In this week’s In-Ear Insights, join cofounders Katie Robbert and Christopher Penn as they discuss the release of Chris’ latest book, AI for Marketers (Second Edition). Learn what’s in the book, which jobs will be taken by robots, and how to set up your career for success. Also learn why social media is in mortal peril due to AI. Tune in now!
Get the paper version here (more expensive because it kills trees).
Listen to the audio here:
- Need help with your company’s data and analytics? Let us know!
- Join our free Slack group for marketers interested in analytics!
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
This is in In-Ear Insights, the Trust Insights podcast.
In this episode ofIn-Ear Insights we are talking about all things artificial intelligence in marketing. So Katie, what’s on the docket for today with AI and marketing? Well today in AI and marketing Chris you just launched your new book. So I figured we could spend some time talking about it. So what what’s the title of the book? What’s the book about so it is AI for marketers and introduction and primer second edition and funny story at the regional was actually a series of blog posts that Ann Handley over at marketingprofs asked for, but it’s there almost three years ago. And she said, Can someone please explain this in something that doesn’t involve like, like super complicated math? And so I wrote a series of 10 blog posts and then
shortly thereafter, just like this be a great book. So
at the time we were
You know, in a, in a different company and stuff and, and schedules and stuff made it so difficult to put things together that I literally copied and pasted the blog post together. So if you are an owner of the original version, you’ll even see references like in the next blog post you’ll so
as has become tradition over the holidays, I like to write something new and I figured it was time to to edit and bringing it up speed, helping the document empathy. The first version I started reading, I’m like select all delete, we’re just going to stop this is not going well. And it ended up being a complete and total rewrite of the original with much greater focus now on
how AI solves marketing problems as opposed to like the cool technologies that are under the hood like nobody really wants to learn about like soft max drop out late and drop out layers and stuff in a neural network not in Mark
Kidding. And so this book is much more business focused problem solves. Are you going to lose your job to a robot? Things like that? So what are some of the problems that marketers are facing today? That AI would be a good solution for? We have.
Well, there’s so many,
what are what are some of the more common tangible problems that someone could get started right away solving? There are there’s a category there’s five categories of problems that we as marketers desperately need to solve, and sooner rather than later for most companies, we call these the five us which because I just like making
fun, fun frameworks. The first one is untapped data. We are sitting on so much data and the reason it’s it’s untapped is well, there’s a couple different reasons. One is a lot of its unstructured. So in your customer service inbox and your CRM and your email system. There’s all this data
But the most marketers don’t have the tools to extract all the information out of it so that’s Problem number one that we can solve with artificial intelligence number two is unknown I identities unknown influence especially on influencer marketing and social media take Who
Who should we pick paying attention to. And a lot of the influencer tools on the market are still pretty primitive. There are ways to use machine learning to figure out in a better, more scientific way, who’s influential. The third is unclear data. And this has a lot to do with the fact that we collect a lot of data, but we don’t necessarily see the relationships between different data points. So real simple example in SEO. One of the things that you do when you’re building a keyword list is you look for
how search does this term and another thing can look at us like how expensive is it or how difficult is it to rank for those terms, but we never actually sit down and plot out the relationships between those two variables using machine
Learning using clustering you can figure that out and go, Okay, I want to prioritize stuff that’s get me volume but won’t be like super hard to rank for. The fourth problem is, we’re very unfocused in marketing is we have a lot of back the truck up for the data on the desk syndrome. That doesn’t help anybody. So we, again, we use machine learning to identify what are the important things to pay attention to in our data. And the last one an area of expertise that that you talked a lot about is that marketers are really unprepared. They’re very reactive, they’re scrambling at the last minute to try and put stuff together. And with machine learning specifically predictive analytics, you can forecast accurately for the future and be able to plan ahead. So those are the big problems that that marketing can solve with machine learning.
So okay, so we talked about untapped unknown, unclear, unfocused and unprepared those are the five us
That marketers face today. If I were a marketing analyst in a smaller company, and I had Google Analytics setup, where would I start? How? What kind of problems? Do you think that someone in that situation is most likely facing? What say attainable solution from AI? Because I think AI in general, people get really nervous and scared that it’s going to be a very large undertaking that it’s going to be a very expensive undertaking. What are some of the
more simple ways and I’m putting simple in quotes, what are some of the ways that marketers can get started?
So there’s two answers here. Number one is, in some ways markers won’t have to do much because true AI is finding its way into more tools. So if you look in Google Analytics in the upper right hand corner, there’s an intelligence button that looks cool but little swirly thing
and that’s Google using its own AI to look at your data and
Hey, here’s some things I noticed Did you know this and and it will alert you. If you have the mobile app, Google send you notifications, okay? Something weird is going on any website like I’m not receiving any traffic today
that’s those are good things to know, if you are an analyst at a company and you can get access your company’s analytics, you could literally be able have your phone be to have its finger on the pulse of of your company’s data. So that’s a good starting point. A second place to look at is there are some tools that at least from a data and analysis perspective you can get started with. One of my favorites personally that requires no coding is Tablo public you can take data out of Google Analytics and do some of these tasks like clustering and just explore the data. Be curious about it, see what’s in there. And then the third if you are feeling a little more ambitious, still no coding but you you will have to learn more of the lingo around AI would be something like IBM Watson studio where you can sign in for
At the free tier to get like 50 hours of processing time in a month, and you can start loading some data and poking, just dragging and dropping different types of machine learning together, the see what happens. Now, again, you need some background in what the little colored blocks do. But you’re not still not writing code, you’re experimenting. And I think all three approaches over approaches that anybody at any point in their career could start experimenting with.
