AI: The future of marketing & the death of the CMO
Jeremy Waite, Evangelist at IBM talks navigating the CMO role; including AI being the solution to what can feel like an ongoing battle for progress.
Jeremy: Hey everyone. You got pictures, maybe? Nothing coming through? So you did a Q&A first? That'll be easier. Let me talk to you about what's in this. I was actually going...I'm going to do a slightly different presentation but cover some of the topics that we were talking about. Hopefully, what you're going to see in a second is Beth Comstock on the screen.
Any of you know about Beth Comstock, she was the CMO of GE. She one of my favorite people in the industry. I'll just make sure that what's in it is going to work. We going to try and do some unusual stuff here. This may or not work. It's a bit risky. A couple years ago in the industry, Beth Comstock was really kind of regarded as a person that was going to reach this tipping point in the marketing industry.
And the reason for that, there was Forrester and Gartner and IDC, a bunch of agencies kind of looking at executives in marketing like Beth. And she became the senior vice chair with GE, one of the biggest brands in the world. And what that looks like was going to happen, in sort of 2012, 2013 was that was a tipping point for a number of reasons.
First of all, that the future CEOs were going to need CMOs because they could understand people and numbers. You know, that intersection of humanity and technology. Normally marketers are good on one side and the analytics, and data guys, and IT guys are good on their left brain. Right brain are seen with marketers. Understand people and numbers.
These are going to be the future CEOs because they understand that intersection in the middle. And she was lauded as being that kind of first person that goes through. And also this was going to haul this new dawn of women in the boardroom and more gender equality, and all that kind of stuff. And it was so fantastic. And she's still vice chair, she's still doing amazing work.
And I thought about this for a while, as I've worked for companies, that have all got, you know, female CMOs. I work for a company at the moment that is run almost entirely it seems by females. From the kind of the VP level of my boss and my boss's boss, all the way through to Janie, all women, which is unusual in technology as well.
But something happened a couple of days ago, and I saw what I thought was going to be one of the maybe the crucial articles I've seen in the last year or two. And if you've not seen it, I encourage you to have a look at it. But basically, what it's saying is a whole number of things. And this is based upon interviews going back to 2012.
We're going back to when Beth Comstock became the CMO and this was like what the future marketing going to look like. Twenty-eight thousand interviews showing a couple of things. First of all, it shows that CMOs, 40% of them been in their job less than two years. It showed that 57% have been in their job less than three years.
And when you consider that digital transformation, which we're going to talk a lot about today, takes over three years to mature. You basically have senior leaders in marketing in a role who aren't going to be there to see out a lot of the projects that they are putting through. And as a result of this, there's a number of other reason why.
But as a result of this, it's now claimed the CMOs are the least trusted people in the boardroom. They're an all-time low. This isn't anecdotal. It's not opinion. There's no bias. This is just based upon a lot of senior executive interviews that show 80% of CEOs no longer have faith in their senior marketers.
It's pretty scary, right? So you compare that with the CIO, CFO, CTO, that drops down to about 10%. And I think we're not going to go into this today. We more than happy to do it in breaks or Q&A, or maybe the panel later on. I think one of the reasons for that is hype cycles. You've all seen the hype cycles, law of diffusion of innovation. The Gartner hype cycles comes out once a year.
Probably the best of research that comes out every year, I think, for marketing. It shows the future of where we going to go, who's invested in what technology, where the tipping point is. If 14% of an audience don't adapt to certain technology, it falls off the cliff. There's no ROI until eventually, it matures and you kind of personalisation, email marketing real time, all that kind of stuff.
What seems to have happened is that marketers have been so obsessed. We're getting seduced by shiny things, that they were trusted originally as this explosion has happened within marketing. And social blew up from 2010, and a lot of people haven't seen the ROI. They've not shown economic value or any real impact on the business. And now it's come down, and now they're least trusted.
And I think that whole thing about the role of the CMO...it's not actually that the role of the CMO is a bad role. It's just that it's so poorly defined. A lot of it is not is setup to make a real business impact. And what's happening in this report is also showing things like CMO is now really just relegated down to advertising, social media coms, looking a lot PR and brand.
What they're not doing anymore is responsible for pricing, infrastructure, strategy, in store, and the real impact stuff that's going to actually drive the business forward. There's been this tipping point that seems to have happened. Good debate stuff for later on. Coca-Cola agree with it. They think that CMO is a faux to the degree that they canceled the role altogether.
