Interview: Optus marketing program includes AI
- The future of marketing — and our addiction to attribution
- The role AI will play — and what people fear the most
- A marketer’s priorities, a melding of analytics, creative and the big idea.
Naomi Simson: At Optus there was a real sense of urgency driving the implementation of AI technology. What were the commercial drivers behind this strategic direction?
Angela Greenwood: It was a few things. It can be very, very difficult for us to understand what the right level of investment across channels is in digital, and how do we move money between these in real-time. Often, it’s very much a siloed approach to how much investment we’re putting into each digital media channel. We wanted to understand if we gave an autonomous AI tool the freedom to move funds between all those channels, and to serve the ‘right’ creative in real-time, what would happen? So there was a healthy amount of curiosity about what we could achieve.
NS: In a marketing environment where the open web is becoming a thing of the past and we’re dealing with the ‘Walled Gardens*’ — each of them claiming they ‘own the customer’ — the notion of attribution becomes even more complicated. We as marketers focus so heavily on attribution, so what was that conversation like on your journey to AI?
AG: We still have a very healthy interest in attribution. We still put a lot of effort into investigating effectiveness — but you can only ever look at that retrospectively, not future-facing. And for us, the amount of sessions we attribute to one single channel — when we know that a customer’s journey is way more complex than that — means we can get really tripped up on thinking one channel or piece of creative is more or less effective than it is in reality. What we’ve been able to see with AI is that some of the creative constructs that work today, no longer work tomorrow — so we want to be able to see that in real-time.
NS: Being able to test ideas at scale was one of the driving factors that attracted you to Albert AI. Tell us about that.
AG: AI can take a lot of the heavy lifting away from some of the lower-level tasks around digital media buying. But what it actually creates is a lot more tasks around how you feed this engine with enough creative to be able to personalise at scale to prospecting leads. Because the possibilities are endless — they’re only really limited by what we’re able to put out there.
NS: Let’s talk about the agency relationship, and the human side of AI and what it means for the people on the team and their concerns and challenges
AG: I was pretty concerned and challenged myself. As performance marketers we have a bit of a reputation for being just a little controlling, so to be able to go from a very detailed digital marketing plan that’s broken down to the nth degree, to a single line item that says, ‘here are the dollars, go do’, that’s terrifying. It was really important to get the set-up right. We had to get back to basic things like getting our naming conventions right and making sure all of our campaign structures were set up correctly.
NS: I can imagine there was a lot of fear for the individuals involved — but as you’ve noted it’s really the low-hanging fruit that AI takes care of, to really free up the people to focus on those interesting strategy and learning pieces. Tell us about that journey for your people, and your agencies, and how they shifted from the execution to the strategic.
AG: It will never cease to amaze me how much people will cling to low-value tasks. So once you separate your people and your agencies from all that, you free yourselves up to think more strategically — like how we’re going to tailor our value propositions to different audiences, how can we actually do that really personalised creative at scale, and how can we take the insights we’re getting from the engine and do something with it? And that’s the really big step that we’ve been able to take. And also, because this type of tool works very fluidly across all the different digital channels, it has actually opened up opportunities for our internal teams to become more cross-functional, and they’re now thinking much more holistically about the customer journey.
NS: What does success in marketing look like for Optus?
AG: It’s a number of things. We want to be as efficient as possible. Digital still pays a really massive role in driving brand consideration, but it’s also really important for us to keep building the top of the funnel through that activity. It’s really important with an AI solution that you’re actually pointing in the right direction — because if you feed it the wrong signals and optimise towards the wrong thing, it will go really hard after the wrong thing. So that’s been a really important learning — how do we make sure that we’re actually optimising the right activity to the right result? And that’s never going to be uniform across everything that we do.
NS: Let’s talk about cookies. We didn’t just set this up for now, we wanted to plan for the future — so how does this AI solution deal with privacy issues and knowing who your customers are?
AG: For advertisers, the most important thing we can do is make the most of all our first-party data, and being a telco that’s a unique position to be in. We do actually have a fair bit of that, so for us, it’s about how we can we utilise that first-party data for the AI to find suitable look-a-like audiences. And then how do we incentivise uses to engage with our own platforms? How do we ensure we get more people using our app so that we are not dependant on the outside world for that view of our customer.
NS: Do you have any particular campaigns you have run that have helped you identify intent to purchase?
AG: Where we see value in Albert and how some of our assumptions have been challenged is around thinking that certain audiences have intent for a certain product, and how Albert then runs through a full range and offers alternatives. We’ve actually found some really interesting cross overs between audiences in terms of intent that we never would have understood before Albert. For example, if someone has intent to buy a post-paid mobile, they actually have a strong intent for accessories as well, so then how do we capture that as part of the ongoing conversation with that customer?
NS: That’s what AI does particularly well — he’s looking for, based on certain previous behaviours, that intention data, and taking all of that information to start predicting ‘what next?’ It’s almost impossible for a human to do that.
AG: You would need a massive team of data scientists to achieve anything similar at scale. And anyone who’s tried to hire a data scientist knows how hard that is. Albert enables us to do all this very rapidly.
We have had a very established test and learn program at Optus for many years, but the speed now and the scale at which we can get those insights is just so much faster now. We are an incredibly competitive category — telco is a blood sport, and it’s very much about how you gain market share, and taking market share from competitors. So we will do anything we can to get a competitive advantage.
NS: What does the future look like for Optus and your AI journey? And I’m talking no more than the next 6–12 months.
AG: For us it’s about how do we get more signals in? How can we get more data in for that to work for us? How do we do a better job of ingesting offline data to optimise towards an omnichannel result, and how do we better leverage our first-party data? And it’s also about the creative side — we have only dipped our toes in terms of what Albert is able to do when it comes to creative optimisation, so for us, it’s about how we set up so many different creative variants to be able to really maximise results.
“Don’t fear the machines. It frees us to be more creative and more strategic and that’s a win for the client and agencies,” Angela Greenwood.
To learn more about the Optus journey to AI martec click here for the case study.
* Walled gardens are sites that don’t have open access to data or don’t allow access to be served through third-party adtech infrastructure. Examples include Facebook, Amazon, WeChat, Apple News and SnapChat.
Originally published at Naomi Simson.