February 27, 2024

Emotions, motivators and drivers: what matters most in marketing

Today's show is a little different. It's not your standard interview style discussion, because today's guest is not our standard guest. He is an Emmy Award winner, an entrepreneur, inventor, and researcher. He's a master of a...

Today's show is a little different. It's not your standard interview style discussion, because today's guest is not our standard guest. He is an Emmy Award winner, an entrepreneur, inventor, and researcher. He's a master of advertising and marketing effectiveness. No-one has worked harder or been more effective at understanding the relationship between marketing effectiveness, emotions, motivations and drivers. And... he built it into a machine learning model.

So, what we're bringing you today is a first hand account from a pioneer in machine learning of how he built his AI, Bill Harvey. What do I mean by pioneer?

Well, Amazon got there in 1998 or 99, but today's guest built a machine learning recommendation engine that was embedded in cable tv boxes in 1997. Not too shabby, eh?

This year, McKinsey tells us that AI is at the top of CEO's agendas. As a CMO, you are probably all over gen AI and tools driven by machine learning too. That's why we thought you might want to hear this.

Hear Bill's full story here:

About Bill Harvey 

Innovator & inventor with 35+ years as thought leader in media research & key solution-provider to Fortune 500 brands. 

Breakthrough inventions include: 

  • The ADI/DMA 
  • Addressable commercials  
  • Passive Peoplemeters 
  • Planning/Purchasing media based on single-source Big Data matching (4 U.S. patents) 

Bill is the original discoverer of DriverTags™ the first proven scientific development of psychological attribute metatags for content (programs, movies, ads, etc.). RMT’s DriverTags™ have the highest predictivity of Nielsen ratings, series cancellations and renewals, and individual viewer likelihood of conversion to specific programs. Most recent work being presented at ARF shows an ability to lift ROI for brands by over +35%. 

Pioneer in New Media, set-top box data, advertising ROI optimization, measurement standards, privacy standards, & ARF Model for Evaluating Media. 

Key consultant in development of Personal People Meter (PPM) & ScanAmerica at Arbitron. Developed first automated Marketing Mix Modeling system for General Foods. 

Co-founder TiVo Research & Analytics (TRA), the first company to merge single-source & Big Data. First-named inventor on TRA’s three U.S. patents. 

Co-founder of Next Century Media & New Electronic Media Science, third-party research companies that serve more than 70 of top 100 advertisers, most major cable & satellite operators, networks, advertising agencies, & other market & media research companies. 

First to turn set-top box data into TV audience data that met media research standards, spearheaded writing of industry privacy principles for ANA, AAAA, & ARF’s CASIE joint task force. 

Held various executive positions at Arbitron (now Nielsen Audio), Interpublic, Grey Advertising, & OpenTV. 

Honored in 2014 by ARF as first recipient of Erwin Ephron Demystification Award, presented to leader with the “unique ability to communicate complex insights, lead organizations through challenging scenarios, & translate ideas into action.” 

Links  

Full show notes: Unicorny.co.uk  

LinkedIn: Bill Harvey | Dom Hawes  

Website: Bill Harvey Consulting 

Sponsor: Selbey Anderson  

Related Unicorny episodes

"It’s a language model, stupid". How marketing should and shouldn’t use AI with Steven Millman 

Selbey Labs

Contact: Gerry Hopkinson CEO 

Website: Selbey Labs 

Other items referenced in this episode

DriverTagsTM  

Dr. Richard J. Davidson 

Weaponizing The Wanamaker Paradox by Paul Worthington 

Next Century Media 

Semasio 

Attention in AI 

Addressable TV 

RMT (Reseach measurement technologies) 

 

Episode outline

Introduction to Bill Harvey  

The host, Dom Hawes, introduces the podcast as a discussion on the business of marketing and introduces the guest, Bill Harvey, as an Emmy Award winner, entrepreneur, inventor, and researcher.  

Importance of AI in Marketing   

The discussion emphasizes the growing importance of AI in marketing and how it is at the top of CEO's agendas, highlighting the relevance of machine learning and gen AI in today's business landscape.  

The Role of Emotion and Context in Advertising   

Bill Harvey explains the relationship between emotion, motivation, and advertising effectiveness, emphasizing the importance of positive emotions and brand attraction in driving behaviour and sales.  

The Start of Bill Harvey's Career   

Bill Harvey shares his journey, starting at Gray Advertising, and his fascination with analysing the effectiveness of ads in different programs, leading to his research on context effects in advertising.  

The Development of DriverTags™   

Bill Harvey discusses the extensive process of narrowing down a list of over a million words to 1562 psychological words, which eventually led to the development of DriverTags™ and their application in program recommendations.  

AI-powered show recommendations   

Bill discusses the introduction of AI-powered show recommendations for subscribers, highlighting its ability to analyse viewer habits and optimize show selections.   

Effectiveness of DriverTags™   

Bill explains the effectiveness of DriverTags™ in matching ads with program context, leading to a 36% increase in sales and reduced marketing costs for brands, surpassing traditional ad recall studies.  

