Jay Henderson is the Senior Vice President of Product Management at Acoustic (formerly Watson Customer Engagement – purchased from IBM by Centerbridge Partners, and rebranded in 2019 as Acoustic). He talks about how machine learning algorithms should be seen as working together with marketers. Offering options and solutions for marketers to assess and consider, rather than being seen as a distrustful ‘black box’ of solutions running rampant by themselves.
You can listen to the podcast here:
Follow Managing Marketing on Soundcloud, TuneIn, Stitcher, Spotify and Apple Podcast.
Transcription:
Anton:
Welcome to Managing Marketing, a weekly podcast where we sit down and talk with marketing thought leaders and experts on the issues and topics of interest to marketers and business leaders everywhere.
I’m Anton Buchner with a special conversation on the rise of artificial intelligence and the impact it’s having on marketing. To discuss this I’m sitting down today with Jay Henderson. Jay is the senior vice-president of product management for Acoustic. Welcome, Jay.
Jay:
Thanks, I’m really excited to be here.
Anton:
You’ve just flown in so you’ve got over your jetlag?
Jay:
Yeah. I got here a couple of days ago.
Anton:
And we’re just pre-empting the official market launch of Acoustic.
Jay:
Yeah. We’re doing a whole series of events in all our different key markets. We’re here today in Sydney to launch the Acoustic brand and the company into the Australian market.
Anton:
Some of you listening might be going ‘acoustic’, what are we talking about? You might like to touch on the heritage, because Watson is a name people will know from IBM. Maybe give us a quick snapshot of what happened out of Watson.
Jay:
Earlier this year IBM decided to sell its IBM Watson marketing business. And we wound up spinning it out into a new company. Really the whole idea behind it was that the business would really benefit from some additional investment and focus. And the best way to achieve both those things was to move outside of IBM.
We formed this new company, chose the name ‘acoustic’ and are in the process of moving outside the IBM Corporation.
Anton:
I probably shouldn’t mention Watson all the time but summarise it a little bit just to get a sense. The general marketer would know the name from winning in chess and Jeopardy but as I’ve learnt more about the Watson brand—maybe you could give a snapshot of Watson customer engagement and how that business unit fitted in.
Jay:
Watson is several different things. It is a set of technology that was used to win in Jeopardy. It is the application of that technology into publicly available APIs that you can develop against. And a whole series of businesses within IBM that were looking at adopting the Watson technology and taking it to market to solve different problems.
There was a whole group called Watson Customer Engagement within IBM. And the idea there was to take the Watson technology and use it to solve problems around customer engagement. The marketing piece of that is the piece we’re bringing over to Acoustic.
Coming with us is all of the proprietary technology that we developed around AI to support marketers. There are a couple of places where we use native Watson APIs but most of the AI inside our product has been created specifically to help solve the problems of marketers.
Anton:
Great. I think everyone listening would have different viewpoints on what AI is and especially AI for marketing. We’ve heard lots of discussion around is it an intelligent algorithm, a brain? What’s your perspective on what AI is being built for, especially for marketers?
Jay:
First of all, I don’t get super hung-up around the definition. In a lot of cases it’s where you’re using advanced algorithms that help solve problems for the business. We do take a particular approach to AI that is maybe a little different. If you look at AI technology, a lot of the core machine learning algorithms have been around for a very long time.
Anton:
Decades.
Jay:
Yeah, 15, 20 years. I used to work at a data mining company called SPSS, ironically now owned by IBM and 15 years ago we had machine learning. So when I think about the excitement you see in the market, where we’re going to hit this inflexion point and really drive a lot of adoption is taking that technology and putting it in the hands of a much more casual end-user.
The goal isn’t for a marketer to have to go back to school and get a PhD in data mining and statistics or to hire a team of 100 data scientists. We’re trying to infuse the AI into the application so that it fits seamlessly into the marketer’s workflow so they’re using these really advanced machine learning algorithms but it’s so natural they don’t even realise it. It’s just helping them solve the problem.
The algorithms and AI are helping the marketer but they’re working together, almost like having a new co-worker as opposed to machine learning is going to replace the marketer.
