Global Marketing
Management Consultants
Global Marketing
Management Consultants
mobile-logo
Global Marketing
Management Consultants
Top

Managing Marketing: The Measurable Marketing Applications Of A.I. (Artificial Intelligence)

John James

Managing Marketing is a weekly podcast hosted by TrinityP3. Each one is a conversation with a marketing thought-leader, professional, practitioner and experts on the issues and topics of interest to marketers and business leaders everywhere. In this special series, TrinityP3’s Anton Buchner, discusses the rise of Artificial Intelligence and the impact it is having on marketing.

John James is an independent strategic consultant for CxOs and Founders, focussed on growth and revenue. He talks about real life, measurable applications of artificial intelligence. Including a tool that he’s building for CMOs to help reduce complexity and improve the operational efficiency and effectiveness of marketing throughout the end to end process of the marketing funnel. 

You can listen to the podcast here:

Follow Managing Marketing on Soundcloud or iTunes

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 today in Sydney I’m sitting down with John James. Welcome, John.

John:

Thank you very much, glad to be here.

Anton:

I’m really excited to have a chat with you. You’re a growth strategist, growth consultant, which means you’re focused on business. You’re a director and I see you’ve been a digital strategist, growth manager and board advisor. So you come with a really interesting background. I’m fascinated to hear what you’re seeing around artificial intelligence today. How does that sound?

John:

That sounds great, let’s get right into it.

Anton:

Delve in. I’d love to start off by getting a bit of background for the listeners. What’s been your trail to today? And what sparked your interest in AI?

John:

I first started at university; I was in an accelerated cohort in Queensland. They had a really good programme there, very practical minded, marketing angle with their education. As part of that we had to intern with two companies as I was studying, mostly at a marketing assistant level, but you’ve got to start somewhere, right?

And in between that I was doing some work at Colmar Brunton—a field work researcher so I got to see—on the boots—I was interviewing people and tabulating data into SPSS in the good old days.

Anton:

You were a data guy.

John:

Yeah, a bit of both—a good cross-section of experience at university. I had a bit of a break in Canada—I lived there for 2 years, worked in some hotels and did the whole gap year thing and that was great.

Then I came back, did some advertising agency work for McCanns, bit at Saatchi. That was a good exposure to big brand work. Then I was in Melbourne working down there for some digital agencies. That was when search was quite new and Facebook was just taking off back in 2014 / 15.

A really incredible yield you could get and the knowledge really wasn’t there so a huge growth market. Obviously since then it’s died down a bit and the yield’s decreased. Then I went to America, was working for a start-up in Silicon Valley and doing growth.

Specifically, it’s a bit of a wide term but it was generating more demand for a marketplace on the demand and supply side and getting both of those parties to transact. So, it was a revenue-focused position there.

Then I came back here and had an idea for a tech application combining all that knowledge together, and I’ve been working on that for the last year and a half while doing some consulting and agency work on the side. That’s probably how I found out about you.

Anton:

It’s interesting we found out about each other on LinkedIn speaking about technology. I put a call out to interview different people on artificial intelligence and your name popped up. You’re based out of Western Australia.

John:

No, actually I put that as a joke, just to see who reads the bottom of my profile. I fly between Sydney and Melbourne. I lived in Melbourne for 8 years, but I’m temporarily in Brisbane and flying down to Sydney and Melbourne every week.

Anton:

So, we won’t relocate this podcast to WA.

John:

No, definitely not.

Anton:

So, you replied saying that you’ve got a huge interest in it with your tech application. Before we jump into that, what do you mean by the words artificial intelligence from your world, and maybe marketers that you interact with?

John:

From my perspective, at senior management level, it’s a bit of a plaything, a buzz word term. But I have been exposed, especially in America and through my business partner who is working for an AI company in Silicon Valley to what I would call true AI, which is what I would describe (if you want to use another buzz word) as evolutionary computational algorithms, which are algorithms which learn and adapt and improve over time without human intervention.

Anton:

ECA—have we got another acronym into the marketing bible?

John:

Well that’s where the true AI is and you’ll probably here another word—neural networks, creating an artificial brain. For me, AI is something that learns and gets better over time without too much human intervention.

Anton:

And what are you seeing in that sense then, from a tech or strategic angle when you say you’re getting into a tech application? Give us a sense of what you’re doing.