Digital Advertising and Somacio   

Bill introduces the integration of DriverTags™ with digital advertising through Somacio, resulting in successful resonance between ads and consumer motivations, leading to a 95% increase in overall sales ROI.   

Validation of Motivational Types   

Bill highlights the impact of altruistic motivations on modern consumers and how it plays a significant role in advertising effectiveness, as validated by the ARF Cognition Council.   

Introduction to Canadian Market and Vividata Partnership   

Bill discusses their work with Vividata in Canada, where they are using data to analyse consumer motivations and how it is impacting advertising strategies and creative briefs.  

Collaboration with AI Companies   

Bill talks about their collaboration with AI companies to add greater sensitivity to human feelings to their models and how it's a new area of focus for them.  

Addressable TV and Programmatic Advertising   

Bill delves into the concept of addressable TV and programmatic advertising, explaining its efficiency and impact on marketing strategies, and how they are positioning themselves in this space.  

Application in B2B Marketing   

Bill discusses the application of DriverTags™, need states, and motivational types in the B2B marketing space, emphasizing the importance of emotional appeals and how they are being used to address the business-to-business market.  

Impact of DriverTags™ in Advertising   

Bill talks about the history of his life's work, the 265 DriverTags™, and how aligning them with ads and audience context can drive deep connections with brands and produce remarkable effectiveness results. 

This podcast uses the following third-party services for analysis:

Podder - https://www.podderapp.com/privacy-policy
Chartable - https://chartable.com/privacy

Chapters

00:00 - None

00:28 - Dom's Intro

03:43 - Emotions in marketing

09:04 - Bill's story: his early days

12:44 - Bill's story: the method

18:33 - Bill's story: early applications

21:33 - Bill's story: marketing efficiency and effectiveness

30:09 - Bill's story: we ain't done yet

37:55 - Dom's Summary

Transcript

PLEASE NOTE: This transcript has been created using fireflies.ai – a transcription service. It has not been edited by a human and therefore may contain mistakes. 

00:03 
Dom Hawes 
Welcome to Unicorny, the antidote to post rationalized business books. I'm your host, Dom Hawes. This is a podcast about the business of marketing, how to create value, who's doing it well, and how you can help your business win the future. Today's show is a little different. It's not your standard interview style discussion, because today's guest is not our standard type of guest. He is an Emmy Award winner, an entrepreneur, inventor, and researcher. So what we're bringing you today is a first hand account from a pioneer in machine learning of how he built his AI. What do I mean by pioneer? Well, Amazon got there in 1998 or 99, but today's guest built a machine learning recommendation engine that was embedded in cable tv boxes in 1997. Not too shabby, eh? Now, this year, McKinsey tells us that AI is at the top of CEO's agendas.  

 
01:00 
Dom Hawes 
And as a CMO, you too are probably all over gen AI and tools driven by machine learning. And that's why we thought you might want to hear this. Today's guest is Bill Harvey. He is a renowned figure in the world of advertising and marketing with a career spanning over 35 years. He's best known for his ground breaking approach to data and analytics, particularly in the context of media and advertising. He is a big fish. A very big fish. And as I mentioned, he won an Emmy award for his invention of set top box data. Now, I first came across Bill when Dynata Steven Millman referenced driver tags in the episode called it's a language model. Stupid. In the simplest explanation, driver tags help brands increase advertising and communication effectiveness.  

 
01:48 
Dom Hawes 
They are a set of scientifically proven behavior driving motivators that we can bake into an ad to help connect a brand to a person's most profound personal motivations. Here I'm talking about things like emotional benefits or core values. Maybe mindsets need states, character and personality types, and Bill identified them and honed them into an AI model. He has proved that the more the driver tags in an ad align with the audience's interests and needs, the more effective the ad is likely to be. And he's shown that the same is true when you align the driver tags in the ad with the media context, so they help drive actual sales behaviour in consumers. Now, if that is not the holy grail of what we're all trying to do, well, I don't know what is.  

 
02:37 
Dom Hawes 
He starts his story today by talking about emotion, motivations, and the other factors that influence marketing effectiveness. Now, I'm interested in this word, emotion, outsiders and those new to b two b assume that those of us who are already in B2B think buyers are purely rational. Surely, they say it's our ignorance that makes historical b, two b marketing. So feature led. And how many podcasts have you listened to? And how many talks have you attended where the amazing revelation is made that B2B buyers are human too, so they must have emotion. So we should seek to invoke that emotion in our marketing to make it more effective. All I can say to that is, no shit, Sherlock. I'm not sure that's insight. More like thin sight to me, a revelatory gem of information that all the rest of us already know.  

 
03:29 
Dom Hawes 
So now it's time to introduce you to Bill Harvey. Bill, you've heard my setup. What's your take?  

