Podcast: Play in new window | Download
Did you know that even some of the most well-known national newspapers are using AI to write some of their articles? It’s true. And you probably didn’t even notice. So how can Marketers leverage AI as part of their Marketing strategies?
We’ve all undoubtedly encountered a chatbot a few times in the past few years, and that is certainly using AI technology. But RJ Taylor, AI platform founder & CEO at Pattern89, reminds us that the key is to make sure you’re not losing that humanness in the process. Machines are great with data, but it still takes a human to differentiate the nuances of those gray areas that test ethical or inappropriate boundaries. After all, an eggplant emoji is just an eggplant to a machine, but humans know better.
Rich: My guest today is CEO and founder of Pattern89, an AI platform for digital ads that analyzes ads across 2,900 different dimensions and compares billions of advertiser data points to predict which ads perform the best on social media. A 15 year tech and startup veteran, he led teams at ExactTarget and Salesforce prior to launching Pattern89. His leadership was instrumental when Salesforce acquired ExactTarget for $2.5 billion in 2013, where he became the Vice President of Mobile Products.
Recognized as one of Indianapolis Business Journal’s Forty Under 40, he’s a father of four, a national champion swimmer, and an amateur birdwatcher. But today we’re going to ask him to dust off his crystal ball and tell us what creative will perform best in 2021. So let’s dive into it with RJ Talyor. RJ, welcome to the program.
RJ: Thanks for having me. Thank you. Thank you. Appreciate it.
Rich: All right. So before we dive into your AI predictions, what’s the rarest bird you’ve ever spotted?
RJ: Oh, you know what I’ve been living in Indianapolis, so there’s not a whole lot of rare birds here. It’s more like backyard birds. But we do have some Bald Eagle sightings which are kind of cool because they’re coming back into the habitat here after being endangered. So I’d say the Bald Eagle. But I don’t know, real bird watchers would kind of laugh at that mention. So that’s why I’m sticking with the amateur title.
Rich: Well, it is a pretty awesome bird. A couple summers we’ve rented lake houses here in Maine. And one year that we did, one of the houses had a little tiny Island out in front of it and there was a Bald Eagle nest. Those birds are pretty amazing.
RJ: They are. Yep. Yep.
Rich: All right. Okay. Enough of that though. All right. So let’s talk about, this is a crazy idea. So just starting off, brand managers and creatives often miss the mark on guessing what consumers will want next. So why should we possibly believe that computers or AI can tell us what humans will like?
RJ: Yeah, it’s a good question. The fact of the matter is both are actually really important and humans know how to react to a situation that we’re in right now. Like 2020 and 2021, these years have been really strange, and humans know how to manage it. Whereas a machine might just, you know, non-compute type thing. But machines can find patterns and outliers that humans can’t even spot. Even the most sophisticated statisticians can’t identify some of the patterns that are emerging from artificial intelligence, but they don’t know how to apply those in an empathetic or ethical way, or even one that doesn’t sound tone deaf. It seems like every day something new happens that we’ve got to react to as humans and as marketers. So it’s really a combination of the two.
And humans are really good at creating things and machines are really good at finding patterns. And so how do you blend those two together and make sure that you’re not making a misstep ethically, or just knowing the current times or those types of things. So I wouldn’t say it’s one of the other, it’s both.
Rich: Okay. All right. So that’s definitely helpful. And before, I guess I just want to define what ‘creative’ means, because that means different things to different people. So when we’re talking ‘creative’ today, what are we talking about? Are we talking about colors, copy, imagery, anything else?
RJ: Yeah. Um, yeah, definitely. In my definition creative encompasses the image, the video, the gif, the color composition, balance, all of the components of an actual piece of creative. And then also text, so copywriting, headline, body copy, those types of things. And in combination, the computers can look at the entire piece as a piece of creative, and then also individual pieces as well. So copy, the image, but creative can be used to describe all of those.
Rich: So I’m still struggling with this one a little bit, because do you have a track record here? Like whenever they do those year-end predictions of what’s coming next, they never look back and find out if any of these people, these gurus, these experts were right, even 50% of the time, the year before. So is there a track record for AI being able to find some emerging trends when it comes to creative work?
