This is my own personal journal of what I’m learning about AI and ML for marketing. I have over twenty years' experience managing and marketing tech companies, including deep knowledge of marketing analytics and attribution. For the last 9 years I've also been advising others in this space; so I focus on what I think the impacts on my profession will be, including privacy and ethics issues. You can read more about me in the "About" section of this Substack. Your suggestions for further education on AI and ML are welcome.
In This Issue:
Book Report
What I Learned This Week
New Section! This Week’s Bookmarks
This Week In Privacy and Ethics
Playing With Midjourney
New Section! Celebrating Humanity
Book Report: “Artificial Intelligence For Marketing” by Jim Sterne
Way back before dinosaurs roamed the Earth, in 2017, Jim Sterne was way ahead of a lot of us when he wrote “Artificial Intelligence For Marketing.”
Professional Disclosure: I received this book for free at the Marketing Analytics Summit in 2019. Embarrassing Disclosure: I didn’t read it til 2023.
Given how fast the space is evolving, it’s easy to think such a book could be outdated, but I felt that my knowledge needed some basic training. I thought at the very least a well-structured introduction to the topics with a marketing lens would be helpful.
I wasn’t wrong. And the book is still very relevant, especially if you are new to the space, or if you are already a data scientist who wants to understand marketing in a no-nonsense way. I have made dozens of notes in pen, scrawled all over the pages. For me, the book’s eleven chapters coalesced into three sections.
Part One: Intro to AI and ML for Marketers and the Data Science People who Love Them
Chapters 1-3 are almost 40% of the book, and with good reason. They are a compact guide to AI and ML that is accurate and timeless, but presented in a way that is not scary or overwhelming for anyone who’s spent time in the technology and/or marketing fields. In fact, having this information broken down and explained so well will take some of the fear out of the subject for many people. We know what this stuff is, folks: it’s just a computer program. Don’t let the marketing of it fool you into thinking it’s anything else - at the moment.
What makes these chapters so fascinating is that Jim is able to put himself in the shoes of both the marketer and the data scientist and help each other understand what the other is facing and how they are approaching problems; as well as how to work with them. So after Chapter One’s overview, Chapter Two talks to the Marketer about Machine Learning. In Chapter Three, Jim talked to the Data Scientist to help them grab onto what we marketers deal with.
Much of Jim’s text throughout the book includes citations and long quotes/interviews from notable authorities in the field, as well as the technology companies themselves. This generated a long list of “blog ideas,” “go and read this,” and more for me. I appreciated the citations as they helped me trust where Jim was going and identify people and ideas I wanted to follow up with to learn more. In particular, Jim’s background as one of the original web analytics influencers helps him create excellent frameworks for explaining how AI and ML will impact the exploration of data for decision support.
Ah, data. As Jim calls it “Machine Learning’s Greatest Asset.” He then goes on to explain how to be a great data analyst, cleverly disguised as a review of “how we used to dive into data.” Here’s where Jim and I might disagree-or maybe it’s just an editorial decision to continue his life’s work to educate everyone on great analytics in a very sly manner. For example, I don’t think the things that make a great analyst will change at all. “Asking really good questions,” as he calls it will never go out of style. What I see is the exciting opportunity to let the actual analysis shine, instead of bogging us all down in data cleansing. And as those who know me best will attest, I love discussing the role of cognitive bias in any discussion of decision support, and here too Jim gives a concise overview.
From there, the book goes on to outline the myriad number of data sources becoming available-even in 2017-and explains the role of data collection by many companies that go well beyond the most publicized reports.
In short: if you are new to marketing, or if you’ve not been in the weeds for a while on marketing data, then Chapters 1-2 alone are worth the price of the book by themselves.
In Chapter 3, Jim speaks to the data scientist. As a very fussy marketer, I can report that the overview of our needs and concerns(from brief explanations of the Four P’s, through marketing mix modeling, and customer journey theory) is simplified, but correct. (So I’ll assume the part on machine learning is, too).