You probably saw...sorry for swearing in camera. Coke doesn't even have a CMO role anymore, right, you all saw that. They've split in half. They said that the sole purpose of marketing is to make money and save money. Peter Drucker said years ago, "Business has only got two functions, marketing, and innovation."
Coca-Cola has looked at that, looked at the changing stuff and said, "Actually, you know what? Because it's not relevant for 2016, 2017, and the world is going to change so much over the next five years, we need to redefine the role altogether." So they've now got a chief growth officer and a chief innovation officer.
One is on making money, one saving money, one is building cool stuff, one is growing economic value, like financial value. So I think that's again something else that we could look at later on about how exactly do we built real value with the responsibility that marketers have gotten. I'm going to show you some technology in a second. Hopefully, this works. I've only demod it six times.
This is brand new technology that IBM have spent an obscene amount of money, and we'll get into that later on. Demod it six times. It's live. It's using an obscene amount of data. It's all powered by Watson. It's a world's first AI powered marketing assistant. No one else in the world has anything like it. And the six times I've tried, it failed three times. It's a machine learning technology itself.
If it doesn't work, we'll see where that goes. So let me just give you like five minutes of intro. I work for IBM, not just because we've got some really cool shit, which obviously, we do. I've been in the industry, a little while. I was head of strategy at Salesforce and head of strategy, Adobe before then. There's a favourite quote of mine from Simon.
He says, "The goal in business isn't to sell to people who need what you have. It's to work with people who believe, what you believe." Anyone can sell to people who need what you have, we can all do that. But how do you work with people who believe what you believe? Agencies, partners, that belief system.
Exactly 12 months ago, I was sitting in front of you guys for the first time giving my first IBM presentation. And just like today, was very rushed. And I arrived about 5 minutes before I was due on, but last time, I had come straight from hospital. Remember? Any guys in the room remember I had just whisked in from hospital. I had been up all night. I had my twins, I certainly did.
I don't think I showed you any photos. Let me show you a photo, a date-stamped photo of exactly what happened the last time I gave a Jellyfish presentation. I'm standing here with my little girl, right? This is Matilda. Oh, fuck. I knew this would happen. Come one. I stood with Matilda, she wasn't supposed to survive, naught percent chance of survival.
NHS never says naught percent, they always give you a bit hope, 10%, 20%, 30%. It's not going to be good, but it's certainly not going to be naught. So she wasn't going to make it. From 20 weeks all the way through. Long story short, we can share over wine later on. Lots of scans. Lot's surgery basically to kill her, to protect the other twin because she had so many problems.
And she was born at two pounds and she was beautiful, and she survived. Had to have heart surgery when she was a couple days old. You never have heart surgery until you're two kilos. She's two pounds, again, she's not going to make it, and she did. And I had just come from the hospital having talked to you guys as well.
And the reason I joined IBM was because what actually happened, it was a very long process of me trying to come over and you'll see where this is going in a second. They were talking about the data that comes off my little girl. Because I was saying, "As amazing as this is, as beautiful as you guys have, you've saved the lives of my twins.
Why is it possible that the data didn't show that she could have survived?" And it sounds like a really stupid thing to say in a middle of such an emotional situation, right? But I'm trying to detach myself emotionally from this thing because I'm an absolute wreck. And basically, they were describing the single view of a patient, heart monitor, all the TPN, and everything from the drugs coming in.
All the stuff that is coming in, historically, from all the little girls that had exactly the same issues as Matilda. And they described this thing about this is how we try and predict 24 hours ahead in NIKU [SP] how to save the lives of little girls and little boys. I was like, "That's amazing. How are you doing that?" They said, "Working on a project with IBM." "Oh my God, that's fucking awesome.
I want to work for that company." That's how this all happened because the single view of a customer in marketing is exactly the same as the single view of a patient. It's the same across lots of industries. If you look in health care and IOT and climate change, and we're doing a lot of cool stuff. We've spent 6 billion on Watson.
We've got a thousand research scientists that try and build the technology behind it. We've been working on this for quite a long time. It was founded in 2011 but a lot of it is out of the thought leadership that's come from the five noble prize winners that we've got in the business. So when I saw that and realised that technology is also what we've taken out to our marketing customers, I was like, "I want to do that.
I want to stand in the front of people and talk about how this is really cool." This is the first time we put them together. Isn't that awesome? This is them today. Aw. So if the rest of this presentation is shit, it doesn't matter because I've got that. Thank you very much. Right. Marketers have a big responsibility.