John:

Well I suppose I was first exposed to AI if you go back to Deep Mind, a company out of the UK, which was started by a neuroscientist and data scientist. He had a bit of both; the brain work and the other work. And they sold it to Google.

Anton:

Did they do AlphaGo?

John:

Yeah, that’s a subsidiary firm. Technically, it’s still separate in Google’s Alphabet Group. At the same time I was exposed on the frontline to search optimisation. I think that’s probably the most practical area where I’ve seen what they call RankBrain, which is an AI-driven module, which runs how websites appear on search queries on Google. I’ve seen that over time, they sacked the whole SEO office in Google and replaced them with this automated solution.

A couple of months ago it had a bit of an issue and they had to roll things back manually so it’s not perfect. But that’s where you would use things every day without knowing that you’re interacting with a product that’s driven by AI specifically. And people use Google every single day.

Anton:

I think that’s a really good point. We hear in marketing that it’s the new thing, the next exciting thing, that’s been happening for the last couple of years, after Mobile First, which we’ve heard about for years. But actually we’ve been using AI solutions for years haven’t we?

John:

Decades. They had their issues back in the day but they’re very much automated self-aware systems now. I’m hearing rumours (from friends) of this AI actually creating its own coding language that the engineers who actually created the AI can’t understand—that’s a bit freaky because this thing is self-aware and writing its own code.

Anton:

That goes into a big cul de sac doesn’t it? When we go if I can’t understand the code, what on earth has been created and as humans we can’t control that or we can’t understand it—that’s a challenge with trust.

How do I trust this computer that’s generating things that I don’t even comprehend or know? Do we need to know or not?

John:

Well, it’s a good point. People fling around the conspiracy theory that AI is going to take over the world—possible. But when I hear stories like that and they have been validated as true it’s pretty scary.

Anton:

We’ve heard with autonomous driving plenty of examples where we’ve been told people can take over the car remotely and we’re going to see all these cars crashing into kids and all sorts of dilemmas. But there’s always a doomsdayer, but let’s keep it positive.

We’re looking at technology—tell us more about that.

John:

In terms of applications a lot of debate around this area tends to be towards two areas in our sector: marketing or growth. It tends to be very AdTech driven or it’s marketing tech to do with the sales side.

I see some really good applications in sales prospecting and closing in automation within that whole sequence especially in B2B software sales. There are some really good applications there, which I use personally.

You can almost sack your entire sales force bar some human interaction when it’s needed. So all the ground work is done beforehand and the follow-up—with that you really need to get involved. And that does reduce headcount. It can contribute to a lot of savings.

The other area I see it is in AdTech optimisation, things like copy images, video. This AB spilt testing gets thrown around a lot and automating that process. I can give you 2 examples.

Evolve AI in San Francisco, the company for which my business partner worked. You may have used Optimise—the split testing learning systems with landing pages in the digital area. They have created an AI and if there is enough interaction on this page in terms of conversion events they’ll split test all the changes you’re going to make concurrently at the same time.

So, if you run an AB test, swap out a variable, do it again and again and again. You can put all of those variables in at the start—it’s called multivariate testing. And over time, once you’ve set it, it will find the right combination of those variables that has the best result and apply that directly to your site.

Anton:

Do you think that is intelligent or is it going back to what we learnt in direct marketing; test all sorts of attributes, certainly it was a lot slower. Whether it was an AB testing or ABC – whatever you wanted to do. But at the end of the day it’s multivariate testing to find the right combinations that may work for some people or some segments and other segments act differently.

John:

I’ve heard personally that they hire a lot of neural network experts within that company. I can’t comment because I haven’t looked at the tech. I would say it’s one of the better examples I’ve seen, but is it just an advanced algorithm with some really interesting parameters, or is it self-aware AI?

They do mention specifically the acronym we mentioned before, so I could argue that it’s more on the AI side of algorithms.

Anton:

And certainly for marketers where it’s become harder and harder to work out how many tests they should do as they can’t keep spending money on doing more and more, it’s probably a quicker way to get the best versions or solutions.

John:

It’s very fast so you have that time advantage. You also have the elimination of human bias which comes into all the split testing. A lot of people don’t even know the difference between 95% and 99% interval confidence when they’re doing split testing. There’s a huge difference between that 4% range.

Anton:

I had a big debate the other day with someone about confidence levels.