 
03:35 
Bill Harvey 
I do think it's oversimplified, but I don't disagree with it. It is one of the necessary components of advertising effectiveness, but it isn't the only one. Emotions are related to motivations. Negative emotions arise when our motivations are thwarted, and positive emotions arise when our intentions, our motivations are succeeding or being encouraged by events around us. So there is a relationship between emotion and motivation. They're both important. Positive emotion tends to be associated with effective advertising. That's true. Sometimes, however, it's very subtle emotion. But the internal feelings are of brand attraction. And brand attraction can be measured in the frontal lobes of the brain by the degree of asymmetry between the right and left lobes, as discovered by Richie Davidson a long time ago. Dr. Richard Davidson. So it's a complicated process. First you have to get people's attention.  

 
04:34 
Bill Harvey 
Then you've got to maintain that attention that is sometimes called interest. And then along the way, persuasion ought to occur. Now, persuasion will be accompanied by positive emotion, brand attraction, all of the biometric and neuroscience of those things, and long term memory encoding. So if you can get all of that to happen, advertising can change behavior, essentially changing the perception of a brand. But then that forms a predisposition, and later, when there's an opportunity at shopping and the brand is seen, those things all interact and behavior sales occur. So it's just a little bit more.  

 
05:11 
Dom Hawes 
Complicated coming on to that, though. When we got together to have our discovery call, I was very taken with, and we're going to come on to that when you tell your story. But the importance of the context that your communication is communicated in, and I think this is going to tease us towards maybe starting your story, but would you mind talking to me a little bit before we get into the meat of your story? About the importance of context.  

 
05:35 
Bill Harvey 
Context effects in persuasion, advertising and propaganda have been known about for decades. Maybe a century back in the old days, people would say, well, if you have a financial services ad, put it in Barron's magazine or other financially respectable publications, because there the context effect will be kind of an implied endorsement by an expert organization that published your ad. So during the 20th century, second half of the 20th century, people did dozens of studies, both in academia and in commerce, of the effects of, for example, a funny ad in a funny program versus a funny ad in a serious program. And it was found that ad recall consistently went up about 15% in most cases when a funny ad was placed in a funny program. And that was found by 70 different researcher studies.  

 
06:36 
Bill Harvey 
So context effect is known, and even today in the 21st century, I just consulted on a project from meta in which 15 different contexts were studied for five different ads and five different verticals. It was a 75 cell random control experiment. It showed, in fact, very powerful context effects, which tended to limit the possibility for getting more than one or 2 seconds of attention in a scrolling environment like Facebook or Instagram.  

 
07:07 
Dom Hawes 
Wow, one or 2 seconds, yeah.  

 
07:09 
Bill Harvey 
And so all those things I mentioned a few moments ago about you need to do all these things, you need to get positive emotion, which usually requires having a character that you kind of identify with on the screen and you tell a story about the character and it's going to take more than 2 seconds. And this partially explains why digital advertising is good at reminding current customers of a brand to buy the brand. It's good in reinforcing that. It's not as good as, for example, television or magazines in getting new customers to a brand. If brand growth is its target, television is going to do the job a lot better than digital is going to do it, although it's always good to use all media types.  

 
07:55 
Dom Hawes 
So folks, that is the amazing Bill Harvey. In a minute we're going to delve into his life story, how he identified driver tags and you will be in awe of what he has achieved. He can talk with such conviction about effectiveness, the subconscious motivations and invoking emotion because he has a lifetime of research, testing and creating in these areas and what he just said. Digital is good at reminding and reinforcing, but it's not good at getting new customers to a brand that chimes with a blog, which I loved, by the way, by Paul Worthington, about how an over focus on digital is stagnating growth. I'm going to link that on the show notes at Unicorny.co.uk Let's get back to our story. Bill, what you've achieved in your lifetime is incredible. Why don't you help us all out, please. Where did it start?  

 
08:48 
Bill Harvey 
Markham, 21 years old, just joined Gray Advertising, wanted to learn everything I possibly could as quickly as possible. And I became fascinated with the idea of ads having different effectiveness in different programs, specifically on television. And I was interested also in ads in magazines and other media that were important at the time, more important than they are today in some cases. And Gray Advertising entertained my desire to analyse masses of data. And what I was able to do, looking at multiple studies, was I realized that there was no agreement between these studies. A J. Walter Thompson study showed this, and a study done by the television Advertising Bureau showed that in some cases, a situation comedy appeared to be the best place to put an ad for food and beverage product. In other cases, it seemed like dramas were better.  

 
09:40 
Bill Harvey 
It depended who did the research and exactly the way they did the research. Then I went to another agency and I continued my journey and my studies of context effect. Here they had 1500 copy tests they had done and they had saved all of the results. So I was able to do a much bigger study. And what I saw was there were definitely context effects, but they were much more complicated because there was an interaction between the ad content and the context content. The ad itself has a certain tonality to its own psychological descriptors, funny versus not funny, heart warming versus not heart warming. Those are some simple ideas that people have attached to trying to understand how to code an ad or a context. But I knew intuitively that it was much more complicated than that.  

