So What? How to do an Audience Analysis

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 episode of So What? The Trust Insights weekly livestream, you’ll learn how to conduct an insightful audience analysis using valuable data sources like Google Analytics. You’ll discover how to utilize this audience analysis to identify potential gaps in your marketing strategy and tailor your messaging to resonate with your target audience. Discover the importance of aligning your content with the interests of your audience and learn actionable tips for reaching the right people with your marketing efforts. Stop wasting time and resources, and learn how to conduct a practical audience analysis with this week’s So What

Watch the video here:

So What? How to Do an Audience Analysis

Watch this video on YouTube

Can’t see anything? Watch it on YouTube here.

In this episode you’ll learn:

What data you have to do an audience analysis
How to use generative AI to help conduct the audience analysis
What action steps to take with your audience analysis

Transcript:

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

Katie – 00:27
Well, hey. Howdy, everyone. Welcome to “So What?”, the marketing, analytics, and insights live show. I am Katie, joined by Chris and John. Hi, guys.

Christopher – 00:35
Hello.

John – 00:36
Oh, that totally worked. That totally worked.

Katie – 00:42
All right, fine. You can have that one. We missed you last week. Last week was a US holiday so we were off keeping all of our fingers and toes by not throwing up. Thank you. By not setting off fireworks. But this week we’re talking about how to do an audience analysis.

We’ve covered this in various forms over the years, but what we wanted to do with it today is sort of bring it up to the now and walk through how to do an audience analysis using generative AI, because that’s the tool of choice for a lot of people. It’s more accessible than learning statistics, for example. It can be faster, more efficient. So what we want to do is walk through what it looks like to do a useful, valuable, actionable audience analysis.

Katie – 01:32
So, what data do you have access to? How do you use generative AI? And then what action steps you should take.

This came from a conversation that Chris and I were having earlier this week on the podcast, which you can listen to at TrustInsights.ai/tipodcast, where we were talking about how a lot of the younger generations are getting their information from specific influencers.

We were talking about the fictitious Christopher F. Pen instead of Christopher S. Pen. Christopher fake Pen is if he were a fictitious influencer and what his responsibilities would be to his audience. First and foremost is understanding who’s in that audience so that he can reach them where they are. So that’s what we want to do today. So, Christopher S.

Katie – 02:25
Pen, the real Pen, where would you like to see her?

Christopher – 02:32
Well, I guess completely unsurprising, Katie, I’m going to ask you, what is this thing that we’re trying to do? We have brought up the five P’s: purpose, people, process, platform, performance. When you think about audience analysis, and particularly in the ways that you, as an executive, would want to use it to promote the business of Trust Insights, what would you be interested in knowing?

Katie – 02:55
I think I want to know the same things that a lot of people want to know. I want to start with the foundational, the basics. I think some of those things should include not necessarily gender, but age. So, gender and location to me are less important than the actual generation of the audience because, as we’re discovering as more research is done, that is a strong indicator of what kinds of content we should be creating and what channels we should be creating for.

So, that’s generally where I want to start. What are the age ranges of the people? And then I want to get into things like, what are they interested in? What do they care about? What are their passions? What’s going to resonate with them?

We can start to get into more of the ICP, the ideal customer profile metrics. But I think to start, I think it’s really helpful to know what age groups people fall into so that you can understand what platforms they’re consuming information from. I think that’s going to drive a lot of your marketing strategy.

Christopher – 04:10
Okay. So, I think then the first place to start is not with generative AI. I think that’s actually probably not where we should go. Where you should go, the first place would be good old-fashioned Google Analytics. Not the most exciting topic these days, but very relevant.

So, I’m in our Google Analytics property here. I’m in the explore hub on the left-hand side. Go to Explore, and I picked out a few metrics. So, we have age interests. We have sessions as a… because we’re talking about audience, we’re not talking about conversions, we’re not talking about sales. We’re talking about just audience.

Christopher – 04:45
So, sessions is generally the best metric for audience because a session has information about how that person started their interactions with you, as opposed to a user, which is the human being themselves, or as opposed to a page view, which is about the content. For audience, it’s all about sessions.

So, I set our timeframe to the last 12 months, because these days with audience analysis, I wouldn’t go back more than 12. Six to 12 is about right, depending on the size of your site, just because it’s been a hell of a four years. It’s been a wild, crazy four years. Behaviors that were common in 2021, when we were all living in our basements, are very different than behaviors in 2023 when we were out and about spreading disease to each other again. So, looking back at…

Christopher – 05:44
…six to 12 months is about right. So, what we see here is we see a nice bar chart effectively from Google Analytics saying there’s a whole bunch of people I don’t know anything about because of data privacy stuff. And then these are your breakouts: 25 to 30, 18 to 24, 35 to 44, 45 to 54, 55 to 64, and 65 plus. What we see is that the majority, the plurality, of our audience is under the age of 35.

