okay, well, let’s do. Oh, I mean, maybe a series of some kind. But I do like a little bit of the specificity here. So I would say they’re not bad. They’re good starting points. And I think that that sort of the caveat is they’re starting points. They’re right. They’re sort of the kernels of ideas.

Christopher Penn 27:38
So what I’m gonna do is, I’m going to tell it, I’m going to feed it back to itself, literally saying this is a list of bad ideas.

Katie Robbert 27:48
Do you think they’re terrible? John?

John Wall 27:50
The pain one is solid. And then you know, the future forward one is always works. But yeah, I mean, it’s on the generic side, this will be funny to see.

Christopher Penn 27:59
Show us your ugly. Bad data practices, show off your bets a myth busting debate, I like I tend to like myth busting five myths holding back your healthcare transformation, the real world transformation, Journey competitive audit, raffle, and a LinkedIn group takeover. So that’s different. That’s, that is a bit more specific and a little bit less, a little bit less generic by giving it a negative prompt.

Katie Robbert 28:25
I kind of like the show as your ugly because I feel like people would at least stop to figure out what does that mean? Like, but again, this is all because we did the work to put together our ideal customer profile, which we’re not going to show all of that on this live stream. But we really tried to dig deep into the pain points. And so ideally, you have that customer profile with those pain points, goals, motivators for demographics, that are going to help prime this model to say this is what this person cares about. Help us now match this to our forecast.

Christopher Penn 29:05
Yep. So one of the things that we forecast for ourselves that people love to talk about is digital transformations. It’s Yes, everyone loves that buzzword. Let’s ask this, let’s take a moment to reflect on the way our ICP might perceive content about digital transformation. Rama your knowledge of our icpm the conventional ways digital transformation is positioned in the media. How might our ICP think about the topic? So now, we’re going to narrow it down on one particular Avenue, but again using an ICP In what we’ve already documented for their pain points things, this is what they would likely think about the topic, positives, opportunity, growth, problem solving, future proof, efficiency and innovation, potential negative perception, buzzword fatigue, as a winner, disruptive and costly, uncertain ROI, lack of clarity, positioning to counter those negatives focused on specific needs data driven results phased approach is focused on people acknowledging the human side of transformation, focus on problem solving a human centric approach and industry specific insights. So that would be if you knew from our predictive forecast, that digital transformation was going to spike and say, June of this year, and we want to campaign for LinkedIn. This would be sort of how the ISP sees that. Right?

Katie Robbert 30:47
And so then, you know, we could say, Yes, digital transformation is a buzzword, but you know, what’s not, is the five Ps, the five Ps encompass everything that digital transformation gets wrong. And so we could do a whole series about that, you know, going through each of the PS or, you know, to the other campaign of show us your ugly, which now I’m kinda like, I like that idea where digital transformation went wrong. And how do you use the five P’s or this is how you could use the five P’s to fix what went sideways

John Wall 31:31
I think I have to start campaigning for digital upgrade because there’s really no transformations left. It’s not like there’s these companies still working on paper and fax over in the corner somewhere like it’s already digital now you just needed to be more digital.

Christopher Penn 31:46
Well, and ROI garage, hmm.

Birth versus reality, the data whisperer series and the transformation toolkits. These are still too buzzword laden and sound like a management consultant who had too much coffee wrote them up with ideas that will resonate with our ICP but are practical, useful, and down to earth for our LinkedIn campaign. Now what emphasize one of the things that generative AI is really good at is having conversations, right? We’re not in search of the one magic prompt here to do it all I know. Today, people have been posting no Anthro Berkeley stare at their 70 Use Case cheat sheet for prompts. They’re all okay. But they’re all starting points. They’re all starting points. They’re like they’re like cocktail pickup lines, right? Or icebreaker lines and not things you would have a genuine deep conversation with a person about. If you think of generative AI, we say this, we said this in our in our webinar. If you think of generative AI as the world’s smartest, most forgetful intern, starting with a canned piece of text, instead of actually sitting down with that person and talking to them, you’re gonna get much better results by having a real conversation. And you can see just from the prompting we’ve done here, for this LinkedIn campaign, we’re getting better stuff when we when we have that that interactive back and forth conversation. So let’s do a toned down approach work workflow confessions, that’s kind of cool about the most frustrating, inefficient or ridiculous workflows anonymously that people have like, oh, yeah, I mean, we used to see this at the old PR firm, there was one person on staff whose job it was to copy and paste results from Google to a spreadsheet. Like that is a ridiculous workflow that that’s not even AI like that you should have automated that years ago. The metrics that matter challenge tech stack, audit, snapshots, spotlight on small wins and office hours with an expert. That’s something that we actually do for our clients. We actually have office hours with them, so they can just bring anything, anything and everything. So these are campaigns for LinkedIn. I think, if you did a LinkedIn live, that was office hours with an expert, that would actually be pretty cool.