 
10:34 
Bill Harvey 
That, for example, television show context scripted shows, they have characters, and the characters have personalities, they have character types and character things like they have honor or they don't. But they also have personality characteristics, like they're chatty or they aren't, or they're conceited or they're not. And those things certainly have an effect on the psychology of the context. Some of that stuff in ads, you have people who are very confident in ads, people who lack confidence in ads. So I knew that was much more complicated than a couple of simple coding classifiers. What I didn't know was, how do I get that list of tonalities that I want to use to code ads and to code programs so I can see do they align? Because that was my hypothesis from the beginning.  

 
11:20 
Bill Harvey 
As soon as I saw that there was a covariance and interaction effect between the specific ad and the specific context, then I knew I had to approach this in a more scientific way, and it would probably cost me money. So some years went by before I got to the next phase of the.  

 
11:37 
Dom Hawes 
Game, when I'm asked what the most important characteristic is for a marketer or an entrepreneur, because, by the way, I think they share a lot of traits. The most important for me beyond doubt is curiosity. And coders and engineers often share this trait, too. You don't get to invent, create, or make something of value unless you ask why a lot, unless you're on a mission to understand how things really work, as opposed to just believing what you're told. Bill has hinted at this already. Intuition about the complexity and sheer size of the task ahead would have been enough to put most people off, but not Bill.  

 
12:13 
Dom Hawes 
Listen to how he relentlessly kept on asking until he got the answer, and be inspired to write down something you intuitively know you can improve, but seems daunting, and I think by the end of his tale, you will want to itch that scratch.  

 
12:28 
Bill Harvey 
What we did was we boiled the ocean. Essentially. We saw that there were over a million words in the english language, and we needed to pick from those million words which words were going to use to code ads and programs, and we needed a way of deciding how to pick those words. So the first thing we had to do was to decimate the list. It was much too big. Million words. We had to get it down to something manageable. So we figured a good way to do that would be we wrote up a one page of instructions saying, we're looking for psychological words. Our instruction sheet said psychological, includes human values like hope and love and education and family and so on. And it includes personality and character traits, as I mentioned before. And it also includes human situations like love triangle.  

 
13:14 
Bill Harvey 
It also includes content descriptors like fast moving or film noir and so on. So we gave them these examples to get their minds going, and we hired 22 coders because there are 22 volumes in the Oxford on a bridge dictionary. So we gave each coder 1000 plus page volume, and they started to fill a spreadsheet with those words.  

 
13:37 
Dom Hawes 
Bill goes into great detail here about how the sausage gets made. It is fascinating stuff, and for anyone who wants to go that deep, we will make the full version available in.  

 
13:47 
Bill Harvey 
The show notes word within the cluster.  

 
13:49 
Dom Hawes 
That has the highest, but for the length of our pod today. And to make sure we continue at pace, let's pick up the story again, where Bill has narrowed down the number of driver tags, but he's hit a problem.  

 
14:00 
Bill Harvey 
So that gave us a list of 1562 words. And then were somewhat bollocked after that. What do we do with the 1562 words? The answer was to come shortly thereafter when I had a company called next Century Media, which won an Emmy in the measuring the tv audience on Moss and also was able to deliver addressable commercials and program recommendations. And the program recommendations was the way we decided were going to further winnow down the list. So we took all the shows on television back at the time, it was only about 10,000 shows reported by Nielsen. So we coded those 10,000 shows again, human coders by the 1562 words.  

 
14:46 
Bill Harvey 
That took a long time, but when we got that done now we had hundreds of thousands of homes with millions of people in them that had our software, which had been downloaded by the world's largest cable operator. So we had all these homes that had got suddenly gotten 500 channels. This was the beginning of the digital setup box era. So they had the first analog, hybrid digital setup boxes, digital analog. And they were able to expand their channel choice from 70 to 500. So I said, these people are not going to know what to watch on television. There's too much choice. But we can help them. And all they have to do is press the a button on the remote, and within a second, the system will function as an artificial intelligence and give them a recommended show to watch next.  

 
15:28 
Bill Harvey 
So a letter went out to all the subscribers that had these new boxes. They were told, if you don't know what to watch, give the AI a chance. It will not recommend anything you've ever watched before because we're keeping track of what you watch. We already told you that, and you already didn't opt out of that, which we gave you the right to, but nobody really did. So in 1997, we did this test, and what we found was I had built this optimization engine, which today would be classified as machine learning, but I just called it an optimizer then. And what it did was, let's say, here's a household that just pressed the a button. They want to know what to watch next.  

 
16:04 
Bill Harvey 
The system, within microseconds, looks at all the shows that have ever been watched by that setup box and all of the codes from the 1562 codes that have been placed on that setup box to characterize the kinds of shows it leaned towards. The system inferred what the weights should because what it was trying to do was explain the conversions. Now, the conversions were the successful recommendations that resulted in people not only going to look at the show that was recommended, but actually watching three out of four of the next four episodes after a recommendation was made, which we called the conversion. So it looked at the conversions, and it looked at which of the 1562 words appeared to be correlated associated with the conversion events, and which of the 1562 words almost never showed up.  