John:

Exactly, so if you want to get geekier; I’ve seen first-hand the start-ups, people using Neil Patel’s confidence interval calculator and it’s completely false. Or they’re measuring the wrong parameters or interpreting it incorrectly. So, you’re removing that human bias from that situation as well as reducing the time.

Anton:

With this example we’re probably still down the bottom of the funnel; it’s at the end conversion point, would that be right?

John:

Yes and this is why I want to talk to you today, because I take it not just at ad-optimisation level which is a click or interactional view. I take it all the way down to a revenue outcome, which is a whole other kettle of fish.

And once you start optimising from revenue back up towards your top of funnel marketing activities, you’ll notice a lot of it is completely wasted. Obviously, we’ve got to be mindful here of the long-term versus short-term, brand versus sales activation, but it’s a completely different way of looking at business. And that’s what I’ve been working on for the last year and a half.

Anton:

That’s great. I see too many marketers focussed (marketer, media agency, digital agency) just talking down that bottom end saying ‘I’ve got an x% increase in conversion rate and look how much we’ve generated in sales’. That’s end of the funnel activity—was it going to happen anyway?

John: Yeah, what’s the baseline? What are the macro and big factors at play here? Has there been a competitor who has entered or exited the market? Half the time I ask them this question and they have no idea.

Anton:

So, maybe give us a sense of how you join up as you go back up the funnel. What are you looking at and how are you approaching it—without giving away too much of your secret sauce.

John:

What I’m looking at—you need to connect a lot of disparate systems together which is why this is very technically complex and again I’m mindful of generalising again because I work in a lot of different sectors.

I’ll give you an example—plastic surgery for certain augmentations to the chest area. The sales lag time on that can be 12 to 18 months. So that first interaction all the way through to a purchase can take that long.

So, if you’re taking cross-sectional data of a 3-month campaign you’re not really seeing that end result. I think the first point is you need to take a very long-term view of this and bring in a lot of variables like brand, different channels, and this is where this attribution complexity comes into play.

Everyone talks about MTA or multi-touch attribution but I think you’re going to find better models and there are a lot of different ways you can configure that. So, I use an unified approach; I consider the length of the sale and then post-purchase behaviour if there’s another sale over a very long period of time.

Anton:

How are you identifying that lag, when you say 12 to 18 months? Are you using some panel or survey data of current customers?

John:

In that respect this is a service sector application. For the product we would take CRM data that comes in via phone or digitally or whatever. And then we would look at that at a later point in time and track where it originated and all the interactions with that person over a period of time.

Anton:

So, the soundbite out of that is identify the relevant data.

John:

Yeah and that’s a big pitfall. You need to know a lot about a lot of different systems. When you look at channels you need to be a channel expert on every single channel and I’m not just talking top level impressions or reach. You really need to know which are the really effective mediums that are pushing a revenue outcome further down the train.

And it can actually be very boring channels. Everyone wants to create a 15 or 30 second video or have their ad plastered on a billboard for example, but sometimes it can just be a follow-up phone call from a sales rep after they’ve inquired or walked past a sign.

Anton:

Or radio as well. To digress slightly, I’m doing a course on northern beaches community radio so the power of listening to the radio. I’m not sure whether anyone listens to northern beaches radio, but we are auditory so going back to some neuro science we listen to things and take it in.

A podcast; we can’t see you, you can’t see what’s around us (anyone who’s listening to this) but we’re listening. We can take in so much more information whether it’s subliminally.

John:

A good example on that. I was talking to a CMO, a close friend of mine in Melbourne, in the health sector and he had a lot of channels going. He put radio in at very specific times late at night when the ladies were at the gym and that radio put a 20% uplift in every other channel.

He took the radio out (and he had to do this—he used exclusion testing over a couple of months) and his budget got pulled so he had to pull the radio and it dropped 20%. So, we could prove a very high correlation if not a causal relationship between investment and radio being that last slice of the pie that uplifted every other result.

I take a very agnostic view of media expenditure. And media is just one way you can get a sale. There are many other types of ways.

Anton:

That’s great. Getting back to the AI side—what application are you looking at and how does it help solve marketer’s challenges all the way through the multi-channel touch issue through the funnel issue?

John:

Let’s get back to the software side of things. I know this is B2B but MadKudu is a really interesting platform out of Silicon Valley. And they use Facebook AI data and bring a lot of sets together to predict the receptibility of a lead being interested in your product before you really have do anything.

