Brand isn't dead, the 95:5 rule & why AI is B2B's most powerful painkiller - Jon Lombardo and Peter Weinberg

22:50The measurement, I need to measure how sticky those memories are, how likely we are to be recalled, not how many people are clicking or converting. So every aspect of the job of marketing just sort of falls into place when you get your head around the 95-5 rule. It also explains why you've turned up in costumes, right? It does. Because the point is to be remembered at the point you're thinking about buying the products, right? Well, in advance. Yeah. I mean, the thing I really love about it is coming back to the point about, you know, Silicon Valley or Galloway, they think it's technology that matters or it's product that matters or it's culture that matters.

23:23Like, I think a Ritz's an idea that has been very influential is market orientation. The customer is the person that matters. And in fact, the 95.5 rule is very customer focused. It's not talking about the funnel, like no customers at the top of the funnel or the bottom of the funnel. The customer is either thinking about your category in market or not thinking about your category at all out market. And most people are not thinking about the category. They're out of the market. And so actually you have to kind of sell to people who are not in the market to be sold to. One of the only real ways you can do it is what we're doing, right?

23:49It's building your brand around. These are people that are doing synthetic research, marketing, AI. I broadly get that about them. They're dressed up in a costume. So I recognize them whenever I see them. I don't have that need today, but maybe tomorrow my boss is gonna say, should I name my after party, you know, data after dark, or should I name it supply chain and champagne? You know, and then they're gonna be like, I need that answer really fast. And the only way I can get that is synthetically, right? And so you kind of have to market to people who are not going to buy it to you. You have to do it through what we're doing, characters, fluent devices, distinctive assets.

24:19That's just a very customer focused way of seeing the world. Like I know that they don't care. So I have to do something that's very easy for them to like low cognitive load to get them to kind of pay attention or just even notice. And then if I've done that, I've probably earned the right once they move in market to be shortlisted and make the sale, but it's very customer focused. Then there's a cashflow analysis. Well, it's both customers. The 95.5 rule anchors you in your actual customer, the in-market or out-market. But it also helps you with your internal customer, which if you're at B2B is sales and finance.

24:47So the message to finance is, hey, 95 % of people are future buyers, which is the source of future cash flows, which is what determines our valuation as a business. Businesses are predominantly valued on future cash flows. So you could say to finance, I'm not doing brand because it's like cute and I'm like obsessed with brand mood boards and personalities. I'm doing it to make us more money when people enter the market. And that's a story to finance. To sales, you say like, hey, the 5 % that we're going after today, they're mostly spoken for because there's great research from Bain and Google saying that something like 80 % of the time people just choose whatever brand popped into their head on day one of the sales cycle.

25:27So then you basically said it's sales like, hey, let's not play that losing game that everyone else is playing. Let us help you fill the pipeline so that you get more leads coming in when people actually enter the market. But maybe somewhat counterintuitively, that's not by running Legion advertising. It's by running, you know, talking geckos saying, hey, when you need car insurance, think of Geico. But people have basically for a long time said right person, right message, right time, which is wrong. It is right person. It is right message. It's wrong time, actually. It's well before they're going to buy.

25:57It's not right time. It's wrong time. If you reach them at the wrong time, assuming they're a category buyer, you'll then have the chance to sell to them at the right time. But people don't think about it. If you reach them when they're in market, it's too late. And this is actually a great finance idea. I was talking to my financial advisor the other day, lovely woman, Courtney, Marilyn. Shout out, Courtney. Sure, she's listening to the podcast. But she sent me this thing from Bank of America where they're like, it's not about timing the market. It's about time in market. And it's the same thing in marketing.

26:23It's not about timing the customer and trying to reach them right when they're in market. It's how long have you been investing in your brand, just like how long have you been investing in the S &P 500? That's the predictor. Talk about some good advice from Scott Galloway. He talks a lot about this, right? Compounding. And it's just like how long and how early did you invest in the S &P 500? I would say that, again, you can go back to Buffett here. Buffett was very early on Geico. And so Buffett very quickly understood, if I can build this kind of memory device, the Geico Gecko, if I can win the mind, I will win the market.

26:56And then that will grow my share and it'll be extremely profitable. But I may credit him with being, I mean, he's actually in many ways a brilliant marketer, a really brilliant marketer. If you go read the old papers he wrote, he's written these like one-pagers that have been circulated around C's candy and stuff. You can actually see all of the hallmarks of mental and physical availability well in advance. He gives incredible advice, I would say, on life, incredible advice on business, Incredible advice on Brandy. But he really understands you win the mind to win the market. Memory is a bit of a moat.