RJ: Yeah. Yeah, absolutely. We started in social because the social trends happen really like in a day or two. They emerge and then they kind of flame out, And on average, marketers are spending about 10 and a half days with their creative. So predicting what’s going to happen in the next 10 and a half days, or the next 30 days, is actually really easy to do when you have a data set as large as we have access to, which is over 200 billion impressions where the data representing 2,000 brands and we’ve tested it over and over again to figure out that we’re actually over 95% accurate at predicting what’s going to be trending.
And you know, we can’t predict COVID, for example. You can’t predict some of these kinds of wilder events, but you can predict based on seasonality, based on what we’re seeing emerging from colors and images, what are people more likely to click on, or purchase, or engage with in some way. And because of that rapid life cycle of an ad or lifespan of an ad, along with a massive dataset, we’re able to be over 95% statistically certain that something’s going to happen. So, yeah.
Rich: Can you give us some real world examples? Like, can you give us some examples for 2020 where it’s like, this is what the AI said was going to happen and here it is?
RJ: Yeah. Well, I mean, the first thing is that it predicted that creative life cycles will actually shorten. And we’ve moved from around 30 days on average life cycle to around 10 and a half days. And that really emerged because marketers are trying to take advantage of what’s going on in the world, and everything that’s going on in the world is changing so rapidly that the lifespan of these creatives algorithmically has just a shorter life span. So that’s one way, it’s just predicting the overall lifespan.
Rich: But it’s not like you can say, “In January, blue is going to be huge”, or this specific.
RJ: Oh!
Rich: Oh you can?
RJ: Yeah. Yeah. So right now we’ve just published our 2021 forecast to actually forecast out what colors are going to drive best performance in early 2021. And it’s looking like things like green and kind of a pinky color. I can give you the hex codes if you’re interested.
Rich: My creative director would certainly be interested, I can promise you that.
RJ: Yeah, well, it’s interesting because you can look back. We look back 11 years and we say what has trended month over month, year over year. And we can look back and say, since 2017 the click-through rate on ads that have this color of this peachy color have continued to increase in performance, and identify those things so that as you’re planning your color palette for next year you can include those or choose to go against the grain and exclude them.
I’ve had some brands that we talked to say things like, “Well, if that’s the case, should we change our brand colors?” And I think, no, you should not change your brand colors. That’s not the reason to do it. But you might want to understand how you can source photos or use text font, text colors, that emphasize these things while maintaining your own brand story.
And that is actually a good example of how humans and machines can come together. A machine might say, “Yes, change your brand color.” But as a human we know, no, we want to establish a brand identity and we want to establish that because it stands for these things. And we want to keep that consistent, no matter what trend hits us. But how can we use these other dimensions, like colors or image dimensions or video dimensions, to still tell our story in a way that performs well.
Rich: Right. And like you said, sometimes you might want to go against the grain, like you know that peach is really popping and you’re like, but if everybody’s going to be doing peach, then I want to do something else because I want my ad to stand out.
RJ: Another good example of that a lot of creatives are looking right now at the Pantone of the year for 2021, which was announced I think last week, it’s that yellow and gray. And so are marketers just going to embrace the Pantone of the year because that’s what the Pantone people said, or are they going to try to use other colors because they know that that color palette is going to be saturated? It’s a question that a human needs to answer, but we can make reasons for and against it.
To note, those pantones are not predicted to perform from our data, but it’ll be interesting to see how that announcement influences color performance.
Rich: You also talked a little bit about computers or AI not being great about ethics or being tone deaf. So for somebody who might not grasp that, how might you explain that? What, what mistake might AI make that a human would be like, “Oh no, no, we’re not going to touch it”?
RJ: Right. Well, I mean a potentially inappropriate one is by looking at emojis. And a machine might say something like the eggplant emoji or the peach emoji are going to drive your performance up. But if you are a brand that is not into provocative or sexual innuendo, you might not want to use those emojis. But the machine will say, “Hey, these are performing really well”, because they tend to perform very well. But they have a subtext that humans understand and that has emerged over time that wasn’t always the thing.