 
16:55 
Bill Harvey 
In those cases that converted, it then started to vary the weights on the words. Words that showed up a lot during conversion events got higher and higher weights. Words that showed up very scarcely, if at all, during conversion events, were deweighted, eventually to zero. So, as that was going on, we noted that the conversion rate was going up. It started at 3%, and it went up and up, and eventually it hit 18%, six times higher than where it started. And I noted that at that point, it had taken offline all but 265 of the words.  

 
17:30 
Dom Hawes 
So Bill and his team landed on a canonical set of terms, the words that define all the effective emotions, need states, desires, and so forth, that we can bake into ads or align with their media context. Now, Bill actually whittled them down even further, but then he found he was beginning to reduce effectiveness, cutting into the muscle, as he put it. At that point, they knew 265 was the magic number. Next step, naturally, was to give these terms a name. I'm sure this aspect of the process is familiar to you. Does it matter what we call something, the programs, the tools, the approaches we create? Well, I tend to think it does, especially when you're trying to lend meaning to something that's pretty complex.  

 
18:17 
Bill Harvey 
We'll have to give them a name. And I said, how about behavior driving meta tags? And everyone said, that's terrible. How about driver tags? And I said, okay, driver tags. We'll call them driver tags. That's what they've been called ever since. Now, ever since we had clients very interested in this, particularly the studios and television networks. They wanted to see how they could make programs better. In addition, there were people in the advertising side of the business who thought we could use this to match ads. When we place ads in programs, we can do it not by blind, random assignment, but by maximizing the alignment between the ad and the context.  

 
18:57 
Bill Harvey 
I think that one of the things that makes it obvious to me that it should work is that why do people sometimes vote with their wallet to buy pay tv that doesn't have commercials at all? It's because the mood of the program gets broken. There's a cognitive dissonance when the mood gets broken, which means that if you set things up so the mood isn't broken, so that the program mood flows into the ads, then you should see a more positive advertising effect. So we did the first study of that, and that study showed that there was higher alignment between the 265 driver tags in the specific ad and in the program the ad was placed in, the increase in sales effect was 36% on average. Now, that's higher than what we're seeing now.  

 
19:46 
Bill Harvey 
A lot of at least 24 companies doing attention measurement now, and they tend to show that more attention leads to more advertising effectiveness. Measuring it sometimes based on did it cause a click through? Did it cause a visit to the website? Did it cause a search for the brand? Did it cause higher brand recall? Did it cause higher ability to guess who was the brand that was advertised in that ad? In some cases, sales effects are measured in these attention studies. It's kind of rare. Not many are actually measuring sales effects. But here we had Nelson Catalina measuring sales effects. And the sales effects of 36% on average is much higher than the increase in ad recall and stuff in most of the attention studies.  

 
20:30 
Dom Hawes 
Wow, that is pretty wild. You know, as marketers, we're always looking for ways to prove our effectiveness. But looking for sales effectiveness from ads is a very tough task, partly because many of us spend our time doing something else, like building brand awareness, and partly because it can be bloody tough to attribute and of course, partly because the results can be properly scary. That's what I like about this. And if you're interested in finding more about proving effectiveness, you need to speak to Jerry at Selby Labs because that's the kind of work that keeps him up at night. What surprised me more was that a brand's use of driver tags could be shown to help bring down marketing costs. I'm going to let Bill explain.  

 
21:17 
Bill Harvey 
Then there was a study that was done for one of the largest CPG brands in the world. They don't want their name used, but they allow us to use the results. And big CPG advertisers said, look, I don't want to do the same study that Nielsen Catalina did. It's going to come with the same Answer. I don't want that. Let's do something more interesting. Let's do something like a Millward brown type questionnaire where we measure all the funnel levels by a survey questionnaire. So they wound up doing, in addition to the set top box data, they did 23,000 completed interviews for a custom study. As you know, that's ginormous. So they did that study and what it showed, interestingly enough, for purchase intent, it almost got the same number as Nielsen Catalina.  

 
21:59 
Bill Harvey 
But 37% increase in purchase intent in the cases where the alignment was, we call it resonance, was above average compared to the 36% that Nielsen found for actual purchase. But they also measured a lot of other things, like first brand mention, also known as the saliency measure. In other words, I ask you, when you think of a bottled water, what brands come to mind? And if your brand, the client brand, is mentioned first, that counts as saliency. So that increased 62%. But the thing that thrilled the CMO of the CPG client the most, because it was such a big survey sample, you could break it out by frequency levels, and then in the low frequency tertile, the average frequency was seven. So they only got seven impressions. At least one of the seven, or whatever the number was of impressions, went over 30%. Resonance.  

 
22:57 
Bill Harvey 
And then we looked and we saw that almost none of those people had two exposures. Over 30%, over 90% of them. It was just one exposure.  

 
23:05 
Dom Hawes 
Just one. Okay, so by looking at the group that only got seven or fewer impressions, what you found was the ad still scored high in terms of its resonance, even when most of that group only got one poultry impression. Wow.  