27:21Like we're trying to build memory around synthetic research. Our characters build a moat there. But that's not an idea that obviously is widely pursued even today. But I always am amazed that the most financial person, the greatest investor of all time, was deeply invested in Coca-Cola and Amex and Geico and Apple. Like he really internalized that in a way that's extraordinarily powerful. And so Buffett on brand is kind of like, that's you should go read about brand, not Galloway. I love your bringing it back to Scott, to my compound interest as well, because the other thing that people forget about building brand is exactly how it compounds and becomes greater and greater return.

28:00We did some research last year, actually, system one called compound creativity. But you look at this gap emerge, those that are consistent about how they present their brand to the world, the return just gets bigger and bigger. and those that keep on changing what they're saying end up flatlining. Yeah, marketers are day traders. They're changing their tagline and they're creative every five minutes. That's the equivalent of buying and selling stocks every five minutes. Just like in financial markets, it's the people. Have you ever seen that famous Fidelity study of the people who had the best performance in Fidelity?

28:34They were like these people who just kept crushing the market. And Fidelity was like, what do these people have going for them? Are they like financial geniuses? Are they mathematicians? and at the end they found that they were just people who had forgotten their password and couldn't change their investments because they couldn't log into the platform. So it's like if we could actually just figure out – That sounds like me. If we could figure out how to lock marketers out of changing their creative, if it was password protected and then we wouldn't tell them the password, all their brands would probably be 100 times stronger.

29:05There's another point we kind of should make also about, in my experience, all B2B marketers have a problem making the case to finance and sales, right? Why they should invest in brand. But you can also, it's not just that the 95.5 rule puts it in customer terms, it puts it in cash terms. So you can say these people, the in-market customers of 5%, they're your current cash flows. These people are your future cash flows. You need a steady stream of future cash flows in order to underpin the value of the business, the market cap of the business. And so you have to be balancing those two things. And if you don't, as you said, you will grow maybe in the short term, but you will flatline and then you will decline.

29:39And that's not a place anybody wants to be. The problem is that most people are not in market, in their job in market, like don't have their job or tenure long enough to make these kinds of longer term investments. I mean, it's overwrought the idea that like all CMOs lose their job in four or five years, but like you have to think in decades rather than in kind of minutes. And as Peter said, that's just not how it works, especially in a world of like chasing clicks rather than memories? Well, I've got a theory that actually brand matters even more in B2B, right? Because if I think of B2C, if you can get yourself on shelf, let's say you're a new toothpaste, if you can get yourself on shelf, you've got a chance because people might buy toothpaste two or three times a year and you're side by side, you've got some physical availability, you have a chance, right?

30:21If you're in a product category that's bought every five years and there's no shelf, like you've got a really, really big brand job on our hands. And yeah, if you look at where people spend their money, in B2C, it's well established you might do 60-40. B2B, in fact, I've been doing a few workshops recently with some senior B2B marketers asking for the average split. And most people come back with 90-10 in the performance. Now, they should be saying to me, 90-10 brand. But they're saying to me, 90-10 performance. A theory about B2B brand and where you should spend your money? Would you like to see some data?

30:58What if I told you, John, that there's a company that can do synthetic research and ask category buyers how often they buy the category? Would that be something you might be interested in, John? Yeah. And you could calculate personalized or customized or category kind of specific. Could you get them on the podcast? Yeah, we could. Actually, oh, I think they're on the podcast. Hold on. John here? John here? Yeah, well, why don't I tell you what we did and then John can show you the data. So as a reminder, what we do at Evidenza is we build synthetic samples. So you choose a category buyer in B2B.

31:31Let's say it's a CEO of a company with 10 ,000-plus employees in Minneapolis in the manufacturing sector. So you get a tightly defined audience. 3M. And then you get AI to impersonate not only that individual respondent, but you do it at scale. So you create 1 ,000 permutations of a CEO who buys ERP software. And then you can actually get these synthetic respondents to take the surveys and get it 1 ,000 times faster and much more cost-effectively than you could have if you even attempted to get a CEO of a Fortune 500 company to take a quantitative survey, which is basically impossible. So in this case, what we did, and John will walk through some of the data, is if you actually look at John Dawes' seminal paper on the 95.5 rule, one of the best things about the 95.5 rule is that you can actually immediately test the theory for yourself because it's a survey.

32:22You just field a survey to people and say, how often do you purchase the category? You know, is it every quarter? Is it every month? It's actually a pretty simple survey. The problem is nobody runs the survey because people don't run surveys. It's too slow, too expensive. So what we basically did is, John, you want to talk about what we did? God, I thought you'd never ask. You thought I'd never ask. I did. Yeah, I mean, we will produce synthetic data that estimates how many people buy in a week, how many people buy in a month, how many people buy in a quarter, a year, two years, five years. And that data is synthetic, but it's rooted often in a real piece of evidence somewhere around people talking about how often they buy the categories.

