Artificial intelligence is no longer a sci-fi dream. Companies such as Spotify and Netflix are already using complicated AI algorithms to collect and combine consumer data to tailor recommendations and target advertising. It’s only a matter of time before brands in all industries start plugging their marketing CRMs into similar platforms to improve customer experience and automate a range of marketing activities.

In this interview, Michael Buckley, managing director of Accenture Interactive Australia and New Zealand, and Tim O’Neill, Reactive founder, discuss what all this means and where they think AI is going in the future.

Accenture Interactive acquired Reactive earlier this year to bolster its digital marketing and technology solutions. Reactive looks after UX and front-end platform development, specialising in apps, ecommerce, websites and social.

 

Marketing: I’d like to have a chat about artificial intelligence, particularly in relation to experience platforms. It seems there’s a growing sophistication in marketing platforms, with web content management systems evolving into experience platforms and that’s been driven by particular customer demand. Can you explain some of the reasons for this evolution?

Michael Buckley: Tim and I have both been in this industry for over 20 years. Traditionally content sat on an open-source platform, and what we’ve seen is that when you see Adobe Sitecore and platforms like Hybris, they’ve managed to bring a digital content system together in one place. So if you have offline or online content – if you have a web portal, a mobile platform, a bus shelter, an outdoor poster, you can now manage content across all of those platforms from one single place. Could you have done that traditionally? Yes you could’ve manually, but now everything’s moved into a single platform to allow you to manage all the content much more easily and more cost-effectively.

A creative agency can access and develop content to put into the platform, a client can do the same thing and a development agency can also access that content, including a publisher. The whole idea in the last five years is to put this content in one single place. Where we’ve moved to is how do you bring content plus data plus the platform together. Now what that means – and this is the artificial intelligence piece – if you’re walking down the street and a beacon through your phone goes off, I can send you a Coca-Cola message on the fly and I can also make bus shelters on the side of the street also react to that same message through artificial intelligence. The messages are becoming far more automated.

 

M: So bringing all of that together requires artificial intelligence? Was it possible to do in traditional computing?

Tim O’Neill: Anyone in the marketing department in the last three years, their job has become exponentially more difficult because they need to create so much more content than they ever did. You’re able to and you want to personalise content to different people, and that’s what some of the platforms have allowed you to do for a while. Then the other thing is the number of devices that people want to consume that content on. As Michael said, it’s not just mobile phones and 20 different types of mobile phones, there’s iPads and there’s bus shelters and there’s digital point-of-sale, so those things have made the role of the marketer so much bigger. But then on the flipside you have data coming in, collected from all those devices, and it’s about centralising the store of data against the customer so you’ve got this much richer view. You’ve got all this content that you want to publish to all these different devices – it’s a big job to do manually but you’ve got all this data and that’s where machine learning or artificial intelligence can sit between the two of those and look at that problem. It can look at the data and work out what content should go to which person on which device.

 

M: For people who don’t know much about machine learning and artificial intelligence, how do you define it? How does it differ from, a) programmable computers, and b) walking, talking robots from the movies?

MB: Spotify’s probably the best example and the simplest to explain. So georgia here loves Elton John’s Rocket Man as a song, so she listens to that song and then she might go and listen to Bon Jovi, then it will suddenly link to other people like Georgia that also listen to those songs and they also like other songs like this. So what happens is you’ve got two clusters of data and they find similarities between those two data sets. Spotify has an algorithm that understands that behaviour.

TO: Extending on that, if you imagine three or four years ago, on your iPhone or iPod, you would manually create playlists of the songs that you’d like and you’d name that playlist… that’s just a thing of the past now. Spotify just creates playlists for you.

To what is the difference between artificial intelligence and a computer, the key really is in the name; placing something that we manually used our intelligence for a few years ago, which was thinking ‘I like Elton John, I’m going to match it with these songs, I’m going to create a playlist’, and instead the computer is using that intelligence and doing exactly the same thing but just doing it artificially. It’s reducing or completely deleting manual effort, so you can spend your time elsewhere. That’s the benefit, it’ll reduce manual effort of the marketing department and let them spend their time elsewhere.

 

M: What are the opportunities for marketers? What kind of brands would be best suited to taking stuff like this on board?

MB: We used the Coke example, they use an experience platform right now to change the content at Times Square in New York on the fly. So instead of sending in your creative a month early and saying, ‘OK, in a month this will go up on the platform’, we’re now able to on the fly react to customers in Times Square.

Not our clients, but clients like eBay – if you’re in the market for a bike, programmatic marketing kicks in, and if you have your experience platform set up with the right content overlayed by artificial intelligence, overlayed by programmatic marketing, you can suddenly start bringing marketing, content, data and the platform together.

And why does that work? Marketers have been frustrated with the waste of media dollars; I think in the high-end brands, they don’t need increased brand awareness, what they need is specific messages to either an existing customer or a potential customer, and understanding who that customer is.