Anton:
I like that definition. I think the idea of infusing intelligence into technology, we have so many technologies. The martech stack, Scott Brinker says is 2,500.
Jay:
Over 7,000.
Anton:
There are so many different types of technologies but I love your point about infusing intelligence or machine learning into different technologies that may help marketers rather than replace marketers.
Jay:
Let me tell you a simple little story about one of the AI capabilities we have that accomplishes that. Think about what a marketer does. They work on their campaign, spend all this time on it, get it approved, creative looks great, hit send. And then what do they do?
We’ll, they’re super stressed out, did it work, did it go out, are people opening or responding? They’re sitting there clicking refresh on their reports. That’s the old way. One of the ways we infuse AI to help make that a better experience is have anomaly detection in the key metrics for your campaigns. So for the things that marketers care about; open rates, response rates, opt out rates—what we’ll do for that business is calculate the normal ranges for those metrics and then if we see giant spikes or dips we’ll alert the marketer.
So, now they’re not chained to their desk, hitting refresh. They can actually go to lunch, get a coffee. There are lots of great examples like that where you’re just making their lives much better or easier.
Anton:
Are we talking more around the comms angle of marketing, at the communications end, whatever the communications channel might be?
Jay:
Yeah. The portfolio of technology we’ve got can help in a lot of different areas. The campaign automation is a big part of our business and definitely a really strong area for us. But we’ve also got a whole suite of analytics to help you understand the different customer interactions. We’ve got technology to help with content management, personalisation as well as to help connect into the other marketing platforms that you’re using.
You mentioned Scott’s 7,000 different vendors, if you actually look at what’s been happening, clearly there has been a lot of consolidation in the marketing technology space. You’ve got Adobe, Oracle, Salesforce buying stuff, us, Acoustic but at the same time you’ve got lots of little interesting companies.
You’ve got one of two vendors from mobile, 3 or 4 for social, a dozen for advertising. And the consolidation combined with that explosion of vendors has created a mess for most marketers. We’ve been focusing a lot on being an open marketing platform in contrast to the walled gardens that are trying to keep everyone just in their application.
So, we’ve been doing a lot of work around something called acoustic exchange, which helps get data in and out of our portfolio and connect the ecosystem of vendors. It’s kind of the connective tissue for the marketing platform.
Anton:
I like that word mess. It is a mess. We hear that a lot when we talk to marketers. It’s confusing. Marketing has become so confusing. The CMO has had a digital division, or social or content division, a whole bunch of silos that have generally invested in their own technologies. So, you’ve ended up with this mess, some of which is working and some of which isn’t.
Maybe before we dive too much into the specifics, how would marketers approach this? And let’s assume we’re not a clean ecommerce start-up that’s just started, but we’re a legacy business that has challenges with silos and technologies as a mess. Because I don’t know who I trust.
Jay:
As you assess as a marketer what you can accomplish as an organisation, you’ll find different things. There are some organisations that are going to be able to buy a full suite from a platform vendor, and that’s going to be great for them. More likely though, what the marketers need to do is assess which areas of their marketing technology stack are working for them, and which ones would be improved by better capabilities, better integration, prioritising them, and then starting to tackle them.
For us when we talk about the need to integrate that’s part of what’s really powerful about that. If you’ve got Adobe or Google Analytics and that’s working for you, that’s great. We know how to ingest data from those applications into our campaign management.
If you don’t have one of those or they’re not working for you, that’s great, we’ve got our own analytics solution and I can spend a lot of time telling you why that’s so great. Marketers may not have budget to do everything at once. They may not have span of control over all those systems so they need to collaborate with their peers.
From a change management perspective they may not be able to change everything they want. So a lot of the time it’s really about understanding what’s working and what’s not and where the opportunity is to change and improve and to eventually drive the business impact, which is what everybody really cares about.
Anton:
Are you seeing the change coming out of the CMO, or the IT department, or is the C-Suite driving these discussions?
Jay:
It’s a really great question. I think the changes are coming from the CMO. Being a CMO is a tough job; there are a lot of demands. The change is often rooted in the desire to improve things. But you poked at an interesting question which is – it the CMO or the IT?