32:55This is a way that synthetic data is often modeled on real-world actions, right? But the data is very interesting because the data shows for CRM, for example, over a five-year period, 100 % of people in that category will buy the product. Cybersecurity, same thing, 100%. Cloud, 100%. Medical devices, only 71 % of companies will buy medical devices over a five-year period. And ERP, only 63 % over a five-year period. So you're going to basically try to get everybody to buy now when, in the case of ERP, maybe only three people will buy over a five-year period. So there's another 40 % who probably won't buy for another three, four years.

33:27So it really just tells you that like you need to kind of think in a five-year horizon. And that's the way to kind of program your brand and demands. But like really like it's a lot that should be brand because most people are just not buying, right? They're only going to buy once every five years or so. So I would spend, like you're saying, it's 1090. And at LinkedIn, we saw I think it was 1585. And the number, I'm not sure the number should be reversed. It shouldn't be necessarily 9010 because you have to compete very fiercely for customers in the market. John Dawes made that point to us. But I think at a minimum, it should be 50-50.

33:58I mean, you should think of it like a hedge fund, right? Like I'm hedging part of it on the future and part of it on the current. Stocks and bonds. But that's not what people are doing. Like everyone is basically short brand and they should be long brand. So we did this for like 20 categories. We did it for 10 B2C categories, 10 B2B categories. We fielded the John Dawes survey. And now we have, you know, you should show the quarterly view as well. I think that's interesting. Yeah, the quarterly view is interesting because, you know, as you can imagine, like everybody buys milk in a week. you know as we just saw only 60 percent of people buy erp in five years so like that just teases out a little bit of the difference between obvious differences between b and b and b b and b b to b and b to c but in any given period there's basically like 11 times more people in b to c who are buying the category so b to c is a very common category means to me it ought to be more activation right more kind of like lead gen depending on how you think about it and it means that brand should be the opposite like basically consumer should be b to b and b to b should be consumer but it's the complete opposite.

34:52But it is category dependent. I think the critics of the 95.5 were always like, well, is that true for every category? Like now you can look at every category and you can see it makes sense to think, you know, maybe in terms of months, if you're selling B2C, it makes sense to think literally in decades, if you're selling ERP, you know, and these are the kinds of like, these are the kinds of things that you need to know to explain to your CFO. So what kind of range? So if 95.5, I'm guessing is an aggregate or an average of of the study. What kind of variance did you see in the data from I mean, the most categories will over a five year period get 100 % of buyers.

35:30But again, like ERP was the one I think we saw where it has the fewest buyers over a five year period. And that's again, two thirds of the people buy the category. So like, that's probably like something you buy every eight years, right? Cars, you buy every 10 years, right? Which is very interesting. You buy every three years. The cars thing is interesting, though, because there's actually B2C categories that are purchased even less frequently than B2B categories, like cars look more like ERP than they do look like toothpaste, you know? And actually things like cybersecurity or CRM are like a little more frequently purchased.

35:59You know, maybe those are more like televisions than they are, they're certainly not toothpaste, but there is this whole range and it makes sense to actually think about like the category context. Well, one I worked on, which I'll test you guys out on then, is I worked on the number one energy drink in the UK, Lucozade, right? And we had a bit of a brand relaunch. I got the data out, inspired by, you know, how brands grow. I thought, well, I better check, you know, How often do you think that one's consumed on average? I would think people buy it maybe like once every six months. I think 20 % of people buy 50 % of the category usually is the Ehrenberg Bass idea.

36:34Yeah, pretty small actually. Yeah, the average purchase frequency was two. Right. So in a year. In a year. But this is the data that blew my mind. Other people probably once every two, three years. We were looking, the mistake we made is we cut the data by a year because we started with the assumption of let's look in a year, right? how many people buy and it came out to them okay fine and we had 26 penetration so 26 the uk will buy uh some point during the year i then said i said to the team just reset that for three years just just out of interest is anyone else in there 46 right so 21 in five people were buying less than once a year so when we had our relaunch campaign i i got the conclusion you came to which is like hang on a minute this is not an eight week campaign yeah this is like a four year campaign yeah like if we're going to genuinely reach everyone and that 20 is huge you know that they're the people we need to get back to the light buyers you know is where your volume comes from so even in energy drinks which you'd think would be a high frequency category that plays out what i love about this data though is and this is like an interesting segue into ai what ai is capable of when john does did his research he's a single professor who has many things going on and over the course of a month he pulled together a paper that referenced i think three categories.

37:46And now, you know, you don't have a professor in a department, you basically have a professor, a PhD in your pocket. And so there's all these things that we would have wanted to know empirically, that you can now start to establish empirically, like there just wasn't category data for 20 different B2B categories previously, there's maybe two, maybe three. And now you can produce 20, you could do it on the B2C side and the B2B side. I mean, it's like you can do it across markets, you can do 150 markets, what's the 95 five split for CR CRM? How does it vary by CEOs and CTOs? I mean, the kind of scale at which you can now field this type of research synthetically, just it's staggering.