Katie – 06:16
And I think that’s really good information to start with because we run into this with some of our clients. And to be fair, we probably fall into the same trap as well, is that we’ve been doing marketing for so long that even though we think that we sort of have our finger on the pulse of what people want, we’re still doing the same things. We’re still sending a lot of emails. We’re still putting a lot of things on LinkedIn. We’re still creating a lot of YouTube videos. And those things are good for some of those audiences, but I don’t know that they’re good for the younger audiences that we also want to be reaching, the people under 35, because, to be fair, nobody in our company is under 35.

Katie – 07:00
And so we don’t necessarily know, if we send them a bunch of email, is that going to work? I think that, for me, that’s where my head is at, is: Are we missing the boat on reaching the majority of our audience, or potential audience I should say, because we’re not delivering them content on the right channels? And I think that’s sort of where I keep getting stuck is I know what we’re doing for marketing. I know what we’re doing for marketing is limited because of our resources. But are we making the most of what we have?

Christopher – 07:35
Right. I forgot, by the way, very critical thing to remember with your data is: Make sure that it is good quality. We had an issue in April with some data contamination, so I had to filter that by saying, making sure that it’s just our website. But the proportions are still the same. 25 to 30, 18 to 24. So again, that top plurality is under the age of 35. People my age and up are in the minority, pretty significant minority.

So that’s who our audience is from an age range. Now, the next thing we can do, just very broadly, let me go ahead and dump this PDF out, is we can swap out our age with our gender, just to take a look at broadly what that looks like. And here, we’re split 50, right down the middle, which is unusual.

Christopher – 08:32
In the past, it has skewed more heavily male, and now is about even.

Katie – 08:38
So, I don’t want to get too deep down the rabbit hole, because I feel like we could, but I’m less interested in the gender identification because I feel like the services that we provide don’t cater to one gender or the other. I wouldn’t necessarily cater our marketing to one gender or the other, or, somewhere in between. Am I wrong in thinking that this particular metric doesn’t matter…

Katie – 09:09
…to us for audience building?

Christopher – 09:09
It does because it’s who we’re attracting, not necessarily on a services basis, not on what to sell, but on, as you mentioned, all the content we’re creating, all the places we’re putting it.

One of the things, and I have no data to prove this, one of the things I think is a contributing factor to this is the fact that we do have a CEO who identifies as female, and we feature you prominently in our podcasts, our live streams, our newsletters and stuff. That communicates to audiences that this company does not have a problem with a woman leader, for example, and that our content inherently will be supportive of that point of view. Otherwise, we wouldn’t be publishing live streams and podcasts and newsletters with you as the headline act.

Katie – 10:03
Okay. That’s fair, because I’m looking at it in terms of what we’re going to create. But I hadn’t thought about the representation side of the conversation, which I think does make sense.

That said, if it were skewing more male than female, I don’t know what we would do to resolve that other than maybe featuring me more. I mean, those are the kinds of things that we would have to look at and take back and sort of figure out in our planning. So, I will be more mindful of this metric moving forward, because I hadn’t thought about it from that perspective.

Christopher – 10:44
The next thing we might want to think about is looking at those interests. What are the interests of our audience? What are the things that… what else do they care about besides our content?

So, this data comes from Google’s DoubleClick network. It is… if it’s available, it’s going to be… in our case, it’s about 25% of the data. 75% of the data does not have this attached, but it’s a decent representative sample.

What we see are the things that people care about: technology, business professionals, shutterbugs, avid investors, movie lovers, TV lovers, etc. Now with this data, one of the things you can do is do those splits. You can add in splits to segment it out by gender to see if there are certain topics that resonate more with one than the other.

Christopher – 11:32
You can also split it out by age, and either add that in or do two separate assessments so that you get a full understanding of what your audience really actually cares about.

Katie – 11:45
I feel like… so in Google Analytics 3, it was called, what, the “infinity audiences”? I feel like it’s an underused part of the system because we tend to skip over what our audience cares about. We, as marketers, try to tell them what they should care about.

How well is that going for people? I don’t know. How well is it going for you, John, when you try to tell me what I should care about?

John – 12:13
Yeah, no, I can show you the scars from last time we did that. No, it’s true.

