So What header image

So What? How to use AI for Email Marketing

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? we walk you through the process of using generative AI for all your email marketing needs, and changes and tactics you should be aware of for 2024.

Catch the replay here:

So What? How to Use AI for Email Marketing


In this episode you’ll learn: 

  • What email marketing tactics you should be using in 2024
  • How to use AI for email marketing analysis and planning
  • What changes in email marketing tactics to be aware of

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 0:36
Well, hey everyone, Happy Thursday. Welcome to so what the marketing analytics and insights live show, I almost forgot to put up the banner because I was clapping along with the music. I’m Katie joined by Chris and John, as always, Happy Thursday guys. Oh, oh, so close.

This week, we’re talking about how to use AI for email marketing, we’re going to talk about what email marketing tactics you should be using in 2020, for how to use AI for email marketing, analysis and planning and what changes in email marketing tactics to be aware of. So last week, we covered what’s going on in SEO, and how to use AI to enhance your SEO and 2024. So we figured we would just sort of continue down this road and tackle email marketing. email is one of those channels that you have a lot of control over. And so there’s a lot that you can do with artificial intelligence, there’s some things you probably shouldn’t do with artificial intelligence. And we are here today to cover as much of that as we can for what’s going on right now. So Chris, where would you like to start?

Christopher Penn 1:52
Let’s start with what what the state of email marketing is, here’s the deal, that hasn’t changed much, right? The basics of email marketing, kind of like the basics of SEO, are pretty much the same year to year. So there’s the technical aspect of things, right, making sure that you’ve got all your infrastructure set up and working properly, enable your your deliverability protocols, which we’ve got past episodes of the live stream, if you want to catch up on those because that, as you saw, it was an entire episode unto itself. And then you have the three pillars of any direct marketing campaign right? From oh, gosh, who was it? Was Bob stone 1968. Right? List offer creative? Are the three pillars. Do you have the right people? Do you have the right offer to put in front of people? Do you have the right creative? And you need to do it in that order, have the right people, right, because if you’re sending email to the wrong people, it doesn’t matter what the offer is, you need to have the right offer. And then you need to have the right creative. Now one of the things that people do the most wrong with email marketing, is they jump to the least important which is creative, which is like, oh, we need better subject line testing, oh, we need to AB test our copy. Well, if you don’t have the right people, and you don’t have anything that those people want. Your creative is irrelevant, right? If if you have a whole bunch of people on your list, who really want steak and all you sell is apples, your creative can be wild, award winning, and you will still not sell anything to the people who want steak.

Katie Robbert 3:32
I’m guessing that this is part of what you’re going to cover or you know, maybe not you always like to surprise me with what we’re going to do. I think one of the conundrums that people run into with email marketing is, you know, the whole if you build it, they will come. But like how do you get people to subscribe to your newsletter? Do you have to like you have to cross promote it on different channels, you have to let people know that it exists in order to find the right people. And so like that I feel like is one of those hang ups with email marketing. That’s not a new problem, by the way. That’s a you know, as long as email marketing has existed, people trying to figure out how to find the right subscribers to their list has also existed.

Christopher Penn 4:20
Oh, yeah, that’s something that hasn’t changed. That is where the people that you want to talk to? Where do they hang out? And then how can you get to them in those places? That can be things like influencers. That can be things like natural search, organic search, SEO. That can be frankly, in the beginning for most companies are buying a lot of ads, right? You’re buying a lot of ads to promote, lists subscribership things and if you want to get up and running relatively quickly, buying ads is the fastest way to do it as long as you’ve got good targeting. And you’re very clear on what your what people want. So the number one thing that you need to do before or anything else is build your ideal customer profile, which again, we’ve we’ve done podcast episodes and live streams on how to do that using generative AI, you absolutely positively need to do that first. Because if you don’t do that, then nothing else, what we’re going to talk about today matters if you don’t have a good ICP. Once you’ve got that, you can even go into a generative AI system and say, like, based on this ICP, what are the probable places that my audience is going to hang out? And then look at those channels and say, Okay, well, can I reach them there?

Katie Robbert 5:38
It’s interesting, because, you know, we often talk about email marketing as its own thing as its own digital marketing channel. But what I’m hearing is that email marketing is so is very reliant on other digital marketing channels to be successful. And so you have to have a more well rounded marketing plan in order for email marketing to be effective.