Rich: It used to be an eggplant was just an eggplant.
RJ: And if you’re my parents, an eggplant is still just an eggplant. But you know, there are things like that. A celebrity can say something one day that means something, and the next day it means something different because it’s used in that way. I remember some world events that have happened where the tiki torch suddenly got associated with white supremacists. And it’s like, wait, no, we’re just talking about a tiki torch, a backyard barbecue. But suddenly it became representative of something. And a machine might not know that or the subtext. It’s also cultural based on region or based on part of the world that some things, you know, in Japan might be a lot different than I might have some context that we don’t understand here. So there’s those types of things that humans really need to be aware of.
And it’s one of the reasons why everybody in social always says don’t preschedule your tweets or your posts. You might unknowingly get yourself into trouble because you have a prescheduled post that doesn’t react to the world as it is happening.
Rich: So my big takeaway from all of this is that, and you said this basically at the beginning, but AI is really good for predicting trends. It takes a lot of data, it helps us make sense, and it says these colors or maybe these words are going to drive a lot of engagement in the next week, next month, next year. But it takes a human being at the control panel to say that may be true, but ethically it’s not okay to say that, or this has meanings that you don’t understand, or this won’t play well here or it doesn’t go along with my brand. So there’s still going to be a human component here, we’re just leveraging data – and so much data that we can’t process it – so we’re using the computer to do that for us.
RJ: Yeah. Yeah. You know, the thing I would add to that is that for so long marketers have been told to act like machines. And creatives have actually been discounted and marketers have been told to act like a machine. Everything has to have data underneath it. You need to learn how to do pivot tables and the lookups and advanced queries to prove your ideas. But it turns out humans are never going to be as good as a machine to compute those things, so why not really emphasize what we are good at, which is human and creativity.
And, you know, the other piece to think about from a machine perspective is that a machine might tell you that creatives are predicted to perform, that people of a certain complexion or with certain ethnic markers or whatever the case is. And AI is not good at identifying race, body type, even gender. And as humans we try to figure out how do brands represent themselves in a way that speaks to our time, speaks to the message we want to tell. A machine doesn’t know how to do that because they can’t differentiate between a light-skinned Hispanic person and a light-skinned African-American person. Or between genders, or a full-figured person and a skinny person. The machines aren’t smart enough there. So it’s really like we get into racism, we get into some of the underlying problems with AI that humans know how to get around. They know what they want the brand represented as, and a machine may or may not be able to help them in solving that specific problem.
Rich: Some brand managers may know how to avoid those situations. I’m also thinking, if we program into our computers that we want to show diversity in this ad, and suddenly it comes up with a Mac and a PC and a Chromebook, because that’s what diversity means to it.
Before I get into what I’m really most excited about, which is what’s coming our way in 2021, I am kind of curious. So you’ve shared some information, you’re about to share here on the show with what the trends are, but obviously your company works with individual companies, you’re a B2B. So how do you work with individual companies that’s more in depth or more specific perhaps than the kind of information you’re going to be sharing with us today?
RJ: Yeah, well, we publish our monthly creative forecasts on a monthly basis to give the world some ideas of the things that we can do as well as to help guide marketers and creatives on the overarching trends. But how we work with individual companies is to look at their customer or customer’s data, benchmarked or against those trends, to understand where are they in line. Where’s the opportunity where the patterns and outliers from a creative standpoint that they could take advantage of? And we typically benchmark and look at creative analytics, look at competitive creative benchmarks, and then identify what insights. And then we can glean from that data and then ultimately forecast or predict what creative trends they can take advantage of so that as they go to put a new creative brief together. Or I’m ready to launch a campaign, and they’re trying to decide between maybe, 10 or 20 creatives, we can actually predict which of those creatives are going to work. So it’s on an individualized basis.
Rich: And I guess that makes sense because we’re a digital agency. We publish information about paid search and SEO and email marketing all the time, and it’s out there for free. But then if somebody comes to work with us, it’s much more tailored and really dialed down into what they need to do specifically. So that makes a lot of sense.