 
23:21 
Bill Harvey 
So the CMO was flabbergasted. What it said to him was that if you put an ad in the ideal environment, you don't need as much frequency, which could save a fortune of money. So that was another finding.  

 
23:40 
Dom Hawes 
What you're hearing here is how an AI model gets made. We're all talking about AI, but how are those tools coded? What's actually going into them? Bill is giving us a ringside seat. I think one of the big misconceptions with AI is that because of a lack of transparency, we feel like it's just out there somewhere as a kind of omnipotent force, and we just plug into it like Wi Fi. So there's a danger that we might either take it for granted or overlook key aspects of its usefulness. But Bill is showing us there's a lot of old school elbow grease that goes into it and that the results, if the work is done right, can be spectacular.  

 
24:25 
Bill Harvey 
The studies didn't end there. Something happened called digital. And digital, particularly since around 2017, has been, as you know, racing ahead of television and all other media types. Other media types are somewhat shrinking or holding their own, in the case of television, but not growing. And all the growth ad spend has been going into digital. So we needed to have a horse in the race for digital. What are we going to do with driver tags for digital? Then we found a company called Semasio. Semasio, which is now operating in 26 countries, started in Europe. What it does is it takes a full text grab of all of the words on a page being visited by an id. So there's all kinds of ids out there. Semasio uses all of them so it can interoperate with trade desk, Google, Facebook, whoever it is.  

 
25:16 
Bill Harvey 
And so what they're doing is they're tracking these ids, and they're seeing what URLs they're landing on, and then they're taking a picture of those URLs, and then they're compounding that across all the URLs visited by the id. So if my id goes to some sites that have to do with human consciousness, some sites that have to do with politics, some sites that have to do with other forms of media research, some sites that are humorous, and so on and so forth. All of the tags that are already on, all of the words that are used in the text on those pages is being boiled down into, like, a tag cloud around my id by Semasio. And that's what they do for a living. We made a deal with them, and we trained their AI to reproduce a version of our driver tags.  

 
26:04 
Bill Harvey 
And when I say a version, we'd already, by that point in time, been asked by our clients, can't you condense the 265 driver tags for two reasons? One is it's breaking our brain, and the other reason is, you won't share those driver tags with us. But can't you condense them? Can you cluster them, in some words, into a small. So, we worked hard on that. We went back to the clients and said, okay, here are 86 need states. And they said, thank you, but it's still too big. Do it one more time. Bring it down by another order of magnitude. So we did that, and we brought it down to 15, what we call motivational types.  

 
26:41 
Bill Harvey 
So when we made our deal with Somasio, we thought, in order to make this a lighter lift to get started, let's just start with the 15 motivational types. Then a national retail chain found out about this and said, we'd like to use this. We'd like to test it out, see how well it works for us. But we want to have you tell us, what are the motivations in this ad that we're giving you, and then we want you to reach people who have those motivations according to the Somasio RMT method of looking at what content they consume. That's what we want. We want the resonance between the ad and the person, not between the ad and the context. And they said later, we'll do both.  

 
27:20 
Bill Harvey 
Obviously, if you put both together, it's going to be even more powerful, but we want to start just with the person. So the same creative was given to a very well known supplier of ids, of lookalikes. So the lookalikes that were being used by this retail chain and which were doing a decent job, that was the control group and were the test group. And Newstar compared us to this other supplier that was using lookalikes. And Newstar said the increase in overall sales ROI, offline and online combined of the driver tags over the affinity ids was 95%. You almost doubled this extremely popular supplier of lookalike models for new to brand. It was even 20% higher than that. So the most recent study worth mentioning was done by the Advertising Research Foundation's own Cognition council.  

 
28:16 
Bill Harvey 
And it was a study of sales IrI sales for 19 different brands over a six year period. What they had us do was to code the ads that were used over that six year period by each of those 19 ads, and by the months that those ads were running, according to Cantor. And then they, ARF Cognition Council did their own analysis of that. And what they found was that 48% of the sales results were explained by the motivations. And when you break down the results, to see which of the predictors, which of the motivational types had the highest prediction, it was wealth success. In other words, becoming wealthy as a motivation, status, prestige. And then altruism was actually the third highest predictor under those two more selfish predictors.  

 
29:08 
Bill Harvey 
So what it says is that in any modern age, with millennials and Gen Z and Gen Alpha, who tend to be more idealistic than the general population, but even millennials showed a greater degree of altruism than the baby boomers did, this altruism thing is not a flash in the pan. This is up there, very close to the impact of the selfish motivations of use my deodorant and you'll become rich and famous, which are the implications of most ads, that your life will better if you use my brand, and it'll better in all kinds of ways. So those are the validations.  

 
29:44 
Dom Hawes 
Bill, that's some story. And it hasn't yet finished. So the first piece you're going to take out and look at taking out into retail, there's three other pieces you're taking forward.  