Christopher Penn 6:08
Yeah, absolutely. Absolutely. If you think about it, if you think about the customer journey, right, awareness, consideration, evaluation, purchase, that’s the buyers journey. Ownership loyalty, evangelism as the owners journey, emails, not an awareness channel, because you can’t just go cold emailing people, right. In some places, it’s illegal. So as a result has to be a consideration and evaluation, which means you’ve got to do something else at the awareness phase, to get people to the email. Here’s the other thing that people forget. email is the alternate sale. Your email marketing is your alternate sale. And what I mean by this is that if you can’t get someone to buy something from you, or do something of economic value, the alt give the other option is join our newsletter so that you can retain them, and then eventually hopefully sell because we know LinkedIn, economic Marketing Institute, whatever I can’t remember, the name has said that it in for the average buyer, night or for the business 98% of your audience is not buying. At any given time only 2% is actively interested in buying. So the other 98%, you’ve got to have something else to quote, sell them. If they’re not going to buy that day, that is the email.

Katie Robbert 7:28
So where does generative AI fit in? So if email as a tactic hasn’t really changed? Where does generative AI fit into all of this?

Christopher Penn 7:39
So the place we want to start with generative AI are the six major use cases, right? Generation extraction, summarization, rewriting, classification, and question answering those the Big Six, the, the generation is probably the easiest for and is the one that everyone knows. So let’s go ahead and pop into, I’m gonna show a couple of different places where you can do this. Again, you need your ICP, if you don’t have your ICP, this is not going to go well. Your ICP is

Katie Robbert 8:12
that episode, you can find it at trust Look for our so what playlist?

Christopher Penn 8:17
Exactly, your ICP should be stored somewhere in Google’s AI studio, I actually embed the ICP right here at the top of the system instructions says Here it is. And this is the details for it. If we go over to ChatGPT, I built a custom GPT that has the ICP built into it, nothing. So you once you’ve built that ICP, you want to have it available as utility all the time and the easiest way to do that is having an embedded in the system itself. If you’re using Watson Studio, Watson Studio, you’ll embed that in your system prompt etc, etc. So rewrite it a generation rewriting of the simplest ways to go so let’s go ahead and take up an email here. I sent out earlier today a pitch and this pitch read like this. Now this is sent to my for my newsletter, so I’m fairly confident in the tone of voice and things. But suppose we want to repurpose this pitch for Trust Insights. I would say okay, let’s here’s my idea. Here’s my pitch. Here’s an email marketing pitch I plan to send out please review it and make suggestions for improvements that would align it better with our ICP. Right so again, this is this is basic rewriting. What do we do? They do the analysis say, Hey, your, your emails a friendly, engaging tone, but couldn’t be, it could work better your pain points aren’t on target, you might want to include some job titles, some ideas for some subject lines, and so on and so forth. And so it’s going to go through and rewrite this email, if we want to do that, against the ICP. And the critical thing to remember here is that you want your ICP to be the reaction to this thing, because we all have as marketers, we all have an idea of what we want to say, where the market is, we know our company, we know our product, we know our service, we know our, our email, newsletter, we’re confident. But are we are we actually we actually writing this email to ourselves?

Katie Robbert 10:43
I think that’s a really good point. And that’s that’s where I think a lot of regular newsletters struggle, you know, because we think we’re trying to reach well, and you know, so let me take a step back. One of the things that, you know, our good friend Ann Handley talks about is right to that one person. And I think while a lot of people try to do that, they may not realize that they’re writing to themselves, versus somebody in their audience. So like, No, I wrote to one person, it’s like, yeah, but you wrote to yourself, so yes, technically, you’re not wrong. You just chose the wrong person to write to. And I think that having your ICP, built in is going to be really helpful. Now, you talked about, and maybe I’m jumping ahead, you talked about generation revision? Couldn’t you use this as question answering and start to ask questions of your ideal customer profile? What kind of emails do they want to get?