38:21So you can take evidence for the first time, often with like a cited study, you know, that you can take it to your head of sales. You can take it to your head of finance and say, like, here's the case, the financial case for investing in brand. And it's rooted in category behavior. It's rooted in customer behavior. Whereas previously, you didn't have the evidenza to go on the offenza. Now you do. Ah, beautiful. There it is. John loves to rhyme all the time. It should be a crime. But it's not. How much John likes to rhyme. How much you whine. There's the clip, ladies and gentlemen. There we go.

38:50That's great. We're done. You do need a jingle, though. I mean, you've got the characters. We definitely need a jingle. Surely the jingle's the next one. We definitely need a jingle. I mean, we've been talking about going on the defense zone with Evidenza, going on the offense zone with Evidenza. We've been singing the Paw Patrol jingle a lot lately. Paw Patrol, Paw Patrol, get there on the double. I don't think. We do. Unfortunately, it's taken. But maybe we could do that with synthetic data. We do. We'll take it under advice for John. This feels like a little brief, doesn't it? Yeah, it's a good brief.

39:15We accept, brief accept it, John. We'll come up with something. I mean, the question, I guess, around this is, do you think that the, do you think it needs to make sense or do you just think it needs to be memorable? Memorable. Yeah, I think so, 200%. No making sense. Nothing needs to make sense. So I think we're on the offends though. People say, we'll be giving demos of our platform to clients and they'll interrupt. I'll have just shown them category entry points for aviation maintenance services. And the guy or, you know, the woman will interrupt and be like, sorry, I couldn't help but notice why is everything in Italian Renaissance branding?

39:45And I was like, well, you just answered your own question. You noticed it. Like that is the point. The point is that it's weird and noticeable. Is there any like logical reason why a synthetic market research company should have Italian Renaissance branding? Not particularly, other than me and John plan to retire to Italy someday, I think. But short of that, you know, it doesn't need to make sense. It just needs to be memorable. Les Binette and Sarah Carter in their book, 66 Ways Not the Plan. I don't remember the exact data, but it's very similar to the character idea. It's like nobody uses characters, but characters are wildly effective.

40:17Maybe it was less than 1%. I think the same thing was true. Nobody uses jingles, but jingles are wildly effective. So you could just probably build a brand on a jingle and a character. Keep that in mind, by the way. And that would be an incredibly like empirical growth strategy. The next time you have us back on your show, John, I vow to you, I make this vow publicly, we will have a jingle. We will have a jingle by the next time we're here. The question is, it will be on the offense or the defense? I'm not saying it's going to be a good jingle. You can ask the audience on that. We can get them to tell you.

40:47But I mean, going back to my good friend Orlando, which I know you guys know well. I love Orlando. His work on the left and right brain and what features in advertising get remembered. And music, melodic music comes out top right in the box. You know, it creates an emotion and it sticks in the memory. Top right is all right. There you go. All right, all right, all right. So we talked about 95.5 rule as being something you can get to very quickly in a way that you wouldn't do before. What are the kind of biggest marketing questions that people are asking you to answer with AI that couldn't be answered before?

41:17Now they can because evidence are here. I think a good way to think about what should you use AI for is basically like how painful is the traditional equivalent? So in other words, like market research, very broad category. But within market research, there's types of research that are pretty fast and cheap. and there's ones that are really slow and expensive. So if you're looking for efficiency gains from AI, the largest gains will be on the most complex, slow and expensive forms of research. So sure enough, like that's mostly what we're doing. You know, clients come to us for segmentations and persona research.

41:52Personas, I mean, they're usually put together qualitatively or talking to sales. If you try to do it quantitatively, like a segmentation, could easily spend half a million or a million dollars. you know, it could take up to a year. So now with AI, you could do 100 segmentations a day across 10 different categories and 100 markets. So the efficiency gain there is staggering. Category entry points is something we're doing a lot of work on. So John and I used to do that with real humans at the B2B Institute. It was phenomenal, but, you know, cost 200 grand, took six months. Now, again, you can do that across 11 categories, 150 markets, It's fraction of the time, fraction of the cost.

42:32So I think it's a good mental model in general is, you know, and this happened when we first took our business to Linda Boff, who was the legendary CMO of GE and now is on our advisory board. We told her about our company and she was like, the reason I love that idea is because, you know, you can be a vitamin or a painkiller, famous thing in startup life, and you want to be a painkiller. Like right now in AI, everyone's like, how can we get AI to develop the creative? That's marketers' favorite part of the job. That's the last part of the job they want to give to a robot. Running a quantitative survey that cost you half a million dollars, you know, took six months and maybe got you fired because it came back with the results you didn't want.