We’re really seeing a huge influx of marketing technologists and sometimes they report to IT and sometimes to the CMO but really someone who is responsible for and owns that marketing technology stack and helping drive the right decisions. It’s not quite all an IT decision or all a marketing decision so these blended roles are really starting to take hold.
Anton:
We keep reading headlines like the CMO is wrestling budget back from IT and so it rests with the CMO but if they’ve got control of the budget what we’re seeing is, they don’t necessarily have the capability or knowledge about a) technology and technology solutions, and b) artificial intelligence when you’re talking about infusing it.
So, it’s a challenge; they’ve got the money but how do they spend it if they don’t really understand?
Jay:
There was this initial wave of investment in marketing technology that largely went to cloud providers partly because it makes sense but also because those organisations could go around their IT. We’re seeing the problems with that and starting to bring IT back into the conversation.
Really, to be successful and create great experiences the customers are going to love, you need to bring IT into the conversation to make sure the systems are going to do what they need to do over time.
Anton:
But ultimately someone has to make a decision so, I get that we need to have technologists, marketers, and vendors or partners together.
Jay:
The CMO is in the driver’s seat there. They’re the steward of the brand. They’re the ones who are responsible for creating customer interactions that people love and deliver on that brand promise. It’s definitely the CMO.
Anton:
So, it’s on the CMO’s shoulders. Let’s put ourselves in those shoes. I’m a CMO. It’s a bit messy. I’ve integrated internally with my technology team etc, and I’m looking to evolve the role of technology and I’m exploring AI. Before we jump into your products, give us a sense of how do we go about it, as the CMO?
Jay:
There are a few different ways to approach the problem. Often the first decision a CMO is making is how are they going to approach it. I think there was a lot of investment initially around hiring teams of data scientists to just focus on marketing. If you have the ability to invest that way it can certainly bear a lot of fruit, be very profitable and help improve effectiveness and efficiency but it’s a very expensive and hard way to do it.
First of all a lot of the data scientists are awesome with algorithms and machine learning, but they don’t really understand marketing so there is a little bit of a gap there. And in order to really scale great experiences you have to bring together human connections with the technology. That gets hard to do if you just try to hire a team of 100 data scientists.
It’s often a blend of getting the right skills for the team, investing in some data science resources and then how do I complement that with technology and capabilities that have been baked into the platforms I’m using?
Anton:
There’s probably an issue around control. We know some marketers or businesses don’t like to let their data out to 3rd parties. And of course the 3rd parties would love to get hold of the data and do better analytics and mining. Certainly, in Australia we’ve seen issues with control.
Jay:
There’s a whole Pandora’s Box around privacy that has been opened. In Europe you’ve had GDPR; that’s rushed in a huge wave of copycat legislation. So, in California you’ve got one that’s about to go live, in Brazil, and here what’s been happening with the Royal Commission. There is some stuff that marketers need to get organised about.
Having the right mentality about privacy, thinking through about how you be transparent about what you’re collecting, being responsible about how you use it; those are all really important concepts for marketers. The privacy legislation will drive the need for some change, for marketers to be more transparent but also to think through what they’re doing a little better.
I was talking with a big Australian bank yesterday and we were talking about the potential for regulation and one of the areas we were talking about was how do they create consent to communicate with customers. Right now, for each channel it’s binary; yes, the bank can contact you or no, they can’t.
We talked a lot about how do they opt down? Either allowing customers to choose certain communications, like only talk to me about these topics or choosing less frequent communications. So, trying to move from this binary in or out to something a little more subtle that‘s not going to be problematic as you have to comply with potential legislation.
Anton:
Which certainly takes a more human approach; humans (customers) interacting, who don’t necessarily want to receive all information all the time, or stop it for a couple of months, weeks or days.
So, data is the theme we’re picking up on. Technology, whether it’s an artificial intelligence learning algorithm, it’s only as good as the data coming in. What are you seeing around either enterprise-wide approaches to data versus siloed customer experience improvements? But it’s really only as good as the data that’s going into those machines isn’t it?