Christopher Penn 11:44
Absolutely, you absolutely can do that. So one of the let’s, let’s pop over to our friend ChatGPT. And do that exact thing. So I’m about to start a new email newsletter, I need some ideas for what to write about, on the topics of generative AI analytics, data science, based on your knowledge, of the Trust Insights, ICP, what kind of newsletter should I produce? Now, one thing we have done, we have not put in our cost our company profile. So that’s missing from here. So you’d want to have that available as well say, Trust Insights is a consulting firm that does this, this this this, however, you have the latest trends and innovations, analytics, insights and best practices. Again, these are things that are based off of our ICP data driven decision making. So you could absolutely do that for ideas at a strategic level or macro level, or even at a hey, I’m about to write this week’s cold open. Here’s my thoughts of what I’ve written. ICP. What do you think of this? What parts if I missed that would be more satisfying to you?

Katie Robbert 13:05
Well, it’s not short of ideas, that’s for sure.

John Wall 13:07
No haha

Christopher Penn 13:08
It’s definitely not short of ideas. Let’s take another example of that fact. We’re going to take the text from this week’s newsletter. please validate this week’s newsletter content. What about it? Would our ICP not like and here is from the almost timeline newsletter, this week’s thing it’s going to go through and it’s gonna it’s gonna yell at me and say it’s it’s engaging, excellent for engaging, technically sophisticated audience. However, to ensure it aligns perfectly with your ISP, it’s like you wrote a two articles to technically dense like, that’s the ISP is not going to enjoy this. It’s too generic. It’s not focused on the applications. You’re over emphasized regulatory compliance, right? No one cares.

Katie Robbert 14:12
If you’re like me and ChatGPT are on the same page.

Christopher Penn 14:14
Exactly. Your call to action is too late in the piece. You’re so focused, you too focused on the technology and have not focused on business goals and ROI. So again, this is from the oldest time the newsletter which to be fair is sort of my pet project, which means that I don’t particularly care if I was to feed in an issue of the Trust Insights newsletter. We might be more concerned like oh, wow, we maybe we need to redo this week’s newsletter if this if this was the same feedback.

Katie Robbert 14:46
And if you want to subscribe to send newsletter, you can find it at trust where it is more closely aligned with our ICP.

Christopher Penn 14:58
Exactly. You know for them Record. Another thing you can do from a generation perspective, if you’re running low on ideas, if you’re running low on inspiration, a lot of marketers have what you call a swipe file, write a list of, hey, here’s the emails, things that I thought were particularly clever, or are writing those particularly clever. You could if you wanted to go into ChatGPT, or Gemini, or whatever, and actually upload your swipe file and say, here is some stuff for inspiration, things, take a look at the writing style, or the tone of voice, or neat language tricks that all these different examples use. And then here’s my email. How would you? How would you gussy it up with things that I’ve identified as really, really well done taking my email and sort of rewriting it.

Katie Robbert 15:55
Can I be honest, I have, I’ve heard the term swipe file before, but I never knew what it was. Did you? I’m guessing John, you knew what a swipe file was?

John Wall 16:05
That’s Yes, I have to give a shout out to Carol, my direct mail mentor who introduced me to that concept. Yeah, just use back in the day of direct mail, you just be getting all this stuff. And as you’d go through, you’d find stuff that’s way better than the rest of the pack. And so you have a folder at your desk where you throw that stuff into. And then when somebody comes to you needing a campaign, you can go to this folder. And you know, if it’s a different industry, or feel free to swipe all the great ideas from everybody else out there.

Katie Robbert 16:33
Oh, okay. Yeah, I blame me getting into this industry later than a lot of other folks. I heard the term I’d never known what a swipe file was until, right this moment I had, I used to work with a designer who had on her calendar, like appointments with yourself that said, tickle file, and I think it was the same kind of idea like things that tickled her that she wanted to borrow from.

Christopher Penn 16:59
Yep, exactly. Another thing that you can do with generative AI in the rewriting vein, if you don’t already use some form of email validation service, you should be right. email validation services are companies that where you upload your list, this is one example called MillionVerifier. You up your upload your list, and it tells you hey, this percentage of this is good, this percentage of your list is bad, nothing. So this is something that, for example, I do both with Trust Insights and with my own stuff. Every week, I feed in the the new form submissions from the website, because let’s face it, there’s a lot of bots that like to dump a lot of crap. When you get feedback from this, you get the bad list. Very often, you can take a look at that list that list and go oh, well, so and so just misspelled the word Gmail, right, like they typed G m i a l. So one of the things you can do is to have a model, evaluate the bad emails and say, fix any obvious errors such as misspellings of common email domain names. Big warning lights, make sure that you are using a generative AI model that with full privacy practices, meaning it’s one of the paid ones that does not train on your data that does not submit your data for human review. For example, if I pull up Gemini here within the Trust Insights workspace, you can see at the very bottom here it says your Trust Insights chats aren’t used to train our models. That is a big difference. From the consumer version of Gemini. The consumer version of Gemini says at the bottom, you might display inaccurate info but here’s your privacy statement. And you go into the privacy statement says yeah, where we were reading your email addresses or PII personally identifying information. So if you’re going to use general AI with that data, you must use one that has privacy protections turned on to safely do it.