43:11That's a real pain. You'll happily hand that off to the robots. So I think it's just a good mental model for AI in general is what is the sort of worst part of the supply chain? That's where you should apply AI first. I mean, the way that some like VCs talk about categories you should try to build a business in is low NPS. So nobody likes it and nobody rates it. And then high fragmentation, lots of vendors. And those things are related. And market research is the most obvious place that that's true. There's all sorts of problems with, on the one hand, survey fatigue. You know, people get tired taking the surveys.

43:43On the other side, on the client side, it's survey remorse. I didn't ask the question properly. I got back the research. It didn't tell me what I wanted. And in some ways, we're solving both of those problems. And once you reduce the risk on kind of the fatigue side and then the kind of client side, then you lower the kind of cost of asking questions. And people, it turns out, want to ask a lot of questions. And the reason they want to ask a lot of questions is they want to understand what their customer thinks and says and does and, you know, who they talk to and what matters. And so I've been amazed by, on the one hand, as Peter says, there are very substantive things people want to figure out, like who's on the buying committee, how long is the process, why do they buy?

44:21You know, what should our segmentation look like? And then there are, on the opposite side, there are very superficial questions, which in their own way are just like a wonderful way of understanding what the customer thinks, which is again, like the date after dark or supply chain and champagne. What should I call my after party? What should it say when you submit a form? Should it say thank you? Should it say, you know, I mean, these are tactical things where previously the cost of complexity was so high or the cost of curiosity was so high, I should say, that, you know, you reserved market research, not just for the biggest brands, but the biggest questions.

44:51Now, as that comes down, all of a sudden, you can have every single marketing decision you make informed by some sort of quantitative research, which nobody really does. I mean, we did synthetic research on this. We asked synthetic marketers, how often do you do, you know, quantitative research to inform a marketing decision? And I think it was like 10 % said, yeah, we do that, you know, always. Microsoft famously used to say they wanted to put a computer on every desktop. And I think our idea would be put a customer in every marketing plan. Because right now, most marketing clients do not have voice of customer or data from the customer or verbatim from the customer in any meaningful way.

45:27Because the customer is too hard to find. But they should, right. Have you ever heard the Jeff Bezos empty chair thing before? Yes, yes. That's a great example. Jeff Bezos leaves an empty chair in every room to represent the customer. But there is a reason that the chair is empty because it would be deeply impractical to fill the chair every single time you need to make a decision with the customer, even though the customer is the most important person in the room, as Bezos is saying. now with ai you can literally fill the chair for every marketing decision you can have the voice of the customer and you know it's very paradoxical you'd be like how did the invention of robots make marketers more focused on humans and their customer but i think it's absolutely what's happening i don't know if you have the same program over here we have like who wants to be a millionaire program on tv we do and they have this ask the audience right where you get stuck on a question you press a button and you get to ask the hundred people that are there live in the show and that That answer is usually the right one, right?

46:17Because it's wisdom of crowds. Wisdom of crowds. It's like that, isn't it? It's like the same thing. You get to press a button every time you've got to make a decision, ask the audience. Yeah, focus group on demand, customer on demand, quantitative research on demand. I mean, it's wild. The internet is basically the world's biggest market research study. And if you think of it in that way, you can start to ask all sorts of interesting questions, you know, no matter how big, how small. I mean, that's the number one thing I think every CMO should think about is how can you bring the voice of customer into your organization?

46:46for as many big or small decisions as you want. But the truth is, as Peter said, you can use it for, people traditionally use it for one big survey a year. You should have one big survey a month. You should probably have one big survey a week. I mean, it's just, that's where the world will go. And frankly, like we'll all understand our customers better. We will build better businesses. Customers will be happier as a result. Like economies will probably grow. I think there's a very positive way to think about this. The closer you are to the customer, the better your business does. I think that's probably my big takeaway from like being in this AI space for a year is like you've got the critics who think everything's going to get worse and then the optimists are kind of like oh well it'll just be the same but faster and cheaper I think John and I at this point truly believe it's actually going to be better like lab grown marketing will be better than man-made marketing it will not just be faster and cheaper it will be more flexible so you can go back and change the sample change the questions It can be informed by marketing science.

47:41So you can be like, how would Ritson position the brand and bake that into your positioning research? It can be more financial. You could say, hey, go read all the earnings calls in this category and append that financial data to the quantitative research. and now all of a sudden you have a market research report that is not as good as the traditional equivalent, it's probably 1 ,000 times better. It's faster, it's cheaper, it's more flexible, it's more financial, it's more grounded in all the marketing books that nobody took the time to actually read. I mean, it's going to be great. Well, on that, we were working on a similar thing with System 1 where at the moment you can afford to use the approach maybe on the top 5 % of your communication.

