Menu
Panel: How Sportsbet, Aus Post, Tennis Australia and Catch Group are harnessing AI

Panel: How Sportsbet, Aus Post, Tennis Australia and Catch Group are harnessing AI

Brand representatives join the discussion at the CMO-CIO Executive Connections event in Melbourne to share how they're tapping machine learning and artificial intelligence and the ethical considerations that go with it

From left: CMO's Nadia Cameron, Aus Post's Silvio Giorgio, Sportsbet's Tony Gruebner, Tennis Australia's Josie Brown, and Catch Group's Ryan Gracie

From left: CMO's Nadia Cameron, Aus Post's Silvio Giorgio, Sportsbet's Tony Gruebner, Tennis Australia's Josie Brown, and Catch Group's Ryan Gracie

Putting creativity and machine learning into the same sentence might seem like an oxymoron, but it’s only by combining both that brands will harness artificial intelligence for customer engagement success.

That was the consensus of a panel of marketing, data and insights leaders at the recent Executive Connections breakfast event in Melbourne, hosted by CMO and CIO magazines. The session featured Tennis Australia chief marketing and insights officer, Josie Brown; Australia Post GM data science, Silvio Giorgio; Sportsbet GM analytics, insights and modelling, Tony Gruebner; and Catch Group CMO, Ryan Gracie.

At Australia Post, AI is about the applications it delivers, and in particular, how it can empower employees to improve customer experiences, Giorgio said. Work to date has largely centred on operational improvements such as parcel tracking. The company delivers about 300 million parcels per year to 11 million delivery points, and has 6500 postman on the road.

The group’s back-end bot, dubbed ‘Dexter’, is designed to help staff scrutinise delivery data and takes into account three elements: Notify me, tell me and help me. 

“Dexter will monitor deliveries every day across the network, identifying parcels that might be at risk of running late and directing people to get them out so we can meet our delivery promise,” Giorgio told attendees. 

“We’re also conducting a trial that provides our call centre operators with the information they need to help customers when they call. We’re trialling parcel delivery success predictions – called happy parcels – where the data would suggest that the parcel will progress well, but could still result in someone calling our contact centres, because they’ve had a bad experience previously. It’s natural to want to make sure that the parcel is going to arrive on time, and when you’ve had that bad experience, you’re more likely to call us next time.

“We need to understand this so we can provide customers with the right information that gives them confidence that the parcel is on schedule as compared to when something goes genuinely wrong – which requires us to have information about what’s happened and when we expect the parcel to arrive. That way AI is really helping our people engage with customers in a different way.”   

At Sportsbet, machine learning and modelling are lifeblood capabilities generating pricing onsite. More recently, these are being harnessed for homepage personalisation, with a real-time recommendation model serving content to customers.

“When we talk about AI/ML, we talk about data products. The reason is it’s much more about what the model can do, than the model itself,” Gruebner said. “It’s the combination of model with data science and technology that’s the key enabler. But more and more, we focus on to what end we’re doing this stuff.”

Over at Tennis Australia, AI has largely materialised to date in performance data utilisation for players. “The interesting thing for us is how you start to cascade that data into something entertaining for consumers,” Brown said.

“During the 2019 Open, we saw some fantastic visualisation of performance data and insights. We have also started putting toes in the water with early marketing cases, such as a ticketing chatbot, so there’s some early experimentation.”

What is opening up broader AI utilisation is a org structure that encourages data scientists to collaborate and share ideas with the people setting business objectives and use cases, Brown said. “I think building an understanding of AI is more about what are the right pilots that align to some of those business opportunities and framing it right,” she said. 

One opportunity, for example, cascades from the performance piece. “When you recognise patterns in play or certain times of strokes, I’d love to see that cascade to a consumer level so you have the ultimate digital assistant that could help you with your own game and your own performance,” Brown said.

“Then I think about marketing itself and how we provide a better sales experience for tickets to our events, personalised content during events – that’s a huge area we want to bring together creativity with right data and technology to serve it up.”

Retail marketplace, Catch Group, is employing machine learning across most of its channel engagement. One use case is homepage recommendations driven algorithmically by revenue data.

“We’ll launch seven events per day and they’ll cascade down the page depending on how much revenue they generate,” Gracie said. In addition, the 20 million emails sent to customers per week are populated using character propensity based on a consumer’s buying and browser behaviour.

“We send highly segmented campaigns based on how customers have touched our channels along the chain. We also show different messages according to the channel a consumer arrived there from. If they’re bouncing off a page we’ll serve up incentives,” he said.

To get to this point, however, Catch Group learnt a few hard lessons about what it takes to use AI effectively.

“We quickly learnt we had terrible data and our data wasn’t organised at all. The analogy was crap in, crap out, and we just didn’t see the sales uptick we were expecting,” Gracie explained. “In the last nine months, we’ve introduced Redshift into the business, we’re using Snowplough [analytics software] and through the dashboard and Periscope, this gives us very granular data. We’re then joining the conduits and teaching the machine this product relates to this other product, and this is what the customer wants at the end.”

Today, Catch Group’s machine is able to show the right product at the right time to customers across digital channels. The team is now using data to create further efficiencies, such as minimising email wastage and improving targeting.

“My advice is if without the right data, you’ll waste a lot of time. We were too far ahead of ourselves,” Gracie said.  

Alongside the technology, Catch Group has a business analytics and data science team feeding the marketing team customer insights.