Katie Robbert 19:03
I also want to make mentioned that sorry, John. In terms of email hygiene, we did an extensive episode on the live stream a few months back, so if you want to catch that episode, you can go to trust And again, find the so what playlist and you can find that email hygiene episode. Sorry, John, go ahead.

John Wall 19:21
No, I was just gonna say one thing after you have Jenai go through that list. Still, make sure you run it through the service before you just directly mail that because you want to, you know, even if you correct it, those could still be dead emails as they’re corrected or worse. There are email addresses out there that are honey pots. You know, if you mail them, you end up getting yourself blacklisted, and that’s a whole nother set of headaches.

Katie Robbert 19:45
Can you explain what a honeypot is? For those who don’t? I do know what this is like the one thing I do know but can you just sort of explain it on?

John Wall 19:55
Yeah, so the idea is that it’s comes from espionage the world of spy thing of having data out there that you know is wrong so that if that ever shows up, you know that somebody is doing something improper. So a good trick with that would be a company in their house list will have five or six emails in there that they know are not used anywhere. So if they ever show up out in the wild, they know that somebody has breached the data, and that there’s some kind of problem. And so a lot of the ISPs have this setup, where there’s certain addresses, if you mail these addresses, you’re proving that you’ve gotten some kind of stolen or broken list. And that will harm your reputation. So yeah, the honeypot is there to catch the flies.

Katie Robbert 20:37
Gotcha. Thank you. Alright, Chris, what have you moved on to?

Christopher Penn 20:41
So this is the implementation, I have taken this CRM data. This is synthetic data. We manufactured this data, you can tell by all the company names all being like super tech, LLP.

Katie Robbert 20:55
People are going to take that big Oh, that’s a good idea.

Christopher Penn 20:58
Yeah, we have ego where LLC, we have altra. NET, C, and a random name generator generator that generates random names. And we have course Yo, this is as if it was from you CRM, sure annual revenue, the number of employees, the day the last contact will accompany. And critically, the email address. So I have loaded in my this dataset, I said, Tell me. First of all, do you know what RFC 5322 is, which is the RFC for proper formatting of email addresses this, this is an international standard that everyone’s required to comply with. And then I’d say here is the data file, identify for me any email addresses that do not conform to this? So what you would do if you wanted to validate a bad email address as it came out of the verifier service? You would essentially do this and start talking to the language models say, Okay, well, which ones are bad? Why are they bad? Can you fix them? And it would go through and do that. So that’s an example of using generative AI to do that list hygiene and part of the list hygiene, you get the export from a hygiene service, and then try and fix the easy stuff, you know, misspellings of common domain names, Gmail, Hotmail, Yahoo Mail, etc, people using commas instead of periods in their in their email submissions.

Katie Robbert 22:20
Isn’t that, you know, just to sort of play devil’s advocate, isn’t that something you could do in Excel? Or in like a text editor?

Christopher Penn 22:28
You could, you could, and I we, in fact, for Trust Insights, we have a Python script that goes through and does programmatic cleaning of the most common misspellings and things. So you can absolutely process that this using a generative AI though means in this case, you don’t have to actually do the work of writing that code or the manual, because, you know, going line by line in an Excel spreadsheet. Okay.

Katie Robbert 22:51
I was, I was just curious, it sounded to me, because, you know, if somebody handed me a spreadsheet of email addresses, you know, and this is Chris, where you and I differ in terms of our usage, my first thought wouldn’t be generative AI, it would be to get into a text editor or to do some sort of like filtering or view lookups in Excel, to do that work.