It’s really interesting to me because this stuff is all based on their ad data. I mean, we’ve talked before at length about how pretty much Google Analytics is just becoming a tool for the ad network. They are more than happy to push you this kind of stuff to give you a better idea of where your audiences are and what kind of things they’re interested in. Hopefully, you can find some ways to map your product to that. It gives you a chance to have a lot more effective messaging.

Christopher – 12:46
Yep. Especially, if… this is a separate rabbit hole for another time, but you can take these topics and you can use the topics with generative tools to create ad copy for these specific topics that involve your product or your service and the topic, and then deploy that in Google Ads targeted to those specific… these segments.

So for example, I’m going to use “shutterbugs,” right? Your ad creative might have a person with a camera, and the analogy might be “Make sure you’ve got a clear picture of your data. Trust Insights.” The ad would be about that, but it would be targeted towards that affinity, knowing that you can target that affinity within Google Ads.

Katie – 13:32
Makes sense. Again, it’s sort of, I feel like, it’s the unsung hero of Google Analytics. It’s data that has been in there the entire time, but we often just focus on sessions and conversions. We don’t take the time to really dig into, “Who are these people who are coming to our website? What do they care about? And how do we keep them?”

Christopher – 13:58
Exactly. So that’s… So, I think that’s obviously relevant and important, is having that information, of all that demographic information being available for assessment.

So, the next thing that you want to do, if we’re doing audience, trying to understand our audience, we probably want to spend some time knowing where our audience comes from. That seems like an obvious thing to say. But let me flip over to this tab here.

So, this would be your Source/Medium and the traffic, the sessions you get. Where is your traffic coming from? Let me make sure that I plug in the hostname, just to ensure that this is correct. Apply. Okay, so, “hostname contains trust insights.”

So, we have 100,000 sessions here. We can see that 44,000… that comes from a newsletter. 16,000 we don’t have data for, it’s direct. We have the company newsletter, which is 15,000. Google organic search. We have some mis-tagged stuff from the “Almost Timely” newsletter. We have a couple of spam bots, which is, we know that we’re fighting those off. Last year, and since this data is a year old, one year inclusive, we have LinkedIn, which is surprisingly high on this list. Now, granted, out of 100,000 sessions, that’s 1.3%. But it’s still, that’s up there. Speaking, Bing, Slack, and so on and so forth.

So, from this, we have a pretty good sense of that audience. Where are they coming from? To your point, Katie, when you’re saying, “Well, how are people getting to us?” these are the data sources for how.

Katie – 15:37
They get to us. I would counter with, these are the only data sources that we’re publishing content on. We are lightly doing so. What’s not represented here, aside from LinkedIn and a little bit of YouTube, is social media. Interestingly, we’re not posting anything on Facebook, and yet we’re still getting a few people from that. Instagram, I don’t see on here. Twitter is still showing way down there.

Christopher – 16:14
Instagram is below your website.

Katie – 16:17
Wow. It’s considered…

John – 16:20
Well, there’s no way out of Instagram, though. That’s totally normal.

Christopher – 16:24
No, there is, the LinkedIn bio.

John – 16:25
Yeah, the LinkedIn bio is the only out.

Katie – 16:29
I don’t see TikTok on there, for example. It’s… the data makes sense because it’s where we are. I feel like that to me… maybe I’m not articulating it, or maybe I don’t know what my question is necessarily, but people are coming to our site from where we post data. But if our audience skews younger, where are they? Where do they want to consume stuff from?

Christopher – 17:07
Right. Well, we do post to TikTok, and Instagram, and YouTube, and most of the sites. I think the only site where we don’t have a lot of heavy publishing right now is, Snapchat, because you’d have to post those stories individually there, and that’s a pain in the butt. But for the most part, we cover most of the social networks.

The other one, we don’t publish too aggressively yet because it’s not supported in our schedule is Threads, but support for that is coming very soon in Agorapulse, which is the platform that we use. So, I would expect to be able to start seeing some of that with the understanding that it appears to be using Instagram’s link shortener. You want to make sure that your UTM tracking codes are solid to account for that.

So, I think we are in a good number of the places that we can be for the size of team that we have and the resources we have. Could we be doing more with YouTube Shorts? Could we be doing more with more creative uses of TikTok than what we use it for? Yes, absolutely, we could. We don’t have the bandwidth to support that right now. But we are in enough places. We’re in… we do get traffic from Bing. We do get traffic from LinkedIn and stuff that makes logical sense.

So, okay, so that is, I think, a good sampling of understanding the data from Google Analytics. Now, there’s a couple other places that I think are worth pulling data from. One of them is Google Search Console.