They use a lot of lookalike audiences, and a lot of that determines the quality of the sample you start with when you do a lookalike and the integrity of that matching process. They’ve got a very interesting system that basically runs on autopilot with very little intervention.

They’re doing some interesting things on that side of the market.

I’m working on (I don’t want to be alarmist here) making the agency versus in-house decision a bit redundant by facilitating operations and connecting strategy to tactical execution, measurement, and then feeding that back into a circular process of growth.

And that removes a lot of the human bias that we see, whether it’s political bias within the organisation or external vendors hiding the juicy data that you actually need.

Anton:

You’re not suggesting that agencies would ever push that boundary are you?

John:

Not at all, but I’m pragmatic when it comes to commercial reality and companies have to make profit, but ultimately the company loses out. That’s why I stick mostly in the private equity sector and work with clients who have a direct interest in the revenue outcomes for the business. Once you move away from that it’s not really about sales.

Anton:

I like that point you made about taking away human bias. We hear a lot of AI discussions around taking human’s job roles and overtaking but also augmenting with humans but taking away some human bias to get a better result is ultimately what we’re trying to do; breaking down silos, preferred choice of channel and solution is great.

John:

I have an example of that. I was on a plane last week next to a gentleman and he was configuring image recognition of cheques. So the big 4 banks in Sydney (all but one) all the cheques that get deposited go to one facility where they’re scanned. They look for discrepancies for fraud or in the amount, and then automatically that whole process is done.

That used to be a manual process with bank tellers. That’s completely gone now.

Anton:

Wow, we don’t know that do we? I got a cheque the other day which was really bloody annoying—I had to go to an ATM and put it in

John:

That is collected. It goes to a distribution centre and automatically scanned into your account, unless there’s an issue requiring manual intervention.

Anton:

So, what’s the end game here? For marketers I guess you’re talking about growth being the outcome—that’s the objective. For marketers what are you wanting to achieve?

John:

I want to make it easier for marketers to adapt to the market as quickly as possible and get the most effective outcomes that are valuable to the business itself. I’ve created a briefing system at the top level that takes my expert opinion in all these different channels and allows you to instantly brief the vendor to do a really good job without having to have meetings or knowledge of what they do.

And that’s where I see a lot of this miscommunication. As a CMO I’m not really familiar with the ins and outs of display advertising for example, but I can use an intermediary who can translate what you want into what the vendor can understand, and they can execute a really good campaign for you.

Anton:

Fascinating. I’m hearing that and thinking capability tends to be a big issue in this space. So both agencies, vendors, and on the client side, marketers that try to get capability to understand both business and artificial intelligence and, as you said, full through the funnel concepts.

This sounds like you’re trying to take away some of those big issues which can pop up, which is great. I don’t actually need to hire all the capability necessarily. As a marketer I could utilise AI to solve some of that.

John:

Correct. Or your saving that overhead of that vendor and that time spent with miscommunication, client back and forth emails. You’re saving that whole problem. So it actually makes it easier for the vendor. So, the vendor likes it. The client can brief on the run, just call up and push a button.

A new product idea at this CMO conference, they get all excited, we need to do this. This allows you to brief an expert vendor in your local area or internationally to execute that idea for you.

Anton:

So you’re getting much more efficient. On the marketing side, greater efficiency, and ultimately, greater effectiveness which is your growth model.

John:

And higher margins for the vendor so everybody wins.

Anton:

It reminds me a little bit of the ads for computers in the 80s where it was all going to be, “free your time up—sit on a beach”.

John:

It hasn’t eventuated has it?

Anton:

They failed to think it through; I need to earn an income to pay. So we still need a job. What are your thoughts around will it be a re-skilling then if we’re improving efficiency, probably reducing some headcount?

John:

Definitely, I’ve seen this already with 2 companies that do a lot of RPA (Robotic Process Automation)—basically that just means sacking a lot of low value jobs like coordinators and things like that. I would be very wary if I was in that sort of role—just sending communication between parties—there’s zero value in that. All it does is open you up to miscommunication.

Appian is a really interesting company in the US that are in that space with a low-code framework. And you hear a lot about UI path—they have automated complete backend systems as well as admin side of things. And they’re going to the front-facing consumer role as well and automating that.

I’m automating the strategic planning side of things which is again much higher up the chain of an organisation.