Christopher Penn 23:15
Nothing wrong with that. Absolutely nothing wrong with that, if you if that’s what you’re comfortable with. You should 100% do that. If you’re a marketer, though, and you don’t know where to start, and maybe your Excel skills are not as strong as you’d like, this is not a bad option.

Katie Robbert 23:30
Well, and I feel like maybe the question is, so it’s not saying, you know, so that’s assuming, you know, to put in RFC 5322. Maybe, again, sort of what you always say, Don’t forget about the chat version of generative AI where it’s more of interactive, you can say, hey, generative AI system, I have an email list that I need to clean up. Where do I start?

Christopher Penn 23:58
Hmm, that’s a great way to handle that. Let’s say like I What is the valid email address? And have it go through email?

Katie Robbert 24:06
Or maybe what happens if I don’t have a good list of emails? Is that going to be harmful to my deliverability?

Christopher Penn 24:13
Exactly, exactly. Or even things like taking a screenshot of your email marketing interface to say, analyze this, like, how is how is my email marketing been doing? These tools can see. All right, so next, let’s go ahead and flush this out here and clear out our system problems. I want to start a brand new session. Sometimes you want to do summarization. Sometimes you want to understand what’s happened in email so that you can maybe classify it, get some good ideas, or in many cases, learn what not to do. So let’s say I’m going to start typing here you talk amongst yourself, we’re going to be using the form submissions from the market. over coffee email list from our website, where people send us pitches all the time, we’re going to look at those. Yeah, I’ve anonymized them. We’re going to look at those and try and characterize the way people are pitching the bargain over coffee podcast.

Katie Robbert 25:14
John, this has got to be your favorite game. Yeah, exactly

John Wall 25:18
Yeah, exactly. This is going to be we’re gonna get a lot of rotten pitches of people whose stories no one would ever want to hear. That’s pretty much the the first grab. But yeah, this will be interesting to see what comes up. Because it’s, you know, there’s a wide array of stuff between folks that actually have an interesting story to tell, versus people just trying to sell us junk. And then some user comments in there. But it’s all pretty much automated junk. I think Chris, and I both got into decision database at some point. So now, just the whole world continues to spam us on a regular basis.

Katie Robbert 25:52
I think what’s interesting about those, you know, because obviously, you know, there are companies and agencies who are using those automated bots, is they like, this is the thing, and maybe you know, this, John, is that, you know, they’re sending out hundreds of 1000s sometimes millions of emails, or are they actually getting responses to these like is that the idea is that, let’s use the bot to send out, you know, 1000 automated on personalized emails. But if we get two people to respond, then we’ve made our ROI.

John Wall 26:29
Yeah, that’s it, you totally hit it. You know, this is what the number one complaint with like email and social networks is that it ignores a lot of the rules of human community communication, you know, the fact that if you send 500,000 And you get two hits, well, those are two wins. So that’s good. You there’s really no you’re not, there’s no pain for the other 499,998 people that you pissed off. So chief statistician, right there. That’s yeah, we’re down to the decimal. But yeah, so that’s the thing, you know, and yeah, it you know, it does work. That is the truth. If they get a couple hits, and then they pass those on to the client, the clients happy then. Yeah, the Spam is just the cost of doing business.

Christopher Penn 27:12
Okay, John, out of 64 pitches. This is like the last 60 days. 48 guests pitches, six spots, pitches to collaboration pitches, five late sales pitches, and three bits of spam.

Katie Robbert 27:30
Yeah, a really bad version of the 12 days of Christmas.

John Wall 27:35
Exceptionally painful version. Yeah, and I that’s pretty good. Like 10 to one ish of guests to sponsors. That’s, in fact, that’s probably a little worse. Like, that was actually good day. We’ve had some more than normal sponsorship pitches in the last month. But yeah, that feels dead on there’s that, you know, that feels good. And again.

Katie Robbert 28:00
So what does one do with this kind of summarization? Like, what’s the use case for this, Chris?