So what’s happening is online marketing, where we had the ability to serve a particular ad to a particular person, is now going offline. When you log into Apple TV or Netflix, that is sitting on an experience platform linked to an artificial intelligence layer that is linked to content like House of Cards.

Tim: There’s definitely a big aspect of true artificial intelligence within experience platforms being ‘watch this space’, you know, ‘coming soon’, it’s really on the cusp of it now. The example of the recommendations engine as one within Netflix or Spotify, they get pretty sophisticated. It’s pretty common now for big experience platforms to be hosted in the cloud, and then you can tap into services like a machine learning application – Google has one and Microsoft has one, so you don’t have to build it yourself. And that lets you extend the experience platform into not just making recommendations but learning and improving over time.

Some of the things you kind of class as lightweight AI, the idea that you can create a test on a website for example, go ‘blue button’ this is ‘red button’, it’s not really artificial intelligence but it will give you an outcome and help to improve the site. But right now the marketing department sets that test up and the results will then automatically take place. The winning button colour will automatically take the place of the losing button colour. The idea that in a few years time you won’t have to interact and set that test up; the website or the experience platform will go, ‘This web page is not performing as well as I would like, I’m just going to see what happens when I change the buttons from blue to red’, and then it will let you know how it went. That’s something that’s exciting and pretty interesting.

 

M: So that sounds exciting and it also sounds scary because it’s a little bit out of the marketer’s’ control. What are some of the drawbacks or fears that would make people hesitate to use this?

MB: In some cases you have a marketer that has a dream and an internal structure that doesn’t enable them to deliver that dream. You hear a lot of CMOs and CIOs, the technology bridge between the two, and how do you bring technology and marketing together internally at the organisation to enables that to happen. That’s a big component to get over, and that’s something that we spend a lot of time with clients because Accenture’s traditional clients are CIOs. We’re bringing our technology prowess and our marketing prowess together, to bring those two conversations together.

 

M: What are some of the main challenges you find there, bringing technology and marketing together?

MB: There’s a few issues that I see. Sometimes the CMO is not necessarily high enough in the organisation to allow them to make a critical decision as one that we’re talking about. I traditionally see the CIO at the board level, and the CMO probably at the board level maybe 50% of the time now, and traditionally it’s grown from 30% to 50%. It’s increasing because the customer is now at the face of the brand and demanding a better experience. The CEO is bringing the CMO higher up in the organisation.

I also see that in the case where the CIO and the CMO don’t agree, they create a role called the CDO, chief digital officer. Sometimes it’s just the organisation is so big that you need to bring in that role to grow digital to a point that it becomes really important.

 

M: It seems like a lot of the answer lies in culture?

MB: It’s partly culture but it’s also partly a traditional structure or a traditional business not being agile enough to change quickly enough. We almost always now bring change management to our strategy to our clients for that exact reason.

 

M: What about investment into this type of technology, is that a barrier? When we’re talking about cutting-edge AI, is that something that is really expensive?

MB: No, I don’t see a barrier. There’s certainly no barrier in an experience platform anymore; it’s almost a given. Is there a way to bring two data sets together to create a very simple AI platform? There’s no barrier. I have no doubt that the algorithm behind Netflix and Spotify is extremely complex and has a very high-end data capability. If you wanted that level of sophistication it’s potentially cost derivative but at a level that can you bring content and platform and data together, it is very easily accessible for every client.

 

M: Looking to the future, what can we expect with brands taking on this technology – have you got a vision or a hunch on where it will all go?

TO: The actual platforms are very much widely adopted by pretty much all of our clients, so it’s more the future is what do those platforms do with cloud computing and with those big data sets and with building artificial intelligence into software which then will become available to all of our clients that have already bought the software.

I’ve been talking in the last few days to one of our clients and one of the platform providers exactly on this question and as I referenced before, they’re really excited about this Microsoft machine learning which they can tap into and then build more sophistication into their platform. Really the onus is on the software vendors at the moment, but part of that is, one of the constraints is really what Michael said before: how much do you want the computer to do for you, and when do you get in trouble? When do you overstep the mark in terms of governance or overstep the mark in terms of brand? Having those kind of controls and measures in place is going to be the big hurdle that they’ll need to get their heads around.

The simple example that I mentioned before is: let’s say the experience platform is managing the mobile app, and the mobile app thinks for itself. It goes ‘These buttons are going to be much more effective if they’re red than if they’re blue. That would really help the performance of this app.’ It might go ahead and do that, but if the app belongs to ANZ and their brand’s blue, then red colours might perform better but they’re not on brand. So how do you tell the computer that?

 

M: That’s a great point, there’s a lot to think about in that and it’ll be interesting to see what different brands do with it. Thanks so much for your time.

MB: No worries. Appreciate it.

TO: Thanks, bye.