Goal in business isn't to sell to people who need what you have, it's to work with people who believe what you believe. We are in a very interesting space at the moment. We're going to get into this throughout the rest of the day. I think this is wonderful. This is the AI startups augmented intelligence, artificial intelligence, whatever we want to call it. We'll debate that later on as well.
This is the AI start-ups that are mainly powering a lot of marketing businesses, and there's 100 of them in 2017. Interesting slide, not unlike ones that you may have seen before. Or you're probably sick to death of seeing if you've been in the industry a little while. This was Scott Brinker's slide, just as I started getting into enterprise marketing for 2011, exactly the same 100 logos in 2011.
Just like in 2017, there's 100 AI logos. 2012, 346, 976, 1,876, 1,376. In 2017 that's 5,187. 4,900 of these companies are unique. There's a lot of companies, they're in multiple categories. Fifty percent of the logos on this slide are start-ups, and 50% of those have got less than three staff, 50% have got less than three staff.
So when we talk about like CMO's are so overwhelmed, and then prepared for all these challenges that they going to face over the next five years because of disruption, because of dark social. You know, 90% of consumer conversations are messaging apps, so command centers don't work anymore. End of 2019, over a third of browsing behaviour, is going to happen without a screen.
So everything is about voice. We've been even speaking to some of the biggest Telcos in the world, one in the U.K., one in the U.S. They're already having conversation about when do we turn off the website because if the user experience is so seamless, you don't need one. So we're going to look at voice in just second. But the issue is that people are trying to navigate the way through this.
This is why you need guys like Jellyfish so much just because how are brands supposed to do that? How on earth are you supposed to do anything with any sense when you've got so many people to choose from? And what we're seeing is a phrase called "Dark Martech". If you're interested, the guy that created all of those slides, Scott Brinker. He's @chiefmartec on Twitter. He's got an incredible blog.
He wrote a book called "Hacking Marketing." One of his terms is "Dark Martech" and he's basically trying to say, the industry is going to change so much over the next 5 than it has over the last 20. And one of the problem is people are buying their way into technology that doesn't work. Or it might look like it might works, by start-ups giving a really bad ass pitch.
And then when you try and put it together to build your own marketing cloud, you realise it's all instructed data and you've got no single view of a patient or a customer at all. And the CEO isn't happy and then the CMO gets sacked, and then it just goes back down again. Interesting stuff, isn't?
So, I said I was going to show you a demo. Hopefully, this is going to work.
This is Watson, you're going to start seeing this little circle in a lot of stuff. You actually just talk to the ad and it responds within the ad unit itself. You could tell it what's in your fridge, it tells you what to cook. We're doing a lot interesting stuff with it.
I'm just going to show today just the most basic use case I could think of, just to make it relevant. What would have been easy for me was to talk about the big IOT climate change stuff, and we're trying to work with TED to try and do this great big education program. And AI is going to change everything. That's great. Well, you guys will probably will go, yeah, but okay.
It's interesting but doesn't mean anything to me. This is technology that, A, is live. It's available today to any customer that has marketing cloud, you know, within our campaign automation tool. I'm not in sales. This is no pitch. I do apologise. But this is Watson marketing assistant. It's the world's most powerful production-based AI.
And it's just like having the world's best marketer next to you all the time. I love this because it's like imagine someone stood next to you all the time that knows every campaign you've ever done. Every metric that's ever been published. Every single customer segment. It doesn't how many million you've got. Watson can process about 10 billion records a second, which is about a billion people every day at the moment.
So what we've done is we've embedded Watson within our marketing cloud. So I'm just going to take 10 or 11 minutes to flag through these pretty quick. And if you've got questions we can do later on. Just to show you what this might look like if you build another campaign. Because what usually happens is people make decisions with their hearts.
They always have done and then they justify them with their heads. Consumer are the same, especially anything that's passionate. Especially anything that is a lot of B2C sales, music, TV, fashion, film, sport, a lot of this stuff that gets shared at scale the most. Especially what lives within private messaging apps the most.
Often, that's driven that's driven by some type of emotional decisions. So what we've done, and we're not going to demo it just now, but one of the big engines that drives Watson is emotional personality insights. We don't really care as much about the 3,000 things you may or may not know about your customer. That's all important.
But in a world where only 50% of people want to give data to marketers now, most people say, "I'm not giving you any data at all so you give me good experience." It's not enough just to say the customer journey starts in email. Customer journey started miles before that, but you got no data, and it's all anonymous.