Jay:
You’re absolutely right; garbage in, garbage out. It’s not always about having all of the data. There was a wave of let’s create a 360 degree view of the customer, put everything in a data warehouse. It turns out that data warehouses are where data goes to die.
Anton:
The great cul de sac of data.
Jay:
I think marketers are trying to sort out how to get the right data, how do I know who the customer is, how do I understand them in the context of right now? What other pieces of data do I need to know to inform my decision about them right now? So, there is a subtle shift into looking for the right context for the customer rather than every single last piece of data.
The other thing I talk to customers a lot about is don’t be afraid to ask customers for some piece of data that you think would really make a dramatic difference in how you serve them. A simple example I love is we were working with a large retail outlet that sells musical instruments. They had this idea, for people who play guitar it’s very polarising; you either love Gibson or Fender.
So, they decided to have a little survey that ran on the website; which do you love? What did they do with the information? They used it to drive the personalisation on every other marketing interaction. On the website, in emails you got a Gibson themed message. It dramatically changed their response rates and conversions and it was really just based on the simple idea of asking the customer this question. It achieved some great results.
A lot of times we get caught up on I’m going to run this data-mining algorithm; segment you into these very sophisticated things. Sometimes you can just do something really simple like ask a question.
Anton:
Getting back to logic. I often refer to my mother who has been a 23 or 25-year subscriber to the theatre—stopped for one year last year. She’s been a loyal customer, stops for one year and gets ‘welcome back dear new customer’. She’s thinking ‘what a bunch of idiots’.
They didn’t understand or use the data; they just put her into a bucket which was new subscribers. We can often get caught up and lose sight of logic and getting the basics right.
Let’s leap into the AI aspect of all of this and assume the data is not bad, maybe some of the data is not bad. What are you seeing around the machine learning aspect sitting behind your products and how that’s helping marketers?
Jay:
When we all embarked on the digital transformation we thought digital is going to be great because we’re going to be able to measure everything. And that turned out to be a problem because it turned marketers into hoarders of data.
We talked about what is the right data and the exciting things that are happening in AI are helping surface insights out of that data in ways that don’t require you to hire data scientists or in ways that are much more automated.
We’re doing some really interesting things around our analytics business. If you’re doing website analytics we can automatically identify when people are struggling on the website. It’s usually during a conversion activity; trying to sign up or purchase and something’s not quite working right so I hit the button 10 times.
So, instead of having an analyst hunt for those things or have to read a report, the system is in the background automatically looking for them and then when it finds one it alerts them.
Then the analyst can give feedback; ‘that was great’ or ‘no, that was normal’. Just being able to automate the creation of some of those insights is really helping marketing departments scale the insights about the customers.
Anton:
So, it’s greater efficiency.
Jay:
Yeah, there’s effectiveness stuff happening too with AI, classic business stuff, we can either improve efficiency or improve effectiveness. The efficiency stuff, there’s a ton of productivity things. On the effectiveness, I can spend more money and generate the same results or spend less money and drive better results. The power of AI can really influence both.
Anton:
Are you seeing it as adding stronger value to the marketing team and the CMO? We can measure everything and everyone fell into that cul de sac of measuring everything but measuring nothing really at the end of the day. Is it giving power back to the marketer to then decide which are the key metrics?
Jay:
The way I think of it is letting the marketers be marketers again. They don’t have to be analysts or data scientists and that’s important. The human aspect is why a lot of marketers get into marketing. As a software vendor, software needs to be helping marketers do what they’re great at.
It’s not that technology is not important but it needs to be an enabler to that creative process that the marketer is going through. It’s a really interesting time because there has been a lot of buzz or even hype about AI. And I think there is a little bit of fatigue around the whole idea of AI but to me that’s important because it’s a signal that we’re on the inflexion point here of it adding value and solving problems.
I think some of our competitors used AI to grab headlines. We’re using it to help solve real problems.
Anton:
That’s important. You’ve got to get back to marketing is helping drive business so if it’s not delivering value, the basics of efficiency in the operations and getting stuff out and effectiveness is in delivering value back to the business; sales, retention, profit growth etc.