Christopher Penn 28:06
So, summarization of an inbox? Well, let’s continue on this. Because I wouldn’t want to do is I want to talk about scoring. So you’ve got this big archive of existing emails. And the when I started off by priming the model with the Trust Insights pair framework, and I said, Okay, what are the criteria of a good pitch, and this is good pitch is relevant, has good value proposition for the audience, concise, professional, and it has a call to action. So writing a now writing the follow on our show. The marketing over coffee podcast is a show about marketing, marketing, technology, tech, and whatever else. And Marketing News. We focus more heavily on B2B marketing in general. Use this criteria plus your own knowledge, no marketing podcast to score the pitches Okay, let’s ask it now. Great. What do you know about the marketing over coffee podcast? Let’s see if we’ve done decent SEO over the years and content creation and Gemini knows who we are

John Wall 29:41
Oh, there it is. Nobel Prize winning marketing over coffee podcast is known for.

Christopher Penn 29:52
Not bad, not bad. We need to be accurate. Exactly. So now we can say

Our goal is the marketing over coffee podcast. So use the knowledge you have shared to judge, score and present the aggregated summary of pitches that meet these criteria. So we’re gonna have it score this. And to your point, Katie, why would you want to do something? Why would you want to assess an inbox? artifice to understand what is being done in the marketplace right now? It’s almost a competitive intelligence to say, Okay, well, suppose we wanted to pitch the In-Ear Insights podcast to other people. How, how would we go about doing that won’t be a pitch that we could develop, and then we’d have this corpus of these pitches. So here we have the scores number pitch to me in criteria 44 out of 64, so 68% of relevant pitches 59% of a value presuppositions fully 50% of concise 90% of professional 84% of call to action. So knowing this, the weak spots are relevant value proposition and conciseness. If we wanted to write good pitches, those would be the criteria to lean on, because we know that what everyone else is getting in their inbox isn’t.

Katie Robbert 31:29
This might be a I hate to say dumb question, because that’s there’s no such thing. This might. So wouldn’t common sense tell us that our pitches should be relevant and should be concise? I guess my question is, why would one need to do this analysis to know that your pitch should have the following elements? And maybe this is where I’m just like, I get baffled by things like this.

Christopher Penn 32:00
Common sense should suggest that however, common sense clearly is not all that common, if only 50% of pitches are concise, fair.

Katie Robbert 32:12
Fair, solid answer.

Christopher Penn 32:15
So this is an example of summarization, classification. And question answering a non email marketing use case for this that by the way we use all the time is taking the contents of an inbox for say, a particular client, and putting that in, and then summarizing, like, here’s what we did for this client over the last month. It’s a way to roll up activities. It’s super powerful, and also is a good way of figuring finding things you forgot to do. So that is classification summarization, question answering other things that you can use gender AI for our technical help. Right, so let’s, let’s flush our session here, asked, What do you know about email deliverability protocols, such as SPF D, Kim, and demark. You’ll recall from the email deliverability episode we did a while back, these are the three big protocols for doing this sort of stuff. And it turns out that the large sophisticated language models really clearly know this stuff. Great. What is the correct format for an SPF DNS entry? So I’m just going to ask these questions, kind of what you were saying earlier, Katie, but the chat and ChatGPT then if you’re unclear about how to do this stuff, and maybe there’s you don’t like the output of the wizards, you want to double check, you can take those deliverability deliverability settings out of your DNS, feed it to a language model and say, Hey, validate this, check this out for me, does this make sense? Did I do it? Right? Is it compliant with the standard? Right, that will? It’s a great straightforward way to have a second opinion on your deliverability work?

Katie Robbert 34:11
With with questions like this, so you know, email marketing, you know, you’re very well versed in it, you know, the technical side of it, I if you said what are the three email deliverability things that you just listed? I would be like, I don’t know. And I would kind of like make up some letters. So how, how do I because I’m not in the technical weeds all the time. Like now that I’m seeing him I go, yeah, now I remember. But it’s not information. I keep top of mind. So you know, that the information in here is correct. But how would I know? Versus if I just did a Google search and said, How do I make sure my email gets through to people? Like how do I know that this is the right information? Because we do know that generative AI can hallucinate So maybe instead of SPF, it says SFP. And I’m like, Oh, that looks close enough. Ask it.