48:21And we're like, well, hang on, what if we trained the AI on Orlando Wood's work? on all the human responses. We've got 160 ,000 times 150, whatever that number is. It's a lot, right? So all the emotional responses and we can tag everything and code it all up and that sort of thing. Imagine if you could just test everything in the world, right? This is why AI will not be average because you won't ultimately train it on the average thing. You'll train it on the Orlando Wood or the Jenny Romanek or the Jeff Bezos or the Warren Buffett or the whoever it is, right? You'll train it on brilliant people and you'll get brilliant insights.

48:57You already see that. I mean, the new models are not trained on the average insight. They're usually trained on PhDs who are grading answers. And so you're getting PhD level answers back. I mean, the problem isn't now that the answers aren't good. It's that people don't have good enough questions to actually get the intelligence out of the Or that there's actually too much answers and they're like drowning in the data, signal versus noise. Drowning in data, but parched for wisdom. Beautifully put, John. There we go. It's there, isn't it? I loved your point earlier about painkillers versus vitamins as well.

49:26Because, I mean, to give our good friend Galloway another outing as well, there was one thing you said on the podcast that's been viewed. That was right? One thing that was right? Well, that's been viewed 10 million times. What is it? I'm dying to know. He said the less sexy the industry you're in, the greater your return on capital. Everyone wants to be actors and bartenders and, you know, models, et cetera, et cetera, which have a very short lifespan. I've got a rhyme for all these things, John, and the rhyme for this is boring is soaring. Well, but you're exactly right. Like Galloway, that is good life advice, but it's also good advice to anybody looking to start an AI business right now, which is like, look for the least sexy thing that requires some sort of real subject matter expertise and that everyone hates.

50:13When we go into a CMO today and we're like, hey, are you happy with your segmentation work? Nobody's like, ah, love it. loved spending a million dollars on that shitty segmentation. That's like a universal pain point. Sorry, what Peter meant is there's no money in B2B. It's very boring. You should avoid it at all. Don't do an AI research business. There's no money there. You'll hate it. Stay away. Stay away. Stay away. Yeah, but it is. It's true. It's too sexy for everyone. He is right about that. He is, totally. It's a brilliant insight. I mean, there's like a company called Viva that's a CRM just for healthcare.

50:46I mean, Salesforce is one of the world's biggest businesses. And yet in that market, that market is so big, it can afford a publicly listed company. It's a pretty big company just in the CRM category. You know, so like there's riches and niches. That is insane, isn't it? Yeah. But you're absolutely right. Go for the boring categories. There is. Solve the pain. And don't be scared, you know. People think AI is just going to like eliminate their jobs. But in many cases, it's doing jobs that weren't getting done in the first place. You know, I mean, segmentation, category entry points, great example.

51:17It's not like people were doing that job, now it's going to be replaced by synthetic research. They weren't even bothering to do that job because they didn't have the time or money. Why don't you talk about the bank tellers thing? I always love that story. Well, before that, I went to an insurance conference recently. It was a marketing insurance conference. And so I was just like, how many people in this room have done a segmentation study? There's probably 60 people in the room. I think probably four people raised their hands. So to Peter's point, nobody's doing segmentations. So if they get to do a segmentation, it'll literally be the first time they get to do it.

51:44And AI will probably be the first time they can actually do it, right? So it's all these things that were never possible. So it's not removing those jobs, it's creating those jobs for the first time. And that's actually really exciting. There's actually a whole new set of category entry points, especially in B2B, that are kind of like birthing as we speak. There's the question like, hey, I need this answer by tomorrow about, let's say, should I show this message when I'm sponsoring a post? Or should I show this message? Or I want to do a segmentation. Or I need data on the 95.5 rule to bring to my board meeting next week.

52:13Previously, there just wasn't the ability to do that. So it's probably going to create a market that never existed. And honestly, when it's a digital market, it's usually 10 times bigger. So it's going to create not just a new market, but a huge new market in B2B. I think that's what people are missing is you're growing the market. You're making, you're creating accessibility to something that up to now hasn't been accessible. If you work in pharma, you want to talk to an oncologist about whether this message is going to make them recommend this drug over that drug. That's something that would have cost a lot of money.

52:39Yeah. Well, that's actually very interesting. So there's things we didn't anticipate about the business. One of them is like how big it's been in pharmaceutical categories, because John and I don't come from a pharma background. But in pharma, it's actually like in many cases, you cannot talk to the customer, not just because it's hard to reach a nephrologist. And we have literally done research with synthetic nephrologists, but because there's so many like HIPAA rules and regulations or, you know, there's risk. Like you don't want the small pool of rare disease doctors to know what new products you're thinking of rolling out.

53:10Now that you're talking to not, you're talking to AI doctors who are not real people, all of the risk is actually removed. Like you skip months of regulatory compliance and like reputational risk of new product ideas or rebrands leaking to the press. So it's just like another example of how it's not all just faster and cheaper. It's like better in ways that you wouldn't even anticipate. We're dealing with complexity there because, you know, you might have government regulations, you know, you might have very technical products that would take you months to even understand before you even get to the answer, but it's able to kind of think through all that.