 
29:53 
Bill Harvey 
Yeah. So that's one area. A second area that we're working in right now is specific to Canada. In Canada, we have a partner called Vividata, who corresponds to target group index in Europe and Australia and other countries, and to MRI and Simmons in the US. So in Canada, we have a number of the largest vivid data subscribers in Canada are practically everyone in the advertising business in Canada, buy side, sell side, industry associations, everybody and some of their largest agency and network clients are looking very closely at these data, which we're using the motivations initially in a self serve system where people can set up their own tabulation, how they want to look at it for any of the motivations. In Canada, they're not called motivations. They gave it the name drivers, vivid data drivers. So the drivers are the 15 motivations.  

 
30:54 
Bill Harvey 
And the way people in Canada, in confectionery and automotive and other categories are looking at using the data is to say, what are the motivations of the people who buy my brand? What are the motivations of the people who buy my competitor’s brand? And maybe I should start to use some of the motivations that are in my competitors brand if I'm trying to take away any business from them. So they're redoing the way they do their creative briefs to include the vivid data drivers as an input, a major input to the creative briefs. And that's just starting. That's a whole new market for us. And a third market. We've gotten a lot of many offers, including from IBM, Watson, and many other well known AI companies.  

 
31:42 
Bill Harvey 
Practically all of the big well known AI companies have wanted to talk to us about how can we add greater sensitivity to human feelings to their GPT four, or whatever it might be? Because right now there's no attempt to model human feelings. It's all about using language to essentially like auto fill. When you crowdsource all of the sentences in the Internet, or some subset of them, when a sentence starts out with these three words, what's the most likely fourth word, next best word, so that doesn't get into the domain of feelings at all, and you might accidentally use a word that might offend somebody. So that's their question. So were working with one AI company, not counting Somasio. We're working with one other, specifically AI company that's totally not doing market research at all.  

 
32:36 
Bill Harvey 
It's just an AI company, and we expect to be doing more work with AI companies in general. That's a whole third area that we're excited about. And then the fourth area that's presenting itself very recently. And it's as a result of this attention craze that has swept the marketing industry that after all these years, my hoping that we get to very sophisticated ways of coding contexts and coding ads so as to maximize advertising effectiveness. Instead of getting the whole sauce at once, it's coming at us just one step at a time. Attention. The very first step of the process is where we're focusing all of our attention on attention, not the fact that we also have to maintain that interest, get it to be positive. Emotion, brand attraction, long term memory, encoding. Ultimately, sales behaviors change.  

 
33:37 
Bill Harvey 
We're just focusing on the attention part and in some cases we're trying to correlate it with recall or sales effect. But we're still not looking at these other bio and neuro effects. We're getting people to ask us, could you combine the driver tags method with the attention method? Like we're using attention method x or no, we're using attention method y. Can you combine your stuff with that? Yes is the answer. And where this seems to be leading is it's time for addressable tv. And if you're going to go addressable, you might as well go programmatic, because that's what the agencies want. They want programmatic. It's the most efficient way. It increases their margins and it reduces their number of people that they need to support and so on and so forth.  

 
34:19 
Bill Harvey 
And if you do that, you might as well use an optimizer, and you might as well optimize on everything you can optimize on. And that's the way we're going to be carried along on the foottails of that movement.  

 
34:30 
Dom Hawes 
Talk to me about addressability a little bit. What is addressable tv?  

 
34:33 
Bill Harvey 
Well, it's the thing were inventing with next century media back in the 90s when we did the driver tag work that I described. Addressable means that you can send an ad to a specified list of people or households rather than broadcasting it. So I'm credited with inventing addressable, and now addressable is just starting to get off the ground in a big way, mostly because of streaming, because streaming is inherently addressable and there are other addressable forms in television. We're going to be part of that, I think, and that'll be a fourth use case or revenue stream.  

 
35:08 
Dom Hawes 
So, Bill, you're really busy. Mean, it's a lifetime's work based on an enormous volume of data crunching and distillation and refining. That's given you these three core sets, tags, states and types. That's driving engagement and effectiveness kind of everywhere.  

 
35:29 
Bill Harvey 
We're just starting the global rollout. But yeah, it is everywhere. We've done it for social media. We've done it for social posts that have logos in them that were started by a brand, but now people are responding to. We've done that for some agencies, influencers. It can go to print magazines, it can go to video game.  

 
35:50 
Dom Hawes 
If people want to find out more about the work you're doing or how the marketers listen to this, if they're interested in how they can find out more or get engaged in some of the work you've done, where would you send them?  

 
36:00 
Bill Harvey 
The website is RMT research measurement technologies. But RMT solutions, it doesn't have in it. It's just RMT solutions.  

 
36:11 
Dom Hawes 
So we'll put a link to that on the show notes so people can go right there and find out. And it strikes me also, I said in the introduction I was talking about, I think specifically in b two B, where the no shit Sherlock moment is ads with emotion do better than those without. But that's not always a no brainer. In b two B, there are still people who are advertising very functionally rather than using emotion. Are the driver tags and the needs, dates and the motivational types, are they attached to people as well as brands or products or media types?  

 
36:45 
Bill Harvey 
They are.  