So one of the things that you just hit upon is really important. And that also brings me to another question I have for you. So you said
that people marketers specifically need to be curious how much of your book touches upon the people aspect of AI and how important is it to implementing AI
the people aspect there’s an entire chapter on your journey to AI and how your company will make that journey the types of people you’ll need and what you should be doing in your own career to to
insulate yourself from the effects of AI within a company. The people aspect is super important. It’s one of the three pillars of making anything work, right? People, process and platform.
the people don’t have the talent of knowledge. So they have to, if you are that employee, you have to be the person who’s proactively go out and getting the training yourself. There’s a chapter called down the rabbit hole if you want, if you do want to learn this stuff, there’s some great free courses, many of them offered by MIT. So can’t get much more bespoke than that.
to level up your own skills if you want to go that route. Now, that said, I think it’s really important to say that marketers don’t have to become machine Linux specialists in a lot of ways. I the analogy I use is like cooking huge surprise, right?
You don’t need to be a professional chef to know what great food is. You don’t have to be professional chefs know what bad food is. You can walk into a restaurant and you’ll immediately see who would want to eat
Here or give it a shot or your favorite place. Like I can’t wait to go there, you know the output and you know what you’re looking for. The same is true of machine learning and artificial intelligence. You have to know your goals, you have to know your strategy. And you have to know what the output should look like. And that’s people in process stuff as opposed to platform stuff. And then either vendors will provide the platform or maybe you know, other parts of your company will provide the platform but you just have to know what is the goal you’re trying to accomplish what’s the plan to get to that goal and then how can you use AI to and and the types of problems AI solves to get there but again, huge surprise and we sound like a broken record, but it really does come down to planning and goals.
People on the listening to the podcast can’t see me cheering, but it really does start with a plan. So Chris, what are some of the things that people get wrong about AI? What are some of the misconceptions the number one misconception by far
And we say this a lot. We say this in every talk we give on stage AI is math, not magic. It cannot make things happen that it does not already understand. When you use AI, you’re providing what’s called training data to tell the complete the software, what to do, how to learn from it. If you provide no data, the machine can’t create anything.
It’s the same as any manufacturing process. No input equals no output, bad input leads to bad output. So that is that’s number one. The number one misconception the number two misconception which is a close second is that AI in its magical powers is always correct. And nothing could be further from the truth
again, garbage in, garbage out. You put bias data in, you get unbiased data out there’s no getting around that you can you have to spend, you know 80% your time can be spent on data preparation on model preparation on feature engineering on making your data as clean as possible. There’s an entire team
chapter in the book on what good data is. There’s another chapter on what biases and these are problems that we as an industry as an in machine learning and data science need to solve. But marketers especially Keep your eye on the goal, keep your eye on the output, if the machine spits out something that does not pass like a basic common sense test. Don’t assume machines, right? Say that doesn’t make sense. Like your machine. You’re telling me I should only market to 36 to 45 year old white males if high income backgrounds but our product is intended for, you know, African American females 25 to 34 and like know the machine is clearly wrong. In this case, something went wrong with training data.
What’s interesting is what you’re saying is the analysis piece of it is actually the smallest portion that it’s the preparation. It’s the continual refining and it’s the continual pulling out insights and double checking that actually takes
All of the time. And that’s where people are still needed to perform those functions because the machine can’t really check itself. And I think that that’s another misconception is that AI and any sort of automation sort of replaces the human all together. But really, it’s just taking out that small part of the actual analysis. But you still need people looking at the data, feeding the data, creating the data, and then interpreting the results that come out again, look at cooking shows and look at the cooking process, right. How much of the dining experience is the actual cooking versus sourcing vendors, sourcing ingredients, the preparation, the getting it be as ready, making sure that the greens are good and not spoiled the recipe that’s the chef’s 20 or 30 or 40 years of experience, the serving staff the atmosphere, the you know, the prices on the menu, but like you said, the cooking part is actually a really tiny piece of that process. Now, ai makes that part go really fast.
But you can’t make everything else goes super fast. You can’t build a restaurant in record time, right? You can’t make crops grow faster beyond a certain point. So there’s there are to your point, there’s so many limits on what the machines are capable of today. Now that said, that will change over time, we are already seeing indications of machines being able to do some incredible new things with particularly around content creation that in the next two years are going to drastically change marketing. And this is this part is not in the book. So this is a podcast exclusive.
I was listening to an episode of this week in machine learning on my drive yesterday and one of the foremost experts in natural language generation said social media as we know will be gone into years. He said, natural language generation is getting so good that it will be trivial for a a hostile actor to create bots that generate true authentic human sounding.