53:44We have a medical devices customer who told me that she's thrilled that we let her talk to fake customers because she tried working through legal and procurement to do research and she literally couldn't do it for nine months. So she's like, I literally cannot talk by law to real customers. So I'm delighted to talk to fake customers, you know? And that's a really fascinating, and as Peter said, kind of unanticipated, unexpected idea, but that just makes something new for the first time, something possible. I mean, it's actually, it is physical availability in a way, although it's more like digital availability in a way that was never possible before.

54:15So like if you pair the mental availability with the physical availability, I'm told big brand or small brand, you get growth. There you go. That's what the book says. I think it's true. The book is right. Yeah. Right. So exciting first year. Congratulations on where you've got to with the business. And AI is just changing all the time, isn't it? I mean, I know I was at our good friend Colin Fleming. I bumped into him at Adobe. In fact, his episode came out last week. It was a great episode, Colin. Yeah, Colin's great, isn't he? I mean, what a brilliant way of, he's got such a good way of communicating complex things in simple ways.

54:45I was going to say he was raising millions at 13. That's amazing, right? That is a brag, isn't it? I mean, that's legit. I don't meet a lot of former race car drivers who become CMOs of gigantic enterprise technology companies. Or as my daughter would say, that's a flex. That's pretty impressive, isn't it? But even at the conference, I was learning what agentic AI is, which I hadn't heard of before. But even the language is changing, going from a kind of, is it deterministic to probabilistic in terms of like not just answering a question, but actually thinking through a brief and an issue and coming and solving it.

55:18Where do you guys see the big change coming in AI over the next year? I think the big change we've seen in our business. So like a lot of times we'd be presenting our product to clients and they'd be like, this seems too good to be true. Like, what is it not good at doing? Which is a great question. And we always said there's a lag in the data with a lot of these large language models. So it's not very good at real time analysis. In other words, like they just announced a new pope while we're filming this podcast. It wasn't pizza ball. It wasn't me. Sadly, I was passed over. An American pope. Who would have thought?

55:52And there's been a trade deal with the US and UK. Yeah, so a lot's happened today. We'll talk about tariffs in a minute. So if you were about to poll, you know, Catholics or supply chain managers, you know, if you told you it's not good at that because it doesn't know that happened. Like our synthetic respondents have kind of like been in a box. They don't know what happened probably in the past two to three months. What's happened now is that almost all the models have become real time. And so we've had a few clients now where we've been running these tests where, you know, like let's say they want to know how people are feeling about the tariffs.

56:23Well, again, think about the traditional equivalent. First of all, trying to, you know, survey supply chain managers about what they think about tariffs, it'll take forever, cost a ton of money. But, you know, the news on the tariffs is changing every single day. So by the time you get the data back, it's like, when was this even based on? Now that the models are real time, you can literally like take a news event and say, how is the market or how are CEOs or CMOs or CTOs, how are they responding to this news? And I think that's very, very exciting, because now you can start to get these like, you know, both timeless insights and timely insights to inform your marketing decisions.

57:00Another thing people worry a lot about, which to me is a bit of an outdated idea, is hallucinations. Like we were on a call today and somebody brought up hallucinations. And Peter and I always like the joke, there is a hallucination problem. There is. It's with people. It is. You know, the models have gotten much bigger and much smarter now. And so their hallucination rates are literally in many cases, one-twentieth the hallucination rates you would get from a person if you surveyed them. So like there is a hallucination problem. It's now the opposite of what it used to be. Did I tell you my Toto Toilet story, John?

57:28I don't think you have. So I was in this demo. I was doing a demo. And there was just a guy who was like agency guy, super skeptical of synthetic research and kept trying to do like gotcha stuff. So I was talking to a synthetic person in the UK and a synthetic person in Germany in this synthetic panel. And the guy, the agency guy, the skeptic was like, ask them if they've heard of Toto Toilets. So I was like, all right, sure. So I go in and I type in, have you heard of Toto Toilets? The German woman says, yeah, actually, I just remodeled my house. We were thinking about Toto toilets. It was too expensive.

57:58British guy's like, yeah, I've seen that in some upscale hotels. You know, I know about Toto toilets. The agency guy's like, gotcha. I gotcha. And I'm like, what are you talking about? And he's like, that's a hallucination because Toto toilets are only sold in Japan. It's a Japanese brand. So there is no chance people in the UK and Germany have heard of Toto toilets. And I was like, that's super funny because I live in Brooklyn and I have a Toto toilet. So like, again, the human is hallucinating. The marketer is hallucinating about what they think they know about the brand. The robots, the aesthetic research is actually right.