Christopher – 18:34
I think if you want to understand your audience, understanding the terms that they’re finding your website for is a good idea. This is… we’re not going to be doing an extensive analysis on this, we just want this data because we broadly know the demographics and the interests of people. We know where they came from. We now have some knowledge of the language that they’re using for which we get traffic from search.

That’s important, just to say, “Yeah, these are the terms that Google associates with us,” and what we get clicks for, where people say, “Yeah, that looks relevant. I’m going to click on that.” Makes sense. So, we’ll want that, too.

The last set of data that I think is useful, for at least your audience, is understanding what people are telling us. That can come from a couple different places. Today, we will use… we’ll export the data from our contact form on our website. We’ll say, “I want the data. I want everyone who’s filled out the contact form.”

Now here’s the catch. About 80% of what comes to our contact form is garbage. It’s sales pitches and stuff like that. One of the things that you can and should do is, export your contact form stuff and then feed it to generative AI and say, “I only want people who are asking sales inquiries about us. Discard sales pitches to us. I want people who are talking about and wanting help from us.” To better understand our… when people do make the effort to reach out, what is it they’re saying?

Those are the four big data sources we have. We have Google Analytics data, and I think it’s fair to say that with the GA data, you can get some useful insights just from that. You don’t need AI for that. You can look at it and go, “Oh, that’s interesting.”

Katie – 20:31
I would agree with that.

Christopher – 20:33
Then we have our contact form data. We have our Google Search Console data. Is there other things that you would be thinking about from an audience perspective that you think is relevant? I can think of one or two other things, but it’s beyond the scope of today’s show.

Katie – 20:47
Yeah, I think the things that I’m thinking are the same. So I would want to know, who subscribes to our various things. Who subscribes to the podcast? Who subscribes to the newsletter? And then, who follows us on social media accounts? But I think, to your point, that’s beyond the scope of today. But those are the other things that, if we were doing this as a project, I would take the time to find out.

Christopher – 21:13
Right, exactly. Yeah. If you were a paying customer of Trust Insights, which you can be by going to this link, it would include this.

The other thing I would do is, I would dump the contents of our Slack group, Analytics for Marketers. I would dump the contents and pull out the conversations, the public conversations for the last year, because, again, that’s a part of our audience. Knowing what people are talking about, especially in the off-topic channels, the Pets channel, the Music, the Books channel, I think those are things that are useful insights from a subset of our audience. I would call our most loyal audience. I think that would be… that would lend some color to those affinities and interests.

Katie – 21:54
I think that makes sense, yeah.

Christopher – 21:56
Okay. So, what do we do with all this stuff?

Katie – 22:00
You tell me. That’s why you’re here, Chris.

Christopher – 22:04
Fine. Let’s go to Google Gemini and start asking questions, as we always do. We’ve got to prime the model.

So, we’re going to start off by saying: “You are an AMA award-winning Chief Marketing Officer with specialization in audience analysis, cohort analysis, demographics, customer journeys. What are your best practices for understanding the makeup of a company’s audience based on digital data?”

So, we start off with our first question, which is a priming question to load the basic context for the task that we’re setting out in front of us. It’s going to go through and spit out a bunch of stuff. This is following, if you have not tuned in before, this is following the Trust Insights PAIR framework.

Christopher – 23:06
If you go to TrustInsights.ai/pair, there’s a PDF you can download totally free, no form to fill out, no blood to give, and you can follow along with the steps.

The next step question is going to ask is, “Great. What are some things that less experienced marketers get wrong about audience analysis with digital data?”

This prompt provides additional content for the model, but provides what’s called contrastive prompting. This is essentially helping the model understand what not to do. This question, I think, is probably the second most important question.

Now we’re going to add a third question on this. This is the “refresh” phase of the framework: “Great, what tips and tricks do you know as an experienced CMO that others might not, that we have not discussed yet?”

So now we’re going to push aside the obvious, basic stuff, and have it dig into additional topics, additional concepts that we might want to incorporate into this analysis. So, now we’ve got the framework for really good audience analysis.

Again, if this was something that you wanted to reuse over and over again, you would want to now turn this into system instructions. You would say, “Help me build a GPT for this,” in OpenAI’s GPT, “or a Gem in Google Gemini.” But you would want it to consolidate all this knowledge.

For today, we’re going to leave it as is. You can see we’ve consumed only about 1,800 tokens. It’s about 1,500 words. That’s a good starting point for this. So, we’re now going to say, “I’m going to provide you some data based on these best practices. Help me assemble an understanding of my audience.”