Fake information that social media will basically be unusable because you won’t have any idea what’s real. And what’s not
a hostile actor, like the, like a hostile actor like Russia, for example, attacking the United States or us attacking China, or Iran’s things. And also, you know, non state actors, like just hackers and things, coding
and creating bots. Today, you can still spot bots. There’s the technology, relatively primitive, but with some of the major advances have happened in natural language processing in the last six months. It will not be long before you can create a bot that will spread disastrously bad Miss information but it will pass the sniff test as human and it will it will come across sounding like it’s authentic. Like for example, this is a silly example
we know that 4g technology the powers your smartphone today operates a certain frequency of the electromagnetic
spectrum now 5g is double that. So you’re doubling that frequency
just the way I say that sounds like oh my god, that could be a bad thing. Right? It’s not there’s there’s very little impact of because it’s all on a certain part of spectrums non ionizing but if I were to create a bot was talking about the dangers of super high frequency microwave radiation and how smartphones should be banned, it sounds credible, right? And that’s something you can generate with machines you could generate millions and millions of tweets and Facebook posts and Instagram photos using adversarial networks to essentially astroturf any issue so things like politics in elections these will social media will have to do one of three things it will you have to go fully paid full or fully verified where you have to like upload a driver’s license just to be able to use Facebook or all go private were the only people up because you speak to the people you explicitly let in your community and you see nothing else.
In order to escape this, this rabid dystopian vision, so there’s,
there’s a lot of on both sides of the things that AI will do that will make marketing great. And the things that less ethical actors will do to make a terrible.
Funny side note, my dad is already creating his own dystopian version of Facebook, where he joins groups just to block people so that he can only see the things that he wants to see. It’s a very bizarre behavior. But it was I thought it was hilarious because I was like, Dad, what are you doing?
Like, well, I just want to join so that I can see what dumb things people are saying. So I can block them like, okay, so he’s systematically creating his own little private version of that, but I can definitely see where that’s a useful thing because
people spend so much time getting their news from Facebook, from Twitter from all of the other social channels.
They can’t distinguish between what’s real and what’s not. And so much of that content now is generated by bots. And I think people don’t even realize that or things are auto posted Facebook for a little while, tried to create that
button that told you the source of the content, but it’s really unhelpful. And so I think it’s definitely an interesting thing
in terms of the future of how AI is going to impact our daily life, not just at in business or in marketing, but really just within everything that we do as consumers absolutely in those private community is going to become more and more important where the people you let into the community are trusted by the way, quick plug. If you want to talk about analytics or AI for marketers join the trustee insights slack group trust insights AI analytics for marketers it is a private group that we’re running completely free with.
You can ask questions like, well, when will my job be taken?
And so Chris, as we start to wrap up, and congratulations on the new book, what is one key takeaway from your book that you would want people to know that they can get the key takeaway? I like to call this book sort of the you’re a useful bs detector for AI, especially in marketing. If you are working with or evaluating vendors. every vendor is going to say that we’ve got a on our product. This book will help you build the lexicon and the understanding of the concepts you need to be able to say to the vendor, no, no, go into more detail. What exactly what kind of problem are you solving with AI? Is it a regression problem? Is it a clustering promises a classification problem? Is that a dimension reduction pump Tell me more? And you’ll very quickly see the vendor go oh, I don’t know it. Just that sales just told me to say it’s AI and they’re like, cool. You don’t know what you’re talking about?
Oh, dear. We get that a lot, don’t we?
Yes, we do. And you know, so I hope this book helps you become sort of the vendor whisper as
of when it comes to things like AI. So Chris, where can people get copies of your book? So this book is available. If you want the environmentally friendly version, go to AI for marketers. book.com, and you can get, you’ll get the ebook version for Apple books and stuff. There’s a Kindle version. And then there’s a PDF if you just want the desktop PDF version. If you would like to kill a tree and destroy the environment, you can get the paper version on Amazon.
Excellent. any parting words? Thank you to everybody in the trust insights community who has been an instrumental part of helping this book happen because without people in the community without you listening without the team without our customers who ask really good questions all the time. This book would be much less informative because you can’t write something like this in a vacuum. You need to have customers and people asking questions all
day long like, hey, how do you do this? Or how would you solve this without the community? It wouldn’t have happened.
Well, Chris, again, congratulations. publishing a book is a big deal. So if you’re looking for Chris’s new book, you can go to Amazon and find the physical copy. Or you can go to AI for marketers book calm. As always, you can find us at trust insights.ai or at trust insights on your preferred social media handle. Thanks a lot, Chris. Thank you, and we’ll talk to everyone soon.
Thank you for listening to in your insights, the trust insights podcast. Please ask a co worker or colleague to follow our show on Google podcasts. Apple podcast. Wherever you listen to your shows got a question like us to answer Watson help solving your data and analytics challenges visit us at www dot trust insights.ai today
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, Data in the Headlights. Subscribe now for free; new issues every Wednesday!
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new 10-minute or less episodes every week.