Anton:

So you’re doing away with 2 levels; the low-level doers, the mundane, repeatable type jobs, but you’re also potentially doing away with the strategic side.

John:

This is the biggest inhibitor to growth in companies I see—this silo mentality between departments. For growth to happen all those departments need to work together, especially product, sales, marketing and any other support services that are geared around that revenue area, data and finance for example.

We often get this friction of political point scoring. That does not work if you want a growth outcome. That’s the biggest inhibitor to growth. My platform helps to translate all those different people and get them working on the same kind of core metric. And then that metric filters down into their department.

Anton:

I like that. You’re talking about alignment which is one of the biggest issues for the biggest businesses: the siloes and lack of clear objectives that overarch all siloes. So, unified metrics, unified objectives which is most important (metrics being the KPIs to measure success).

John:

Big time. We use this OKR term that John Doerr uses and again there is a lot of bias when it comes to choosing metrics. How do measure the metric? Which metric do you measure? You need to have a lot of experience to know how that works.

Anton:

So are you talking ROI of AI?

John:

Yes. I’m working on is an autonomous business growth engine, using humans sparingly to generate financial outcomes for businesses.

Anton:

What are you hearing around long-term versus short-term? So long-term with AI it’s difficult to implement, difficult to get right, maybe it’s little steps and then progressively iterate. But the CMO and the board don’t want to take too long these days because they need short-term sales.

How are you overcoming it and what are you hearing?

John:

It’s a good point. I suppose that question is two-pronged. That interaction between sales and brand activation or short and long-term effects, and also how much of that can be automated?

It’s very complex and that’s why I don’t want to make general statements. But I built it in a way that can adapt to different sectors, products and services. I talked about sales lead times—that’s just one of the variables that you have to look at. You have to know the customer decision-making process as well.

Probably not a good answer to that question.

Anton:

It’s a trade-off. There’s probably no answer. It’s probably the old ‘which bit of my marketing is working’—50%—I’m not sure which side.

John:

I think the hardest thing with AI and automating things is this dynamic environment that everyone is exposed to. Markets are changing very rapidly these days and it’s not going to get any easier. So if you have the agility to make better planning decisions based on unbiased data, then you’re going to win in the long-term.

In terms of decision-making over time, AI will help maybe remove some of those sources of bias and make things quicker and easier. That will give you a competitive edge in the market.

Anton:

Which is ultimately what most leaders are wanting.

John:

Pretty much.

Anton:

Fascinating. Could you leave listeners with one piece of advice when looking for AI solutions?

John:

Sure. That’s hard. 99% of them are fake—that’s number one. So, if you hear of a start-up using an AI, it’s probably an open source AI platform or something and ticking off a tick box when they’re raising funding—make it sound good to investors.

I would go with tried and proven applications, and again this is very expensive. But the 2 companies I would be looking at would be Google and Facebook who are doing great things there, maybe some of the other established players in the market but there aren’t many because it’s so complex.

I’ll give you an example, Google’s Vision AI is really good or their OCR (optical character recognition)—you can tap into that Google Cloud and use that straight away. And the same with Facebook’s AI with Improving Campaigns—that’s very powerful.

Anton:

It’s back to test and learn isn’t it? Maybe this is the future, it seems to be a long-term trend, from all analysis, but get in, start small, get your head around it, test some things, don’t see it as the all-encompassing solution just yet, but maybe work towards how it can improve efficiency and ultimately effectiveness.

John:

Exactly.

Anton:

It reminds me of CRM in the 90s.

John:

Yeah, pretty much.

Anton:

We’re out of time but it has been really fascinating to talk to you today.

John:

Thanks very much.

Anton:

I just have one more question. When your technology puts you out of a job what are you going to do?

Ideal for marketers, advertisers, media and commercial communications professionals, Managing Marketing is a podcast hosted by Darren Woolley. Find all the episodes here

Want more articles like this? Subscribe to our newsletter:

Fill out my online form.

Anton is one of Australian's leading customer engagement consultants. With an eye for discovering greater marketing value and a love for listening to what customers are really saying about a brand. Anton has helped take global and local businesses including Microsoft, Nestlé, P&G, Gloria Jean's, Foxtel and American Express amongst others to the next level. Check out Anton's full bio here

We're Listening

Have something to say about this article?
Share it with us on Twitter, Facebook or LinkedIn

Tweet
Share
Share