Gartner calls it personification, it says it's going to go in the biggest disruptors of marketing over the next few years. Right message, right person, right time, right channel, when you don't know anything about them. You might know a few anonymous attributes. But if you know anonymous attributes about a big enough audience and you can make a lot of assumptions, that's basically what Watson is doing.
But we do that based upon emotions not on engagements and clicks. What mood are you in? How excited? How extrovert? How confident? How authority challenging? You know, and we can look at emotional attributes of the person and then figure out the next best option that could serve pre-roll or an ad or an email or a YouTube banner or whatever else.
So what's happening here is part of the engine is behind the scenes with this campaign. Now I've hove it over Sleek Fit Three launch. So this is just email campaign. It's very basic. It's exactly what you'd normally expect to see within our marketing solution. This is probably about 18 months ahead of anything anyone else in the world has got in the moment.
A lot of people talk about AI, really all they've just got is very good predictive engines. This is a true AI powered by machine learning. So if we hopefully try and wake Watson up, we can talk to him. Hey, Watson.
Watson. Hello. Would you like to learn more about campaign Sleek Fit Three launch?
Jeremy: Would I like to learn more about campaign sleek fit three launch? I certainly will. He's already opened up an exact of the context of where I am. As I do this, and I'm going to try and do it really quick and respect your time and breaks and stuff. Bear in mind that if this isn't perfectly seamless, no one is going to use it. It has to be as good as an assistant that's sat next to you.
This is why the machine learning bit is so important. The companies that are already having the most success with this. We've got it rolled out to a few big companies at the moment. The ones that are spending about $7 in tech...sorry $7 in people for every $1 in tech, $7 in people for $1 in tech. And by people, that's agencies and partners as well.
The companies we're seeing that don't have success in this type of technology are the ones don't pivot into tech, and it's a little of implementation in people afterwards. So that ratio seems consistent across the industry, seven to one. So and that a lot that in this case is machine learning to try and get Watson up to speed. Yes, how does this drop compare to all of the campaigns in the past year?
Watson: I do not know what drop means.
Jeremy: Of course, you don't. When I was Adobe campaign, was referred to [inaudible 00:17:41] when that was all rebadged and I've always called an email campaign a DROP. So we've got to try and show Watson exactly what I mean. So every single team, every individual, every group, every agent is going to use different language, different slang, different phrases.
You train Watson so whenever you log in he knows exactly what type of language you like to use, just like an assistant would when you're training on the job. Drop is the same as campaign.
Watson: Okay. Got it.
Jeremy: So how does this drop compare to all campaigns over the past year?
Watson: This campaign is ranked 238 out of 250 campaigns over the past year.
Jeremy: Well, a couple of things have happened there. First of all, we had a good feeling that this was wrong, we didn't know why. Executives made decisions with their hearts not with their heads. We had a good feeling but we didn't know why. So we couldn't actually go into the data and try and figure out, which is why we've just got to ask using natural language to try and figure out what the route might be.
So something else has happened. I said how does this drop compare. This. You guys know what that means because you're all human, a computer doesn't. If you look at AI machine learning intelligence, this is often a very complicated thing for a computer to understand what you're actually referring to.
And Watson has seen straight away I'm referring to this campaign we've already opened up. But what he's also done, he just said your feeling is right. This campaign is ranked 230 or 250, it's not good. We know that something bad is happening, someone has trusted us with their email address. We don't want to piss them off. We want to give them the best experience possible.
But what we can't afford to do is bounce backwards and forwards to agencies and data sciences, backwards and forwards to segment the audiences and figure out the problem. This is why we got to try and fix it as quickly as possible. Now, what he's also done is if you just see behind here, each campaign had 273,000 records.
Plus 250 campaigns, that's 62.5 million records he's searching in a couple of seconds to see every single one you've ever sent, and he's looking at them all to try and figure out why. And he's saying that this one isn't performing very well. So why isn't it performing very well? What metrics did you use to calculate this?
Watson: Here are the metrics used.
Jeremy: So exactly what you'd expect to see. In 2017 there's about 525 different marketing metrics you can measure. He's shown that these are the metrics he's using, which is exactly what an email marketer would expect to see. Click through rate is okay. Our open rate is terrible, should be 5%, this is 0.5%. Like I said, this is massive data set that we're trying to play with. This is live.