Christopher – 25:22
I’ve said “Write in an outline format using Markdown.” The reason for it is because that is a programming… well, it’s a formatting language that is super easy for machines to understand and it is compatible with the way that Gemini in particular spits out data. If you’re using ChatGPT, it does this automatically, and not necessarily… well, I don’t love the way ChatGPT’s formatting comes out, but that’s fine. If you’re using Anthropic Cloud, it does something similar. It doesn’t really matter on the formatting side.

So, let’s go ahead and hit “Go” on… so, I took PDFs while we were going through Google Analytics. I was taking PDF exports from that dashboard. You can get that data programmatically through the API if you’re a coder. But I figured for today’s show, we would just do it the non-coder way.

So we had PDFs of each of the sections we looked at. I took the distilled down sales pitches, and I took a text file from Google Search Console. Those are the data sets that we’ve split out here.

So, let’s take a look here. “The Trust Insights Audience Analysis,” we have “Data Source Methodology.” Here’s the “Gender Demographics.” Yep. “Age Distribution,” followed by their behavior: acquisition channels. Almost Timely, website engagement, qualitative… let’s see. So, it is essentially creating the recipe.

Katie – 26:47
I have a question, Chris.

Christopher – 26:48
Yes.

Katie – 26:50
So, you just did something that I am a little surprised at. You took PDFs of the data and not CSV files of the data. Is there a reason for that? Because my instinct would have been CSV files, not PDFs.

Christopher – 27:08
The reason I did that, I always do that with ChatGPT. I conditionally do that with Google Gemini. The reason for that is, if you provided a CSV, the tools, in an attempt to be helpful, start loading Python and try to write code to interpret it. I’m, “No, I just want you to understand the text, and I don’t want you to try to build software around this.” ChatGPT is the worst at doing that.

Katie – 27:32
Okay, that’s helpful. That explains what I was running into when I was working on our project yesterday. I’d given it CSV files and it was taking forever because it was building Python code. I couldn’t figure out why it wouldn’t just do exactly what you’re saying, which is to just… “Can you read the thing, or no?” So, that’s a really interesting pro tip that I wasn’t aware of. I was doing that same thing because I thought, “Well, the CSV file must be easier for it to read,” not a PDF. So, pro tip: If you don’t want your generative AI to start building Python code, you just want it to answer your question, use a PDF, not a CSV.

Christopher – 28:14
Exactly. Now it’s spitting out here basically the analysis of the data and trying to sew it together.

So here’s the “Methodology.” Here are the “Limitations.” “Quantitative analysis: email, location data was not provided.” So we might want to provide the location data because it does say that might be useful.

“Behavioral channel: Email is the dominant channel.” Again, this is not a huge surprise.

“Website engagement” we have “sales inquiries.” The different types of… “general consulting and training,” “speaking engagements,” and “specific requests.”

“Industry-wise: marketing agencies, higher education, industrial distributors, real estate…” and so on and so forth. That’s all coming from our website contact form.

Now, again, as we were saying earlier, if this was a paying client, we’d be pulling that data on HubSpot as well, because we want to know the specific industries that… that data, by the way, you can expert export from HubSpot via API.

“Integrating AI into marketing,” “navigating GA4,” “developing AI strategy,” those are the kinds of things that people are interested in.

Then it attempted to build some personas. “Event Emily: corporate, showcase GA4 expertise, develop a training programs for agencies, actively pursue speaking opportunities, refine website content messaging, leverage email marketing, expand data collection.” So that’s its analysis, based on the pieces we’ve provided.

This is true of all generative AI, but especially in a situation like this, the more data you provide, the better it’s going to get. So, it even identified it’s hamstrung by some of the data we didn’t give it.

Katie – 29:54
What’s interesting, because we’ve done previous episodes, and we were talking about this yesterday, too, of ideal customer profiles. This is very similar to an ideal customer profile. But what’s different, at least from where I sit, is that the ideal customer profile is who we want to attract. This audience analysis is who we are attracting. And now we just need to sort of do the Venn diagram of the two.

Christopher – 30:27
Funny you mentioned that.

Katie – 30:31
It’s almost like I’ve worked with you for 10 years.

Christopher – 30:33
This is excellent analysis of the kind of audience we are currently attracting. I’m now going to provide you with our ideal customer profile, the kind of audience we want to attract, “Read through our ICP, and then compare and contrast our audience profile with our ICP.”

Now, let’s go into Gemini. Go into my Drive. I’m going to move this off-screen here because there’s stuff that people, the general public, should not see. Enjoy my cats.

Katie – 31:19
John, I feel like you and I should have little skits prepared for when Chris is computing stuff and it’s just kind of like dead air, because it happens enough that we should have some sort of… something ready.

John – 31:32
Dance break.

Katie – 31:33
Little dance break.

John – 31:36
This segment brought to you by Giphy.