I think it’s great that it’s giving more of a focus on the business metrics, which is important for marketers. What are some of the other pitfalls that you’re hearing because when we look at the hype in the market there is a risk in all this of going, ‘it was too overhyped’?
It is very complex because you need fairly good or clean data. You need to prioritise what that might be. You need to look at systems and what systems are being integrated and what ecosystem it’s within. And then you’re obviously testing something and looking at the analytics and the machine might be telling you which anomalies and results to look at but what other stumbling blocks you’re seeing in approaching AI?
Jay:
The reason there is so much hype about AI is because the potential is so great. People see the potential which is why it’s creating excitement. When I talk to customers about how to adopt AI and how to get the most out of it, a lot of it is not terribly complex. Pick some place to start, show value, move on to the next place.
Anton:
Crawl, walk, run.
Jay:
Demonstrating value around an AI project is not hard. It’s often identifying a meaningful problem to solve, that’s going to produce good ROI, and then promoting that internally, and then moving on to the next one. This idea of experimenting is really important.
I was talking with a Telco customer yesterday in Australia. They were saying it used to be working on a digital transformation project and then that ended and we realised we still needed to change more. People are moving towards agile marketing and taking the ideas behind agile and applying them to marketing.
Particularly for these AI projects it’s a great idea or model to identify these short, high impact things that you do. Prove out the use cases and then either kill them off because they didn’t work or you promote them into your production machine and then move on to the next one.
I think this idea of being more agile and technology allowing you to be agile is really powerful.
Anton:
So, taking more of a ‘short-termism’ approach maybe? Because most major tech transformation projects get bogged down 6, 8, 12 months later they’re still going.
Jay:
It’s almost like short-term wins that help you get towards the long-term goal.
Anton:
With a clear goal obviously in mind.
Jay:
Yeah.
Anton:
Can you share any other experiences, on the positive side. Let’s not dwell on all the negativity. What are you seeing?
Jay:
Most marketers want the same thing: they want to create great experiences that their customers love. And that’s a powerful thought. Marketers have great intentions. Then you think about all of the terrible customer experiences that are out there. There’s a big gap between what marketers want to do and what’s actually happening.
To me, some of the most exciting stuff in our market right now is where technology is helping marketers close that gap and bring closer together what marketers have in their heads and want to deliver as an experience with what they actually do. And AI is the catalyst for that.
Marketing is great at the human connection. Technology is going to be the thing that lets them scale that up and create those personalised experiences. I think AI is going to change marketing more in the next 5 years than the change we’ve seen in the last 25. We’re on the precipice of such a massive transformation of marketing. And there are lots of examples of marketers who are adopting technology in the right way or driving differentiation.
There is huge opportunity for great experiences to differentiate brand and drive results that people want to see.
Anton:
Which is back to the marketing objective isn’t it? I’ve got an idea and I’m just going to float it and see if the machine can solve it. In my simple brain I’m thinking anytime somebody clicks to a website for example, can that site intelligently identify me and only let me access products that are relevant to my profile?
Jay:
There are some things you can do. There are ways to understand your location and that can help influence what you see. We can also progressively build the profile of who you are as you’re clicking around. So, as you’re expressing interest in some things or typing terms into the search box, all of those things will help drive the personalised experience.
There are other things you can do, buy 3rd party data that would help tell you more about that person in an anonymous and privacy sensitive way. So, there’s lots of opportunity, even if someone brand new shows up to say they look like those people who have come before who responded well to this message; let’s personalise the experience in that way.
Anton:
If you look at a premium product that’s not for everybody, luxury products for example—could the intelligent brain do some credit risk and financial checks to say, ‘o.k. this product is at least affordable (in a positive social way) but then based on like.
So, I like this style of brand, product or thing, therefore it’s only available for this sort of person. So Jay, as you click on this and come to the site it will serve up only this product for you because all other ones aren’t relevant to you and only you could access it. You can access the next level of information because you are the right person.
Jay:
Yeah, and when you think about the different attributes marketers have historically used to target, it’s some pretty basic things; age, income, gender.
Anton:
Classic demographics.