This is available today, 20 customers at one set. It's not all smoking mirrors, sketchy stuff to try and wow you. This is really bad ass, isn't it? This should be like a round of applause after every interaction. It's freaking awesome. So there's something wrong with it but we need to figure out is this something that's going to fix itself?
Like it's an anomaly that's just a spiker or is it something that's going to trend down and going to cause us a huge problem? So we can just ask him. How is it trending?
Watson: The performance rank of this campaign is trending down.
Jeremy: Okay, so it's not going to fix itself. We need to do something with it. I've got a feeling it's probably the subject line. Now going forward, Watson could actually help to construct the subject line itself. We could get into that later on as well, but this has been a manual campaign. And we've got to try and figure out what is it that's gone wrong?
Now we know that people made decisions with their hearts, right? They look at a subject line in half a second, they decide in one and a half a second whether or not to click on it. I think that the subject line probably just isn't emotional enough. It's probably too transactional, which is a lot of emails are. And it's all kind of this very discount-driven thing. So let's have a look. Watson, what is the subject line for this campaign?
Watson: The subject line is, "Don't let your health skip a beat, get Sleek Fit Three."
Jeremy: And what is the sentiment of the subject line for this campaign?
Watson: This subject line has a negative sentiment score.
Jeremy: Sixty-two and a half million emails, sentiments analytics on every one minus 60% sentiments. Something is clearly wrong, we need to do something with it. So we need to try and fix it. This is going to take about another three minutes and then I'll leave you alone with your heads exploding. Minus 60%, well, how do we fix that? Okay, take the emails for the past year with the highest open rate and analyse the subject line sentiment.
Watson: Here are the sentiment scores for the top three mailings with the highest open rates and the bottom three mailings with the lowest open rates.
Jeremy: So that's all of our best campaigns. We can see there, we've got 88% positive sentiments for "The future is now with Sleek Fit Two." We can see all the worst campaigns there. We can scroll all the way down, 62 million records. So now we can fix it pretty quick, even it's just based upon knowing the one that works.
And our creative team can go in and craft something even better later on. But we know we need to fix it pretty quick to start with. Watson, what is the sentiment of "The future is amazing with Sleek Fit Three?"
Watson: The sentiment for this is positive.
Jeremy: Eighty point five percent positive. So we know for this audience, that's most likely to work. It's going to have a positive interaction, so let's look at changing that. Okay, hopefully, this is where he has to do all the hard work. Modify this campaign subject line to "The future is amazing with Sleek Fit Three."
Watson: The campaign subject line has been modified. Would you like to see this mailing template?
Jeremy: So that's pretty much done, right? We've understood what is, we've changed it from minus 60 to plus 80% positive. That's awesome. We've found the problem. This is super, super quick. I showed one of the biggest agencies we work with in the world this. And they said, "What you do in about 50 minutes sometimes takes us 30 days to fix."
And they were like, "Looking at the future, what does that free people up to go and do something even more interesting. Later on provide a meaningful customer experience instead of getting bogged down with all the automative stuff?" So this is where we'd feed it back and say, "Just go ahead and fix it." We don't even need to do that, do we? Just get Watson to do it.
Yes, of course, I would. Now, we haven't done the full demo, so I haven't shown you the whole thing. But we've seen the subject line has been changed. What you will have seen for 100 million customers, every single person has a completely unique experience. They have a different product with a different colour and a different colour and a different image.
Because Watson has automatically tagged every single one. We changed the subject line. We've got the right ones that we know have got the highest propensity for people to click on. And obviously, that feeds back into the analytics, etc., etc. It's pretty bad ass, isn't it? So you guys can get access to this. This is all completely open source.
One of the things I love about IBM, we just give people cool stuff to mess with. Go and play with it. Go at ibm.com/Watson. Give it to your developers. All of those APIs are available because Watson is basically just a big platform full of APIs that you plug stuff into and just mess with it. Figure out how to get Alexa's style experiences within whatever app mobile responsive site, customer service, voice to text to figure out all the things that are going wrong in real time. Pretty awesome.
So let me leave you with a thought, I mentioned about how we should believe. What the people believe. Where are the people who believe what you believe? This is about people. Technology is freaking awesome and I love this shit. I mean, we spend a lot of time talking about it today but we can never forget technology is nothing.
What is important is we have faith in people that are basically good and they're smart. And, hopefully, if we give them the right tools, like Watson, our agencies, like Jellyfish, they'll do something wonderful with them. Amen? Thanks, everybody.