Christopher Penn 35:20
So I’ve said, let’s talk about email marketing, you’ll be an email marketing expert, I’m having trouble with email marketing, getting emails delivered, I’d like you to coach me through diagnosing my delivery issues. Does it sound good. So it says, understanding the fundamentals. deliverability versus deliver delivery is the technical process of an email, reaching a mailbox. deliverability refers to whether it lands in the inbox rather than the spam folder, many factors influenced deliverability. Let’s start investigating basic checks, typos, spam filter, trigger words, bounce rate, spam, complaints, and so on and so forth. So this can essentially start to walk you through, diagnose this, and you can say, hey, I’m a non technical person, I don’t know how to doing this. So I’d like you to give me step by step instructions for how to check my spam complaints. Or how to check my DNS. I don’t know i know i we pay for WordPress, is that it? Let’s say no, let’s that has nothing to do with email. And say we have a I bought my domain name through GoDaddy, it was a great, here’s how to start finding those settings within GoDaddy, or here’s some technical articles. So using the tool as a coach, as a, as a second opinion, is a great use of it, especially if you start asking it for more articles and resources. A tool like perplexity, for example, would be a great choice for this. Because what it will, what perplexity will do is it will find you the actual articles to read.

Katie Robbert 36:48
And I think for someone like me, well, and, you know, I feel like I’m fairly representative of a lot of marketers, I would be looking for a tool like perplexity because I also want to see like, where are you pulling the information from? What are the articles? What are the videos that I can watch, to walk me through this, because that’s one of the things that at least to my knowledge, generative AI doesn’t do is it doesn’t generate a video for you to watch when it says these are the instructions, a lot of people and I’m putting you into this category, Chris, prefer learning by watching a video, whereas I prefer learning by reading a list of instructions. But you have to cater to both kinds of learning styles and I don’t believe generative AI itself creates those walkthrough videos.

Christopher Penn 37:39
No, it does not generate does not create those walkthrough videos. But in many cases, if they exist, a good AI enabled search engine will find them present them to you.

Katie Robbert 37:52
That makes sense.

Christopher Penn 37:54
Another one is just asking these things questions like, Hey, I saw an article about this find me additional sources. So there was a recent piece, I forget who which publication was in, but said hey, using emoji in your subject line increases open rates. And it turns out that there are both there are sources that both agree with and disagree with this entity. Turns out it depends on your audience. So one of the next things you would do is say, Go back to your tool that has your ICP loaded in it and say, based on this research, and you might want to to copy and paste some of the articles into it, and your knowledge of our ICP. Good. I start using emoji and our subject lines based on everything you know about the demographics and the firma graphics and the people that you are in your community. It might say, Absolutely, you should be using emoji. In fact, you should make your entire subject line emoji or by saying nope, your audience will just be turned off by this this is this is not appropriate for your audience. Don’t do it. But, again, really emphasizing using that ICP will help guide these decisions.

Katie Robbert 39:06
I think that is a question that we can do offline that we absolutely should ask both so for those who aren’t aware, Chris and I are constantly at odds about the use of emoji in email. I am anti emoji Chris is pro emoji. And John refuses to tie break.

John Wall 39:24
Know the one thing you know we spent so much time with bureaucrats and any company over $20 million you know when they see an emoji they’re just like, whoa these crazy kids that’s risky you know if we’re selling stuff on Tiktok then yeah, you know emoji it up but for our folks, our boomers know the way too crazy

Christopher Penn 39:57
That covers the city next major use cases of generative AI generation, you know, write the email, write the subject line, et cetera, et cetera, with your ICP, or using this swipe file as inspiration, help me do this thing, extraction, get data out of other things and just think, Oh, I forgot RFM analysis.

Katie Robbert 40:24
I was just thinking the same thing. Clearly not. Clearly, you know, it’s okay.