 
36:46 
Dom Hawes 
So if there are human characteristics attached to those, and you're able to model and synthesize groups of those, presumably driver tags, needstakes and motivational types can also be used to address a business to business market.  

 
36:59 
Bill Harvey 
Indeed, there is somebody in b two B who's investigating it right now, and we've had this conversation about rational versus emotional, and obviously they're more skewed to rational appeals rather than emotional appeals. So I explained to them that if you were to try to use, like, facial emotional cues, don't expect necessarily to see a big smile on someone's face because your ball bearings have a higher tensile strength than somebody else's ball bearings. But in the frontal lobes of the brain, you would expect to see the brand attraction or repulsion effect. So you could call that a motive. Some people do call it affect or know. These are just words.  

 
37:40 
Dom Hawes 
Well, folks, that was the one and only Bill Harvey, and I am so pleased that we have been able to bring his story to you. He is a living legend in advertising effectiveness, a pioneer in artificial intelligence, and what a deep dive he's just given us. I am blown away not just by the sheer scale of what he did, but by the results too. Like, wouldn't we all like to see a few 48% increases in our work? What's more, as I mentioned earlier, there's a heck of a lot of rich detail that Bill revealed in the interview that we simply don't have time for here. But for those of you who do want to dive even deeper into the ocean of how Bill built his model and what he's doing next, please go to the show notes and click on the full interview link.  

 
38:26 
Dom Hawes 
But get yourself some scuba gear first, or maybe a diving bell even, because he goes really deep. So let's just recap a little bit of what we've heard today. Bill talked first about the need for emotion. Yeah, it's old hat, of course, to all of us and B2B's, but his point was made that it needs to be nuanced. It's got to be the right emotion in the right moment, and it's nothing without first grabbing someone's attention and then aligning it with their motivation. Tick agreed.  

 
38:58 
Dom Hawes 
Then we talked about context, the power of matching the mood and nature of the ad to the context of the media, and Bill explained why this can severely limit the brand growing power of digital, particularly in fast scrolling environments where it might be easy to reinforce a purchase decision or to remind someone to buy something, but it's much harder in, say, a couple of seconds to build a connection with a new brand. Then Bill gave us basically the history of his life's work, a 30 od year quest that started with every word in the english language and which led to the 265 driver tags. The core emotions, desires, need, states, personality, types, values and so on that can drive deep connections with brands when we align them with our audiences and the context the ad runs in.  

 
39:48 
Dom Hawes 
But more than that, it was a bit like watching open heart surgery on AI over Bill's shoulder. Like who else out there is actually explaining how they code and create an advanced AI model like this? What he also showed us was that by aligning the driver tags in ads with the audience and the context the ads run in, we can produce some remarkable effectiveness results. And not just the more touchy feely stuff like brand preference, ad recall, or even click through rates, but actual money in the bank sales, pounds and pence, dollars and cents. That is the real treasure that Bill was diving so deeply for. Speaking of which, it's high time I got back on the never ending hunt for marketing wisdom. So look out for our next episode, which will be all about how to handle the early majority market.  

 
40:43 
Dom Hawes 
In the meantime, I've got a favor to ask you. If you like what you're hearing here and you want more of it, please subscribe today. Every subscription we get helps make the show bigger and better, and that simply means more Unicorny insights and ideas for you. Thank you. See you next time. You've been listening to Unicorny, the antidote to post rationalized business books. I'm your host, Dom Hawes. Nicola Fairley is the series producer, Laura Taylor McAllister is the production assistant, Pete Allen is the editor, and Ornella Weston and me, Dom Hawes, are your writers. Unicorny is a Selby Anderson production.  

Bill HarveyProfile Photo

Bill Harvey

Harvey

Innovator & inventor with 35+ years as thought leader in media research & key solution-provider to Fortune 500 brands.

Breakthrough inventions include:
• The ADI/DMA
• Addressable commercials
• Passive Peoplemeters
• Planning/Purchasing media based on single-source Big Data matching (4 U.S. patents)

Bill is the original discoverer of DriverTagsTM the first proven scientific development of psychological attribute metatags for content (programs, movies, ads, etc.). RMT’s DriverTagsTM have the highest predictivity of Nielsen ratings, series cancellations and renewals, and individual viewer likelihood of conversion to specific programs. Most recent work being presented at ARF shows an ability to lift ROI for brands by over +35%.

Pioneer in New Media, set-top box data, advertising ROI optimization, measurement standards, privacy standards, & ARF Model for Evaluating Media.

Key consultant in development of Personal People Meter (PPM) & ScanAmerica at Arbitron. Developed first automated Marketing Mix Modeling system for General Foods.

Co-founder TiVo Research & Analytics (TRA), the first company to merge single-source & Big Data. First-named inventor on TRA’s three U.S. patents.

Co-founder of Next Century Media & New Electronic Media Science, third-party research companies that serve more than 70 of top 100 advertisers, most major cable & satellite operators, networks, advertising agencies, & other market & media research companies.

First to turn set-top box data into … Read More