Part Two: What Can Marketers Do With AI and ML
The next 25% of the book is a detailed exploration of the practical applications of AI and ML to marketing,s starting at the top of the funnel (“Getting Their Attention”) and proceeding through retention. Here, Jim shows us actual companies working on the problems at hand. Setting aside the fact that the specific technologies may or may not be with us, I was frankly mad at myself for not reading this book sooner just so I could get a picture of how things were already evolving.
You don’t have to be on top of every latest toy, to read these sections and start to see how our entire world in Marketing is going to change in every way. I suggest you pick the issue that relates most closely to the problems your organization has right now, and focus on one area that is changing. For example, if you’re struggling to retain customers, see the in-application bots list and go check out the ones that are there and try to find the new ones that have arrived.
Personally, I have these chapters book marked as references and starting points as I work my way through teaching myself about the subject. You’ll see me referring to them in issues to come.
Part Three: Marketing Operations, Ethics, and Prognostication
The last 35% of the book is also worth the price of admission. Here, Jim talks about the tools that are transforming how marketing teams are being organized and managed, as well as providing a framework for how to start onboarding AI into your organization. Important: Don’t panic, you are not late! We are right on time, but we must not put it off any longer.
Jim provides countless practical real-world examples of how many kinds of companies are already integrating AI and ML into their work. This section will help you start to understand AI and ML beyond the theoretical and apply it to your work. His chapter on helping marketers onboard AI contains a well-integrated smorgasbord of ideas from a variety of sources that pull together key concepts in a helpful way.
As someone who has been concerned about ethics and privacy issues for some time, I also appreciated Jim’s structured evaluation of the risks and challenges of Machine Learning and Artificial Intelligence. He explains concepts such as Overfitting, Correlation vs. Causation, and Hallucinations. He explains in concise prose that the ethics of these technologies is a very personal and moral one, but doesn’t shirk from demonstrating how this technology could bring about real and in some cases irreversible harms. Jim then wraps up the book with some thoughts about where the world might go, and even 6 years later I don’t think we’re close to answering them.
Yes, You Should Read This Book.
After all, you’re reading this Substack. So if you’re like me, and you want to find a structured way to understand the basics, without having to watch tons of videos or suss out your own blog posts, pick up the book.
If you’re very technical but not a marketer, you’ll also benefit from reading this book. You’ll do a far better job of supporting your colleagues and adding real value to them and your company’s bottom line- if you understand how they think and what their problems are. After all, empathy is the core of every great marketer’s skill set.
What I Learned This (Two) Weeks
1. It’s about tools. And it’s not about tools.
As I expected, everywhere I look someone’s hawking some ChatGPT-enabled tool or plugin. And yes, I’m coming up to speed on these tools and I will have something to say about them.
But it’s also not about tools, as all of my clients will tell you I’ve been saying for a long time. Most of the problems we face in business and in marketing are people problems. Just like installing Google Analytics doesn’t automatically make a team’s decisions magically transform into data-informed ones; using AI and ML tools does not mean your organization is optimizing for the future.
My advice continues to be: understand the root problem(s) first; then pick the tools. Here’s a piece from my company about our own internal process to select a project management tool.
2. Holy Crap, GPT4 Does What?
A roundup of “welcome to our new overlords” pieces that blew my mind.
In the “think how it can be misused before you roll it out category,” AI fools voice recognition to verify identity by Australian government agencies.
3. SEO can be gamed by AI. For Now.
For whatever reason, the desktop version of Twitter won’t let me open this tweet and copy the images over, but you should go see how easy it is to get Bing to index some copy. It’s a nice arbitrage while it lasts…
4. Here Come The Agencies.
#ShockerNotShocker, agencies the world over are figuring out how to be better advisors charge clients for more while paying less by using AI (Paywall or email required, sorry.) As a consultant myself, I can’t be too hard on them. This is a tough business and I believe that knowledge and expertise should be rewarded. I just want everyone to be careful about how fast they get on this bandwagon and who they pay for that info because it’s a) not rocket science and b) they aren’t that far ahead of you, in most cases- we’re talking six months or less in this case. Obviously, this does not apply to AI/ML-first shops.
If you’re on the brand side, you need to read that piece and understand the ways agencies can and cannot help you in the AI/ML age. It’s got some excellent examples.