Christopher – 31:45
So, again, one of the nice things, and this is… it’s a semi-pro tip: You should have things like your ideal customer profile and your audience profile, and things, in PDFs or text files or Markdown or something like that handy so you can just drag and drop it right in.

One of the reasons I love using Gemini is because it’s integrated into the rest of Google Workspace. I know where in our Google Drive this stuff is. I can just plop it right in and I don’t have to sit here spending time building it. It’s just, it’s canned. It’s ready to go.

All right. Let’s see. “Similarities: Focus on data-driven marketing, interested in AI and advanced analytics, professional roles, emphasis on business impact.”

“Differences: Your current audience appears to be more diverse in terms of industry representation, whereas your ICP prioritizes specific sectors. Your existing audience seems to include a wider range of company sizes, where the ICP is focused on mid-market to enterprise. Your ICP emphasizes actively using CRM and marketing automation, which might not be as prevalent in your current audience. Digital transformation maturity: Your ICP targets companies actively engaged in digital transformation. This level of maturity may not be as widespread in your current audience.”

“Actionable Insights: Refine targeting and messaging, develop content tailored to the ICP pain points, identify key decision-makers, leverage industry-specific channels, and highlight your digital transformation expertise.”

So, it is saying that we’re attracting an audience that may not be as focused as we want it to be. These are the five things that we should be doing to do better at that, which is definitely interesting.

Katie – 33:10
And I think that what opens up are those really good marketing planning conversations. I can see value in attracting all of these types of people because awareness is at the top of the funnel. You want to have good awareness. Do you need to be as specific if your purchasing audience is a little bit more niche? I don’t know.

Those are the things that I would take this and take it back and start to map out: Where do we narrow down the audience in the customer journey? Is it okay that we’re getting a wider audience at the awareness phase, and then as they go, how do we narrow them down through the different phases of the customer journey to get to the ICP?

So, the ICP needs to be built into the awareness phase, but it doesn’t exclusively need to be the awareness phase.

Christopher – 34:16
And that’s a really, I think, very critical point. The critical point that you manufacture there is that there will be people who are in your audience who are supporters, fans, boosters. They’re never going to buy anything from you, and they don’t need to, but they do need to talk about you. They do need to refer you to people who can make those buying decisions.

One of the dangers of this sort of analysis is that you get so laser-focused on one type of person that you forget, particularly in B2B. But any complex sale, including things like real estate, higher education, etc. There’s a lot of ways to influence a decision-maker, and it may not be the obvious, “Oh, you connect with this person on LinkedIn and send them a FedEx,” and stuff like that.

We’ve been in situations where an executive says to the intern, “Hey, make me a shortlist of companies that do this.” The intern goes and Googles it. Well, if you want to be on that shortlist, try to influence the Chief Executive. Or you could influence the intern, but through search and content marketing.

So, what I did was I said, “I like this. Give me some ideas for highlighting digital transformation.” I gave it our Services pages as well, so that it had some additional ways to incorporate our expertise.

“The five levels of digital transformation for marketers, data chaos to customer clarity, beyond vanity metrics, case studies, host a webinar, AI for digital transformation, a roadmap for marketers, white papers and guides, service-specific, strategic consultation.”

Christopher – 35:51
“Here’s some ways to change our services, the way we talk about them, as a way to show people our digital transformation expertise. Showcase our marketing mix modeling capabilities, which, by the way, are better than ever, thanks to generative AI, data intelligence.”

So, there’s a lot of ways that the tool, given, again, given the data we’ve given it and its background knowledge, now says, “Here’s how you should be doing your marketing to broaden awareness of our digital transformation capabilities.”

Katie – 36:27
And see, Chris is smart. I know he’s saving all of this, because I’m going to ask him for all of it afterwards.

Christopher – 36:34
Exactly. So, that’s… that is the framework for audience analysis. The hardest part by far is getting all the data together.

Katie – 36:45
Oh, yeah.

Christopher – 36:46
The prompting… we’ve been prompting this thing for 11 minutes now. It took 25 minutes to assemble the data, and from three very easy sources! That does not count our CRM, our email marketing system, our social media management system.

Some things that we might include, if we wanted to do a deeper dive… go into our AI transcript system and pull all the sales input calls we’ve ever done in the last year and put all that data…

There’s so many sources of information you can bring in: surveying people and doing focus groups and one-on-one interviews, and shadowing whatever the business you’re in. Gathering that data is what’s going to feed this modeling capability. The more you have, the better you’re going to do.