Christopher Penn 40:30
RFM analysis, if you’re unfamiliar, is recency, frequency, monetary value. It’s a classic analysis tool, particularly for retailers use it a lot, but you can use it in email marketing, and you should be using it in email marketing. So let’s take a look at what an RFM dataset would look like. In email Marketing. I have here, same synthetic data set. So this is the so if you if you’re trying to be evil, and you’re gonna copy and paste two emails, a share on screen, you’re in for a rough surprise. And I have three different metrics, I have the number of emails that addresses opened, I have the number of days since they’ve opened the last email and have the deal size from our CRM. So recency, frequency, monetary value, if you have information like this, you can feed it to a tool like ChatGPT has advanced data analysis, or have your any language model, generate the code necessary to perform RFM analysis. And what that will tell you is it will tell you, we’ll help you cluster your database into people who are maybe frequent openers, but don’t buy anything. People who are who have bought something, but haven’t you haven’t heard from him in a while, or people who just occasionally show up. And then based on that, you would have email marketing strategies that map to each segment. So if you have people who have bought from you in the past and have a high monetary value, but have not been recent or frequent, you know, that’s something you hand off to your best sales folks to say, Go get them back, they bought a lot of stuff on the past, they probably still need to somehow go get them back. On the other hand, you have folks who are like, big openers, big, frequent openers, but they never buy anything. Those might be your evangelists. Those might be folks who they can’t afford, whatever it is you’re selling, but they’re they love everything that you create, find a way to work with them as influencers, and say, Hey, can you help me share this newsletter? Can you help me share this email newsletter with your audience, and get in front of people and buy something? Right? So that’s, that’s a technique that you can use. What did I do with I had some Python code that I’m earlier about. So I had Gemini, go through and start framing out an application to ingest that data set and then produce the analysis and the different clusters so that later on, you can run it. So you can either do it in ChatGPT natively, which will take a long time. And if you have a lot of records, it doesn’t do very well, like over 1000 records, it does not do well. Or you could have the tools build you code to do it for very large datasets. So for example, the almost timely newsletters close to 300,000 subscribers. ChatGPT is like, wow, I’m not doing this. But Python code will have no trouble with it.

Katie Robbert 43:30
Well, and I think it’s worth reiterating that in order to do this kind of analysis. With statistical tools, or with generative AI, you have to have really good data governance and very clean data to begin with, you know, generative AI is not going to fix your data governance problem. In this instance, it can give you some ideas of how to approach your data governance, but you still have to fix it. So yes, you can use these systems to do to help with your analysis, but you have to have clean data to begin with.

Christopher Penn 44:03
Exactly and good. Good software, good email marketing software. If your email marketing software doesn’t spit this data out. It’s you’ve got a problem. If you’re if you don’t have a good CRM, you’ve got a problem this is not going to work for you. So generation extraction, summarization, rewriting, classification and question answering of the six major use case categories of generative AI we’ve talked through things like building your ICP, to do the generation or to do rewriting, extraction to fix bad emails, summarizing your swipe file or summarizing pitches already gets you get some competitive intelligence going, classifying the contents of an inbox or classifying again, your swipe file, doing RFM analysis, validating your DNS settings to improve your email deliverability. And ultimately, using that ICP with your generative AI to ask questions like What do you think of this? What do you think this is? Have this content calendar for the next quarter. What do you think about this email the subject this paragraph, use these tools. As advisors right there, really, you can program them to be great advisors.

Katie Robbert 45:14
And we don’t have time to get into it, but I will make a quick pitch to also use the five P’s. Once you have your ICP, run through the five P’s to figure out the purpose of creating a newsletter in the first place. Do you want it to be an awareness tool? As Chris mentioned, it can work as awareness but it’s better if it’s further down the funnel. So you have to really understand why am I doing this in the first place? Am I creating a newsletter because everybody else has one? Or am I creating one because I have value to add to my ICP? So just my quick five Ps pitch.

Christopher Penn 45:48
Exactly. John, any final parting thoughts?

John Wall 45:52
You know, identify yourself as promo G or con emoji and get the testing running.

Katie Robbert 45:59
Con Moji over here. I don’t think that’s a word, but I’m calling emoji.

John Wall 46:03
Well I’m pro emoji. A really good cat emoji is the best I could come with that. I’m not satisfied with that. So what about anti emoji? Yeah, anti emoji. It just doesn’t hit like pro emoji. No emoji. No emoji.

Christopher Penn 46:17
There you go.

John Wall 46:18
That’s an upgrade. I’ll take that’s the new new champion.

Katie Robbert 46:21
Hashtag emoji.

Christopher Penn 46:22
Got the rhyming? Emoji no Mojo merging.

John Wall 46:26
Oh, that rhymes too. That’s that

Katie Robbert 46:32
Haven’t read that.

Christopher Penn 46:34
And with that, thanks for tuning in. Folks. We will see you next time. 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 I podcast and a weekly email newsletter at trust Got questions about what you saw on today’s episode. Join our free analytics for markers slack group at trust for marketers, see you next time.

Transcribed by

Leave a Reply

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

Pin It on Pinterest

Share This