TL; DR? Before engaging agencies or consultants right now, think about what you need. Offerings right now are falling into three basic categories; a)saving money/speed time to market on creative and/or data analytics; b) thinking holistically about AI/ML integration on a strategic/corporate level; and/or c) developing entirely new campaigns using new tools and ideas. The dangers of outsourcing too much to AI are real, and this separate article does a good job of outlining them. Bonus points for Robocop references.
“…as sophisticated as the largest platforms may be, there are components of the digital advertising supply chain that no advertiser should be willing to entrust with them.”
-Eric Benjamin Seufert
5. Yes, Emily Bender Is My New Professional Crush
In the face of Sam Altman refusing to share the source for GPT-4, and the world figuring out they tested the model on their own training data, Here’s Venture Beat getting very excited about Emily Bender and Stochastic Parrots. (You did read my last newsletter so you got that reference right away, I know it). I won’t stop banging this drum. To that end, it’s worth revisiting this 2020 piece on the messy and not-as-altruistic-as-they-started origins of OpenAI, which is a name that will probably get changed soon.
Breaking news! Here’s Emily with a blistering takedown of the weird request to slow AI down. There’s a reason she’s literally the only person I have Twitter Notifications turned on for.
6. How Bing Works, By Bing
When should you use queries in classical search mode, and when should you use conversational or chat Mode? Bing does a great job of explaining and offering options. TL; DR: the less you know what you want, the more likely it is that chat will serve you better.
New Section! This Week’s Bookmarks
Here’s where I’ll drop links to things I’m bookmarking for future use.
“An epic list of crazy text-to-image ideas” from Creativndie.com. Clearly, a way to create fantastic content for your customers is to help them create prompts on AI to get stuff done. Great opportunity if your market warrants it.
SEO Pro Jessica R. screenshots her ChatGPT process to create landing page briefs. I LOVE this kind of real-time information.
This Week In Privacy And Ethics
I’ve been working on privacy and ethics issues in marketing data since 2015. I also chair the committee on Privacy and Ethics for the Digital Analytics Association, a committee I created in 2020. My opinions are just that and do not reflect those of the DAA, or my colleagues on the committee.
Here’s a series of perfect examples of how the model used to train generative AI is steeped in bias, and how that bias affects the outcome. In this case, it’s a lesson in the “smile-based” culture of the west. This week, I learned that Russians don’t smile as much, and consider it untrustworthy. Bonus: NPR report that generative AI images that smile are more likely to be believed, so don’t trust that smiling image of a stranger online.
You’d think people would learn from Google’s “Don’t Be Evil” slogan it had to retire. But nope, OpenAI did not. I’m VERY wary of any capitalist promising to help “all of humanity” with their product, and you should be too. Let’s keep betting on those folks who want to augment humans, not replace them.
Whoops:
Playing With Midjourney Prompt Hero
Midjourney was having bot issues today, even after I subscribed, so instead here’s an image I found on a site everyone can use called Prompthero.com. Sadly, those “hallucinations”/bullshit things show up, but it’s good to keep a note of them. The actual prompt is in the caption. Ironically, it might be the most appropriate prompt for “stochastic parrots,” ever.
New Section! Celebrating Humanity.
All this talk about the future of AI has me thinking a lot about what it means to be human, and how special we are. In each issue, I will share things that I think are uniquely special about humanity- joyful, compassionate, creative, and/or weird.
This week’s entry is a piece posted on Facebook by the International Association of Dance Teachers. I can’t imagine any AI being able to envision this, let alone execute it. It’s joyful, energizing, and celebratory.
What’s Next
I’ve ordered this book on Marketing AI and will write a report on it after I read it.
I’m about to record my first ever solocast as part of my “Stayin’ Alive in Tech” Podcast, and it’s all about AI. I’ve had so many questions from you all that I want to summarize some thoughts quickly.
I will be taking some time now to update my website to reflect the current state of my consulting business, including offering consulting packages.
If you work for a company that’s integrating AI into its marketing technology, I’d like to talk with you, on or off the record. Please send me an email or ping me on LinkedIn.
{blush}
Thanks!!