58:33You know, so it has a lower hallucination rate in some cases. So that's been a fun nuance. That's a great one. Yeah. And there's been a lot of those moments where people are just like, no, no, no, no, no. Or we have times where, you know, marketers will present, a CMO will take our research to someone in marketing and they'll be like, oh, no, no way. And then they'll take it to someone in sales and salesperson will be like, oh, yeah, we hear that every single day. So there's just so many assumptions that humans have. You know, that's going to be one of the interesting wrinkles with synthetic research is, you know, you actually see this in medicine.

59:04So there was this newspaper article in the New York Times of doctors with AI and doctors without AI and then just AI by itself. And everybody wants the story to be that the doctors with AI were better than just the pure AI, but the pure AI actually won by a large margin. So if you're having an operation, you want the AI. Yeah, because the doctors were overconfident in their opinion. They kept overriding the AI. So it's like the human is not always right. The machine is not always wrong. There's a different angle on the doctor story too. There's another study on doctors versus AI. And in fact, people enjoy talking to and spending time with the AI more than the doctor because the doctor is very busy and very tired and has to get to the next patient.

59:45The AI doesn't have to do that. The AI is infinitely patient. It's infinitely empathic. So there's a weird way you can start to think about AI is like it has a heart. Like I'm not surprised it has a brain. Like there are large language models. They're trained on tons of data. They're kind of genius. But it actually, that's not surprising. It's the heart that's surprising to me. And then also like another thing that's a very big trend obviously is the multimodal nature. So like they have brains, they have hearts, they have mouths. They can talk to you. They have eyes. They can look at stuff. They can listen to you.

1:00:11And so like you now basically have, like they are called neural networks, right? And it is called artificial intelligence. And you're kind of getting, you can very quickly see you're getting kind of a digital twin, so to speak, that is capable of doing all the things that humans are capable of, which is that in some people inspires hope and in other people it inspires fear, you know? Well, I think it was a bit of a branding thing. I mean, coming back to Colin, I thought the whole agentic AI sounded a bit scary. It sounds like kind of like the FBI taking over matters, right? But he just said, think about it as your own personal minion.

1:00:41And I was like, oh, right. Oh, that's great. Somebody who can do all the work for you, all the boring tasks, doesn't have to sleep, gets on with it, makes your life better. Yeah, the marketing of AI is very interesting. And, you know, these Silicon Valley companies that are way ahead of the industry are going to have to figure out how to tell normal everyday people what it does. And you're going to see all – I mean probably the best metaphor I've seen for, you know, agentic AI is this idea of self-driving computers, you know. So it's like you can sort of begin to understand what that means. It's that you're not asking the computer what – Galloway is not going to say, hey, what hotel should I stay at in Istanbul and then go book the Soho house.

1:01:17The robot is going to go self-drive computer, going to go book its own, your reservation for you and probably at the lovely suite that Galloway likes to stay at at the Soho house. I mean that's obviously scary to some people because they see that as I'm going to lose a job if I'm a driver, right, or if I'm a booking agent, I'm going to lose a job. But there is lots of examples of automation. Like the history of technology is basically the history of automation, right? where like at some point you had ATMs and that replaced bank tellers. There's actually more bank tellers now than there was in 1960.

1:01:46And then you had kind of like self-checkout at grocery stores, but there's actually more cashiers now than, you know, there were in the 1960s. And the same thing is true now with Amazon has robots in the warehouses and there's more warehouse jobs. So automation has a weird way of creating more jobs, not removing jobs. I mean, that's the hopeful case for it, obviously. Like we have to see how it goes. I don't know if that's hope or cope, but that is an interesting way of thinking about it. It's one of them. Well, I think that's a perfect place to end, actually. You always say, always end on a high, as they say, which is good.

1:02:15And I tend to agree with you, actually, because history tells us that with every technical revolution, as you say, it creates more jobs. They're just different to the ones that are replaced. But we haven't made ourselves redundant yet. Not yet. Brands will never die. I don't think marketers will ever die either. Well, listen, I look forward to the new Evidenza jingle next time we meet. Thank you. That is our vow. I promise to you. The brief is out there. But thanks for sharing the updates. Go on the Offenza. with Evadenza. And congratulations on an amazing first year. Thank you so much. Thank you for having us, John.

1:02:44Lovely to see you, John. Have a good one. Take care. Thank you very much for listening or watching Uncensored CMO. I hope you enjoyed that. If you did, please do hit the subscribe button wherever you get your podcasts. If you're watching, hit subscribe there as well. I'd also love to get a review. Reviews make a big difference on other people discovering the show. So please do leave a review wherever you get your podcasts. If you want to contact me, you can do. I'm over on X at Uncensored CMO or on LinkedIn, where I'm under my own name, John Evans. Thanks for listening and watching. I'll see you next time.