Jay:
Yeah, so, there are these new areas that can be really powerful and help drive results that marketers are starting to experiment with to figure out how that would influence. You’ve got to overlay that with the regulatory environment. In financial institutions there are fairness in lending laws and things like that that change the scope or how you solve the problem.
Anton:
But the machine would know all that. Whatever industry we could feed in all the regulatory compliance data. But in some discussions I’ve had marketers who are hoping to take personalisation to another level. We’ve heard about personalisation; lists, CRM for the 90s and direct marketing.
Jay:
Let me give you an example. One of the ways we’re driving personalisation into the platform. Using machine learning to drive personalisation—that’s been around for a long time. For one of the personalisation products we sell, it’s the number one reason people say they buy it.
Now, if you look at how they use the product, it’s also one of the least used features we have. We went through a whole design thinking exercise where we interviewed them and asked them why they weren’t using it and it boiled down to ‘it’s kind of scary, and if it doesn’t work I get fired’.
It’s almost a little bit of a control thing. They don’t want to turn over control to the machines. So that resulted in us designing some different capabilities. We still have the machine learning; it’s going to experiment and run tests and identify men between 18 and 24 or suburbs of Sydney who respond to this type of content but then instead of saying now that I know this let the machine decide what to personalise instead.
It will translate what it learned to a rule, present the rule to the marketer and ask ‘did you know, men, 18 to 24 in these Sydney suburbs are responding to this content. Would you like me to start serving them this content’?
It lets the marketer compare it to their intuition about what should or shouldn’t make sense for the business and it gives them the control to say yes or no. By working with the marketer and putting it into their workflow and giving them control we’re driving much better adoption of this fancy technology.
Anton:
We’re not threatening the marketing team or leader, we’re working with them. I like that thought. It’s almost augmented marketing or relationship. That’s fantastic. So going forward, the next few years, you talked about being on the precipice of big leaps, what are you seeing coming down the line in terms of next stages?
Jay:
I think there are a couple of big things happening. For me, one of the most exciting trends is starting to bring together the marketing technology and advertising stacks, and again it comes back to business results. Today the onsite or on-brand experience is very different to what’s happening in advertising.
We’re doing some very interesting things that are very simple to connect the marketing technology ecosystem into advertising. My favourite example is a bank here in Australia used one of our marketing technologies to pull a list of customers and what products they owned.
They activated that into their demand side platform and used it as a suppression list. So, basically they used it to stop marketing products to customers they already owned. We’ve talked about some science-fiction things with AI but that’s basic blocking and tackle. But it turns out they were spending about 20% of their media budget marketing stuff that customers already owned.
So, there’s huge opportunity on the advertising side to drive better ROIs as you bring these things together in addition to creating a much more integrated customer experience. It’s awesome that it can deliver the business results that folks need. As I look over the next couple of years I think you’ll see massive coordination across those two big platforms.
The other big trend I’m excited about is the interesting things happening in emerging channels. What’s really cool about the APAC region and Australia and New Zealand is that this region is very much on the forefront of what’s happening with group messaging platforms like Lime, WeChat, WhatsApp and driving some very innovative ways of brands interacting with people through those channels that are on the forefront globally.
The US is behind on that front and it’s one of the very exciting things that’s happening in this region.
Anton:
The risk with that is, as users, we like the closed garden. As soon as advertising gets bombarded into it, we look at it as an invasion of our privacy but you’re saying if we can make it a better experience, there’s positivity.
Jay:
Yeah and there are some places where it’s a very efficient channel and way to communicate with a brand. I agree there is danger and opportunity. Like any big thing that might happen you’re trying to balance the tension there.
Anton:
You’re American; we’ve got to be positive. I was having breakfast with two Californians and I love the positivity despite everything else we’re not going to talk about. Jay, thanks for having a chat and lovely to meet you. Good luck with the official launch.
Jay:
Thanks
Anton:
I’ve run out of time, but I’ve got one last question. As you work together with your robot on your side, what are you going to name it?
Ideal for marketers, advertisers, media and commercial communications professionals, Managing Marketing is a podcast hosted by Darren Woolley and special guests. Find all the episodes here