Cool. All right. So let’s hear it. What does the future predict? What are we going to see in 2021 as far as creative goes?
RJ: Well, there’s all sorts of interesting trends and countertrends that have emerged. I already mentioned the shortening of the creative life cycle and the lifespan really dropping down. As we look to the types of imagery and creative that’ll drive the best performance, we’re going to be still seeing a lot of impact from COVID and sort of the isolation associated with that. So, electronics, TV’s, phones, now represent around 30% of all creatives have a phone, tablet, or TV in them or computer. That has spiked greatly in this year because people are alone and they’re literally just spending time on their phone. So it’s a little bit sad of a prediction, but it’s also real.
Selfies will continue to increase in their popularity, as well as we’re going to start to see things like dreaming of a future that’s better. Spas, outdoors, playgrounds, fishing, parks, oceans, those types of things, really continue. Sad for the cat lovers out there, the internet has seen enough of them, especially as it relates to cost per click. So we’re going to see cost per clicks continue to rise for images or videos that include cats. That doesn’t mean don’t use them, but just know that they might draw your prices up. As well as gyms, because as people were sensitive to that type of imagery right now, so the cost per click is going to continue to increase there.
Rich: I’ll believe that cat thing when it happens. And maybe we’ll get back together next year and I’m going to bring up the cat thing. It’s the very first thing. And you will be able to figure it out right then, because I don’t think the Internet’s had enough cat. So I’m just going to put it out there right now.
And are there any trending animals because although right now we’re recording this,
I brought this into a quiet room in my house so we could have a quiet conversation. I think when we did our pre-interview you saw that I had the cage there, you may have asked me about it. I have a hedgehog. So is our hedgehog is going to be trending in 2021? The world wants to know.
RJ: You know what, I don’t have any predictions on hedgehogs. But we can make predictions on dogs and cats and even some birds and obscure animals, like llamas, those types of things. But no, I don’t have anything. I’ll dig a little bit deeper.
Rich: All right. Let me know if he finds something. Yeah. Yeah. That’ll be interesting. All right. And we talked about colors, we talked about images. I’m kind of curious, hashtags obviously play such an important role on so many different platforms. Do you guys ever kind of go down that rabbit hole or is that something that is just so of the moment it’s hard to protect?
RJ: No, definitely. We are very interested in copy recommendations. And then there’s kind of two pieces with hashtags. One is in the copy. And then with some of the recent changes at Facebook, we’re seeing more and more texts overlaid on top of the creative itself. So you can imagine an image or a video with a hashtag on top of it. So we’re seeing that body copy typically between the 40 to 60 character level drives the best performance. And also when it includes a hashtag. Now, not predicting on the specific hashtag that you’re using, but the fact that it does include a hashtag. However, when you have a hashtag text overlay, that typically increases the cost of your ads and decreases the performance.
Rich: Really? That’s interesting.
RJ: So hashtag in the body copy, go for it. Hashtag overlays, I would recommend avoiding that right now. If that’s in aggregate, it might specifically work for your brand or for your objective. But in aggregate, we’re actually seeing that cost more money, which is, I think kind of interesting. Like, as you think about branded hashtags or trying to put hashtags over content versus just the body copy.
Rich: Well, just as a small business owner, hashtags used to be such a big part of my life when Twitter was a tool for small businesses. I don’t really feel that Twitter is a tool for small businesses as much anymore. I think it’s more of a tool for celebrities, Kardashians and the POTUS personally, but that’s my own issues.
Let’s talk about copy just for a moment here. Because I want to talk and again get your thoughts on GPT3 – which stands for “Generative Pre-rained Transformer 3”. And according to Wikipedia, and I had to look this up, it’s an auto regressive, which I also had to look up, language model that uses deep learning to produce human-like text. So in other words, computer’s writing text or copy. What role do you feel that GPT3 will have in creative work moving forward?