Katie – 37:34
But I think that goes back to where we started the episode, which is with the five P’s, to really understand: What is the question you’re trying to answer? Because you’re right, there’s a lot of different places where you can get data, but they might not all be useful data sets to answer the question.

So for me, I wanted to understand more of, “Who’s in our awareness stage of the funnel? How are they consuming content? What are we missing?” When I go through the analysis, I’m going to find that information. I’m not looking for, “Who’s converting?” That’s a different part of the funnel. That would change the kind of data that I might be looking for, the different metrics.

Doing your requirements upfront, even for something like this, is essential, because you’re right, gathering the data is what’s going to take you the longest. If you’re gathering the wrong data, you’re not going to get to the answers that you’re looking for.

Christopher – 38:35
Exactly. And even with the tools that you have, there are ways to segment that data out.

One of the questions that, Katie, you always ask me is, “Yeah, your newsletter is great. It’s got almost 300,000 subscribers, and every Sunday, you harass the internet with it. But is it the right audience?”

One of the things that you can do, for example, in GA4 is you can say, “Okay, well, now let me restrict this down to maybe just the Services pages on our website. Do we then see those… the source/mediums change for the pages that… where we know there’s high sales intent? Team… your Team page, your About Us page, your Products and Services page, your Pricing page, if you’ve got a page on pricing, and your Contact page.

Those are the four pages that… where there’s higher sales intent, and you might want to build a filter, or do it through the API, for something like Google Analytics. Say, “How does the audience change from everybody to high sales intent?

Clusters of content?

Katie – 39:38
I could derail things, but I’m not going to.

Christopher – 39:40
You got time?

Katie – 39:43
Why is the About Us page considered high sales intent? Because, again, I’m an N of one, so this just might be my own personal opinion. I have more of just a natural curiosity. Sometimes I’m, “Who is this jackass?” I go to the About page, but I have no intent on buying. If anything, it’s doing the opposite. Why is going to an About Us page considered high sales intent?

Christopher – 40:13
By itself? It’s not.

But in a visit where there’s also a visit to a Product or Service page, it is. Statistically, that has been our experience, and we’ve seen that in software like Demandbase, for example, which will highlight the pages that people visit before converting.

Are these pages? In fact, we used to do that with our old digital customer journey, and you could see the pages, the About Us, the Team page is very high up.

We know this anecdotally, too, because in, I would say, half of the input calls that we’ve had with prospective clients, one of the questions they’ll say is, “Okay, this is great. Who on your team is doing the actual work once we engage?” We used to get that all the time with the PR agency, and that was always an awkward moment because it would be, “Not us. We’re just here to sell you something, then somebody else…” Thankfully, Trust Insights, this is it.

Katie – 41:05
John, do you feel like you have a good grasp on who makes up the Marketing over Coffee audience, or is this a type of analysis that you would find valuable?

John – 41:16
It’s very similar. It’s pretty much the same audience, I would say. I mean, none of the stats look different. It’s all the same thing of, it’s people who are looking for cutting-edge marketing. Totally matches the mission of both places. It’s just weird how for Marketing over Coffee it’s so convoluted in that… I’m… we’re only interested in three one-thousandths of a percent of the audience, which is the folks that want to advertise back to the same group, which is an absurdly tiny slice. But otherwise it’s… yeah, it’s more of the same thing of, how focused do you get and how broad of a net do you cast? You’re kind of always fighting that challenge of, you do want to get the interns and just the random public because they make references. They’ll remember your name and pass you on when the time comes.

Christopher – 42:06
So, one thing that you mentioned is really important, and we haven’t talked about it. We might want to talk about, either on this episode or a future episode, is: All of this data and everything that we’ve put together is really good. It’s very useful. It tells us who is in the audience. In understanding the audience, it does not tell us who’s not in the audience.

It does not tell us who might be valuable that we’re missing. Is there a segment of the audience that we’re not paying enough attention to? But there’s a hint of it in the data. “Yeah, you got 11 visits from Pinterest, but you haven’t posted on Pinterest in six years.” Is there a possibility that data from Pinterest… that Pinterest might be a valuable platform?

One of the challenges that a lot of companies have that we don’t have, thankfully, is: We have access to multiple GA accounts. We have Marketing over Coffee, we have my personal account, we have the Trust Insights account.

One of the things you might want to do is do a comparative analysis to say, “Is there a different audience on ‘Christopher S. Penn’ than there is on Trust Insights? If so, is that audience potentially valuable? Should we be doing more to address it? Is there a different audience on Marketing over Coffee that we’re not talking to?”

You can also, with some of this data, you could use SEO tools — Semrush, Ahrefs, etc. — to say, “Okay, well, what is… here’s our general starting keyword list. What are the related terms?” Then say, “Why are those not in our Google Search Console? Is there an audience that would use that language?”