Katie Robbert 34:27
Yeah, that’s and that is a really light lift. Because really, it’s just a matter of scheduling the event. And letting people know that it’s going to happen. The prep itself is minimal. Because you’re saying these are the things I’m an expert on. So as it’s a it’s an AMA session, so you just say, you know, so I mean, that is a really smart idea because it’s engaging. It’s not people being talked at, they can bring their questions, their questions about digital transformation there. about how to use generative AI with all of these things. There are questions about process. And it’s an opportunity to showcase the things that we are the most knowledgeable in, and that we have solutions that we can answer questions. So that when someone goes Hmm, I wonder who knows the thing? Where can I turn to to actually hire someone to do the thing? We’ve shown that we can do the thing?

Christopher Penn 35:27
Exactly. So let’s quickly recap. We started with survey data. We said, Hey, here’s who our audience is. We then went to trend data to say, Okay, well, we know who our target is, what are the kinds of things that they would be searching for, we went over to our SEO tool, and just took a look there. From there we we got past data, five years worth of back data, that we were then able to have a generative AI forecast forward finding on the next four weeks, that term is going to trend, and then we spot we took that information. And then we fed it to general AI with our ideal customer profile, to turn it into actionable stuff. And this is the part that I think everyone has missed with generative AI. And with analytics, analytics tells you what happened. Surveying tells you why predictive analytics tells you what’s likely to happen. Generative AI tells you what to do about it. general AI will help you close that gap because for a Trust Insights has been in existence for six years now. And we have seen more analyses just go on the shelf. And stay there than ever being used because people will never share how do I take this analysis and bring it to life? How do I take this analysis, do something with my data? Generative AI is that next step? That is that we used to have this whole hierarchy of analytics, and it was prescriptive analytics. Well, that’s what generative AI is. Generative AI is prescriptive analytics. So I want to beat to death. The fact that if you’ve been doing all this stuff so far, right, you’ve done your diagnostic descriptive analytics, you’ve done your diagnostic analytics, maybe you have predictive analytics. Now it is time to bring it to life with generative AI and that’s what this can do for you was what we just did today.

Katie Robbert 37:27
And, you know, we sort of went through in the scenario that we have the ideal customer profile using that information if you don’t have that yet, but you do have a predictive forecast, you could even use generative AI in a very simplistic way to say I have these key words what are some suggested topics that I could write out some suggested campaigns, I know at a high level my audience are you know marketing managers or marketing directors in you know, the mid range size companies, you could just at a very high level list out that information without going through a larger ICP exercise and say, you know, I want to target people who are you know, in these size companies in these roles in this geographic location with these titles, so you have those four pieces of information, give that to generative AI and say and these are the keywords I want to rank for helped me put together a content plan. Yep.

Christopher Penn 38:33
The other place to look for that data. Your inbox right you have a customer service inbox right you have a call center you have something where customers have talked I hope customers have talked to you copy and paste all that information into a secure system please use make sure using systems as you will see at the very bottom of screen here says your trusted we’re using Google workspace so your Trust Insights jets are not used to train our models. That’s that’s what you want to see. Because you will be used using customer data. But copy and paste a few pages of that data in and say infer a customer profile. From this information right if you go into your customer service inbox, you will see things people have their name their title and position in the organization in their email signatures right John Wall business you know business partner at Trust Insights Katie Rivera, CEO Trust Insights. You don’t have to go through a elaborate exercise. Just gather up 1520 emails from your inbox and say, Okay, this is the collection of people and let it build a profile for you.

Katie Robbert 39:36
The other thing I would say is if you have someone like a John Wall on your team, just start a recording and ask John Wall a bunch of questions about what people are saying get that transcribed into something like otter and then give it to generative AI to say help me make sense of you know, all of this information like start to categorize that category because Catterick is oh my goodness, I can never grow zation Thank you is something that generative AI is really good at? Yep.