RJ: Yeah, well, GPT3 and other natural language generation solutions – and LG is the term that a lot of folks in AI use – are on the rise. And we saw some pretty stark advancements in GPT3 and in LG this year. So I think that marketers definitely need to be aware of them and understand how they impact or how they can be taken advantage of. GPT3 was heralded as kind of the end of copywriting by humans.
So I actually have a Master’s in creative writing and I have an English undergrad as well. So people in that circle kind of all freaked out for a second, because they’re like, wait, am I not valuable anymore? We’re seeing some newspapers say that they’re generating over 20% of their content from natural language generation or GPT3 solutions. However we have the same creative challenge that I’ve described before, which is a machine can write a pretty awesome press release or even a sports summary, it’s really good at that. And actually the advancements that we’ve seen it makes it sound more human versus machine generated, which I think is paying attention to things like tone and voice.
However, newness comes from humans. Newness like the new voice, the way of speaking, comes from us. And that’s both on text as well as in imagery or video. We can look to these other solutions and say, Oh, these things are going to take over, but machines optimize for the top performing. Therefore everything’s going to end up sounding the same if everybody uses GPT3 or a natural language generation solution.
And the same thing algorithmically on the creative side with images or videos, every video or image is going to look the same because they’re optimizing to the same norm human. And ‘human in the loop’ is the term that student used in AI. Human in the loop interrupts that cycle where we come up with new ideas and we introduced newness in the same way that natural language generation solutions do. And I’ll tell you, I’m more worried about an LG or GPT3 having a misstep because you got to have a human read every word that machine writes. And there’s some famous examples of how machines have been trained and then they automatically start spouting out racist or otherwise inappropriate language pretty quickly. And you unleashed one of those things on your website for SEO or for some other purpose. And you’ve got a brand disaster on your hands.
Rich: Nobody wants AI generated Tourette syndrome on their website, for sure.
RJ: Yeah. Yeah. It’s risky right now. And then you have to make sure you have a good Q&A process. And then another secondary one so that you don’t get into big, big problems. So something to watch, because I think it can help us with rogue things like, no pun intended, but things that we do over and over and over again. But for newness or what we want to have a new clear brand voice, that’s still human.
Rich: Yeah. And actually I think I told you a couple of years ago during my presentation at the Agents of Change Conference, I talked about disruptors and one of them was AI and the ability for AI to write. And at that point it was brand new, but it turned out that 800 articles the year before had been written, it wasn’t the Washington Post, but it was something like that. But then when you dug a little deeper, most of them were sports scores and reports on stocks and they were well-written. Better writing than some of the people I know. But the bottom line is they weren’t really filled with life. And there was an example with a human writing, the Denny’s Quarterly report, and talking about it hitting a grand slam. Which of course, anybody in the U.S. who’s ever been to a Denny’s know ‘grand slam’ is the name of their breakfast. So at this point, AI is not going to come up with that joke, right? It’s like, that’s going to be one of the things, at least for now, jobs as copywriters are safe.
So anyway, this has been fantastic. I really enjoyed this. I hope people got a lot out of it. If they want to learn more about you or about your company, where can we send them online?
RJ: Sure, sure. So pattern89.com is our site, and those creative forecasts are free and available in our learning center. So come on and see us at pattern89.com. I’m @RJTalyor on Twitter, so you can find me there and tweeting about birds and kid stuff and digital marketing and AI.
Rich: So it’s you and not a bot, is what you’re saying.
RJ: It’s true. Yes, it is me.
Rich: Awesome. RJ, absolute pleasure. Thanks so much for stopping by today.
RJ: Thanks for having me.
Show Notes:
RJ Talyor lives by the mantra that the future of marketing will be powered by AI. His company, Pattern 89, is doing their part by using AI to predict future marketing trends. Follow him on Twitter for his latest tips and insights on all things AI in the Marketing world.
Rich Brooks is the President of flyte new media, a web design & digital marketing agency in Portland, Maine, and founder of the Agents of Change. He’s passionate about helping small businesses grow online and has put his 20+ years of experience into the book, The Lead Machine: The Small Business Guide to Digital Marketing.