This is where language models are really powerful, is: Is there an audience that uses that language that we’re not talking to? Because we don’t know what we don’t know. We don’t know that blind spot, that empty space.

Katie – 43:56
So, you definitely just hit on a couple of things that maybe for a future episode we could tackle. I mean, you’re right. I always want to know, since we do have these multiple streams of audiences coming in, “Where… where the value lies,” for lack of a better term.

John has an audience with Marketing over Coffee, and so on and so forth. I want to know, within those different streams, who’s a buying audience, and who’s a, “Here for the ride,” audience. Both are important, but we’re going to treat them differently. We need to get better at segmenting versus everybody gets the same thing.

This is… again, sort of where we started on the podcast earlier this week, it gets into that personalization. If we are sending the same message to everyone, then we’re missing a good part. If we’re sending the same message to the different age groups, different generations, we’re missing a good part. We’re missing the boat, basically, by not tailoring our messaging to these different segments of the audience. That’s something that I would want to use this analysis for. That’s my action. It’s, “Great. What do I need to do? I need to understand how to reach these people where they are. Not ask them to meet me, but I need to meet them.”

Christopher – 45:25
So, I asked… these people… I asked… oh my God, that was terrible English. I said to the model, “Okay, here’s a strategic question. The white space, the underdist… what part of the audience might we be missing that would eventually be a valuable audience, based on all the data we’ve uploaded?”

It came back very quickly and said, “You’re missing… Katie, you’re missing the data-savvy executive: the CEO, the COO, the CFO. These people are expected to be data-savvy.”

Here’s the evidence: Multiple inquiries come from CEOs and founders. Speaking topics align with executive concerns.

Here’s what this audience needs: Not data science — aka, not me — but data strategy. They don’t need to code, but they need to understand what’s possible with data, how to ask the right questions, how to foster a data-informed culture, bridging the gap between technical and business case studies that speak their language.

This is using the language model and essentially asking, “Hey, here’s this data set. You know all language. What is this data set not have in it within this particular probabilistic scenario of marketing and AI?”

Here’s how to reach them: Executive-level content, white papers, exec summaries, tailored briefings, speaking at executive events instead of marketing conferences. Partnership with executive coaching firms to help introduce the company to these executives through their coaches.

Then here’s… it gives very specific examples of how you might want it to do that, kind of targeting the kind of executive networking…

One of the things that is always a push and pull, sort of a balance that we struggle with from time to time, is figuring out: How do we balance the different voices of the team at Trust Insights? This, to me, feels like the thing that, Katie, you should hang your hat on, from one CEO to another. Here is how… here’s how Katie successfully manages an incredibly technically complex environment for the company and for clients. Here’s how Katie thinks about data strategy, as opposed to data science. Nobody cares about, whether DPO or SPPO is a better model for opti… for optimizing a language model. Direct preference optimization versus… never mind. I care about that, but no one else does. But knowing what should we be doing?

Christopher – 47:56
I mean, one of the big things, if you go into some of the CEO forums online, especially the ones that are relatively safe spaces, people are saying, “I don’t know what the F to do about AI. Everyone says it’s important, and I don’t get it. I know I should get it. How do I get smart on this?”

As we think about, Katie, the kind of content that we should be creating, but especially you, in the newsletter and in the podcast, and your own content on LinkedIn… got to help your fellow CEOs out.

Katie – 48:30
All right, you heard it here first: Less Chris Penn.

Christopher – 48:35
That’s right, I got my own stuff.

Katie – 48:36
You’re sticking him back in a box.

Side note, when we worked at the agency, we did try to develop “Chris in a Box,” and it did not go well, because you can’t cage the animal.

Christopher – 48:50
I am ungovernable.

Katie – 48:53
I think it’s a good place to stop.

Christopher – 48:55
It is a good place to stop. Any final thoughts?

Katie – 48:59
I got some work to do.

John – 49:00
Dig around and make a mess. Go get in trouble.

Christopher – 49:06
All right, we will talk to 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 TrustInsights.ai/tipodcast and our weekly email newsletter at TrustInsights.ai/newsletter.

Got questions about what you saw in today’s episode? Join our free Analytics for Marketers Slack group at TrustInsights.ai/analyticsformarketers.

See you next time!

Need help with your marketing AI and analytics?

So What? Marketing Analytics and Insights Live

airs every Thursday at 1 pm EST.

In this episode you’ll learn:

Transcript:

Leave a Reply Cancel reply

Subscribe to our Weekly Newsletter

Pin It on Pinterest