Christopher Penn 40:07
There’s a very good chance if your organization is mid sized enterprise that you have laying around recordings from a focus group, or customer advisory board, or one on one interviews, or surveys, or polls or quizzes or any of those things, somewhere in your organization somewhere in the bowels of your servers, your server room, is that data, go get it, right. Remember that the power of of AI is twofold. You’re the differentiating factors for any organization to win are twofold, one, the quality and quantity of your data. You have this information somewhere in your company, and the quality and quantity of your ideas. So hopefully, we’ve given you some ideas today, on this live stream about how to tie in the four layers of analytics we started with, with Diag, descriptive, like here’s the survey data, we went to diagnostic, we, we started pulling together things like search data, we went into predictive to forecast what’s going to happen. And then we went into prescriptive, which is the use of generative AI, you have, for sure descriptive analytics data, all of your organization, you may have diagnostic, you have you might have predictive, and you have the tools for generative. So if you if you are not sure how to do this, let us know this is something we just did it for ourselves. But we can help you do this. But if you’ve got the stuff and you just don’t know how to put it all together.

Katie Robbert 41:39
John Wall’s heart aches

John Wall 41:42
We are the worst case scenario for this really, right? Because we’re gonna be under $20 million. And we don’t, you know, all of our products are bespoke and custom consulting. So yeah, to really hit the point home of for any business over $20 million, you already have more than enough data laying around, you know, just point us in the right direction, because you’ve got, you know, support logs, or you’ve got customer service contact info, you’ve got product feedback, all that stuff that gets ignored in large organizations, that is all data that can be taken advantage of and leverage to actually make positive change.

Christopher Penn 42:17
Yeah, and it’s B2B and B2C I’ll you know, we have one B2C customer, we look at Google reviews, right and Yelp reviews and stuff like that. And we gather hundreds of these things every month and we can infer customer profiles from that joke Hey, you know Sally salad lover loves really dislike the anchovies that were on the salad bar last month like annual Scotts Valley Come on, Sally. But that information is stuff that can go into customer profiles and then use gendered AI to do things like if you had a restaurant Okay, here’s that here’s we’ve noticed that Sally salad lover profile Yeah, seems to be frequent to the audit the restaurant more. Here’s our current menu. We’ve got we know we’ve got to have a new menu for April. Take that ICP feed agenda of AI and say here’s some ideas or here’s some things we know we can make. What other ideas could we put on our menu for next month to appeal to Sally salad lover?

Katie Robbert 43:23
Sally salad lovers Jalisco lover, Sally, Sally lover, Sally, Sally. That’s all I can think is that it’s a tongue twister. But no, I mean, you’re absolutely right not to take away from, you know, the importance of what you’re saying is that there’s a lot you have a lot of data that you can use to find those marketing opportunities. And you can use generative AI, even if it’s, Hey, I want to find some marketing opportunities within my own data set. Where should I start? You can ask that question to generative AI and say, help me figure out how to even approach answering this question. And it can say, Well, do you have this? Do you have this? And you could say I have three of those eight things? Or you can say I have CRM data, I have marketing automation data? How should I be looking at that data to find marketing opportunity? So you don’t have that conversation with generative AI? Rather than taking it upon yourself to say, Oh, I don’t know how to do this analysis. I don’t even know where to start. Let the machine figure out the hard stuff. You take all the credit.

Christopher Penn 44:26
I mean, you volunteer at Bay Path Humane Society, you volunteer at a local animal shelter. One of the things you you could the shelter could do is just do a quick demographic profile. Even from the Google Analytics data, here’s who visits our website, boom, feed that into agenda of AI build us an ICP from that. Compare it with the actual adopters of dogs and cats say Does our website demographic match our adopter demographic and if it does great, now, here’s my demographic helped me write some fun descriptions for the ad was to help them be more appealing on the website so that the probability of someone booking an appointment to come in is greater.

Katie Robbert 45:07
Absolutely, there’s, there’s no at this time, there’s really no limit to the types of ways that you can use generative AI to just figure out how to self help solve a problem. Right? know if you’ve done an analysis, give it the analysis, say what else should I do? What am I not looking at? What am I not asking to find those marketing opportunities like so here are the five campaigns I’ve done? Here’s how they went, what else could I do? Find those marketing opportunities?

Christopher Penn 45:37
Exactly. So that’s gonna do it for this week’s show. Thank you for tuning in. And we will see you all on the next one. Thanks for watching today. Be sure to subscribe to our show wherever you’re watching it. For more resources. And to learn more, check out the Trust Insights podcast at trust AI podcasts, and a weekly email newsletter at trust Got questions about what you saw in today’s episode. Join our free analytics for markers slack group at trust for marketers, see you next time.

Leave a Reply

Your email address will not be published. Required fields are marked *

Pin It on Pinterest

Share This