“These teams are sitting much closer together than they used to. They used to operate in silos and BA never really feed that data down to the marketing team,” Gracie said. “Now we have huge collaboration between the two and on a daily basis, and we come up with new lists we want to target. The BA guys get excited because they see the results.”

An example was a recent SMS campaign sent to 120,000 people, which generated upwards of $300,000 in incremental revenue. “That was exclusively targeted to people who had not opened our emails, push notifications or looked at the website,” Gracie added.

“We knew they were ripe for an incentive. Those deep insights are driving far more effective targeting.”

Foundational capabilities

So what other foundational capabilities must be in place to harness artificial intelligence? For Sportsbet, the main thing is the real-time component, firstly in terms of ingesting data to run models, then actioning data where it needs to be actioned. Key to this has been dynamic cloud-based infrastructure capacity.

On the people side meanwhile, the big challenge is skills. “We talk a lot about data scientists but they’re just one part of the chain. We have data engineers that sit around with business analysts, front-end developers, technology practitioners and then importantly, the business owners who can understand the business application and bring that to life,” Gruebner said.

“One of the big challenges is how to bring them all areas together.”

Brown also believed the foundation for AI success is people and partners. “One thing we’ve done for many years is a strategic relationship with Victoria University bringing across PHD students, data scientists and embedding them into Tennis Australia,” she said.

“This leads to rich conversations with really smart data individuals sitting alongside our data scientists or professionals team… harnessing all the right capabilities. We combine this with partners like Infosys, which helps us on the technology implementation side.”

But it’s only through creativity and business thinking that you make magic happen, Brown said. “Storytelling is one of the most important skillsets in the team to bring to life what you’re learning and allow it to have impact on the business,” she said.

Can versus should

And just because you can do AI, doesn’t mean you should. There are plenty of ethical considerations that need to be front and centre, panellists agreed. For example, while there’s plenty of interest in AI at Australia Post, Giorgio is proceeding with caution.

“We are explicit about AI and machine learning being here to help provide intelligence for our people to provide great customer experiences. We focus on improving service performance, customer experience and the safety of our people. We have a lot of posties on motorbikes, and we’re looking at ways to improve people safety every day,” he said.  

“We use AI as a tool so our employees can provide even better customer experiences. AI can never replace the personal interaction our customers love from our Posties. We want to keep our posties in jobs.”

Australia Post has also developed a governance framework to assess AI applications, largely driven by its role in the community.

“Another piece of the framework is satisfying the ‘creepiness test’,” Giorgio continued. . “For example, while as individuals we may have grown increasingly comfortable over time with sharing our location from our mobile devices, as a customer of Australia Post, one might not be so cool about us knowing what’s in your parcel – which is why we don’t.”

“That’s more around using data: What of my data are you using and am I comfortable with it? The second element is: What is the engine doing with the data.  

“We developed the framework to keep us in check, and we challenge ourselves with these questions often. It’s important that people understand we do have an ethical framework and that it’s in practice, then showcasing the case studies of how it helps people, is very powerful.”

Over at Sportsbet there’s a saying around AI: Do more with the same. “We don’t try and automate for the sake of automating people out of the workforce. We are trying to help supplement our employees’ jobs,” Gruebner said.

“Measuring a lot of this stuff is difficult too. Something like personalisation is a long loyalty play, and it’s difficult to work that out. So we take an agile approach to how we roll these things out. We try to understand what impact we are having then try to build an ROI based on that. There’s no magic bullet for how you measure these things.”

Along the theme of ethical AI use, one recent launch Gruebner was proud of was using machine learning to encourage responsible gambling.

“We now have a model in place to identify customers likely to have problems with gambling so we can interact and intervene earlier on. It’s not just improving customer experience, it’s also customer safety,” he said.   

Last lessons

As for any other lessons to date around AI, Giorgio said teaching his data team to broaden out beyond the analytics and to the application is important. “The other lesson is come up with a solution that leverages what the business already uses. And make it easy as possible for the people as close to the action to have access to information,” he said.  

“Providing someone with insight is like giving them an IKEA flatpack: They still have to put it together.”

Gruebner and Brown also advised organisations not to underestimate the power and wide range of skills required to harness AI well.

“Don’t forget the human factor,” Brown said. “I go back to Ivan Lendl, who is the most passionate user of data when he’s coaching. The data might say the player’s forehand on court was a winner and fantastic, but if his player walks off the court and says they weren’t happy with their forehand, you have to listen to the player and understand that emotion as well. Data plus human is still very important here.”

Finally, don’t forget about the creative aspect of using data and AI, Gracie said. “The one thing we keep throwing out is creativity. Marketing is underpinned by creativity, so just putting shit messages out there for the sake of it, won’t work in the end,” he warned.

“Let’s not forget the message and the medium need to match. Let’s keep marketing a creative discipline.”

Follow CMO on Twitter: @CMOAustralia, take part in the CMO conversation on LinkedIn: CMO ANZ, join us on Facebook: https://www.facebook.com/CMOAustralia, or check us out on Google+:google.com/+CmoAu

 

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags digital marketingdata analyticscustomer experience managementdata-driven marketingbrand strategyartificial intelligence

More about ANZAustraliaAustralia PostCMOFacebookGoogleInfosysSportsbetTennis AustraliaTwitterVictoria University

Show Comments
[]