S1 Ep 27
This week, Colin Peacock is joined by Professor Michael Townsley from Griffith University to focus on data driven defences in e-commerce loss privation, focusing on a case study which looks at one major e-commerce retailer’s journey to build a global digital loss prevention capability.
They reflect on how the smart framing of fraud as a “lost sale” helped engage multiple internal stakeholders, and how serious investment in data collection and analysis has driven impressive improvements, including a jump in order acceptance rates and millions in recovered revenue.
They also look at emerging risks like account takeovers, crypto payments, and AI-enabled scams, and how retailers must stay agile as bad actors evolve. If improving digital loss prevention is in your remit, check out this week’s podcast.
Podcast summary
In this episode of the Sell More and Lose Less podcast, host Colin Peacock talks with Professor Michael Townsley of Griffith University about E-Commerce Loss Prevention, focusing on a global e-commerce retailer.
They discuss how the retailer has built a digital loss prevention capability, focusing on the framing of fraud, serious investment in data, and forensic analysis of fraud patterns.
The conversation highlights how better data and stakeholder engagement enabled higher order acceptance rates and significant bottom-line gains.
It also explores emerging challenges such as crypto payments, AI-enabled scams and the adaptability of offenders, all of which matter for retailers looking to strengthen digital loss prevention.
Cleaned transcript
Introduction
COLIN:
Hello. In this episode, I’m joined by Professor Michael Townsley of Griffith University in Australia.
We’re reflecting on another of our Learning by Doing case studies with a global retailer in e‑commerce, and the lessons they’ve learnt in setting up and establishing a digital loss prevention capability.
Key takeaways – complexity and context
COLIN:
Tell me, you’re studying this, you’ve got a great research project that we’re doing with you.
What were some of your takeaways, three or four takeaways you got from the meeting?
MICHAEL:
Yeah, absolutely.
Before the takeaways, let me say that I was really impressed with the presentation.
I don’t think I quite appreciated the complexity of that organisation.
They had a supply chain that is a bit more involved or internal than some of the other retailers that we’ve seen present.
COLIN:
The complexity… Well, the geographic spread was amazing.
MICHAEL:
The geographic spread was amazing, right?
They’re dealing with everything from Mexico to Australia, UK, Chile, China. It’s crazy.
COLIN:
Yeah, yeah, yeah.
MICHAEL:
So that, I think, was even… Even though we thought it was very impressive and interesting, I think extra marks for the complexity in which they’re operating.
Takeaway 1 – framing fraud as lost sales
MICHAEL:
So, my first takeaway was I was really impressed with the framing of the fraud problem that was presented.
The presenter said something like they’ve got a mantra about any incident of fraud is a sale not fulfilled, or words to that effect.
That had cut through with other parts of the organisation.
I thought that was a very deft use of common language across an organisation so everyone is on board or takes seriously a fraud problem.
I just thought that lots of retailers could take a lesson from that sort of framing of the problem.
COLIN:
Yeah, a lost sale, right? Yeah, it was very clear. It was good.
Takeaway 2 – serious investment in data to reduce e-commerce losses
MICHAEL:
The second thing, they did talk about a journey, but part of that journey was around some quite serious investments in data collection apparatus in a hard-to-measure problem.
That was really telling in some of their results, which were only achievable by being really, really serious about measuring, in quite precise ways, some of the fraud problems they had.
In some senses, you sort of implied that they weren’t aware of these problems, or by no means the magnitude of the problem.
But they put a lot of effort into measuring indicators, or coming up with indicators, that then could be used to shape the fraud problem.
Again, many retailers could take a lesson from that.
COLIN:
And a long-term investment as well.
I’m not sure whether this is your third point, but linked to the data, one of the things that came across for me was that one of the changes that they’ve been able to introduce, because of the team that they put together, was they’re able to accept more orders.
They became a bit less scared of the fraud and moved from, I think, 95% to 99% acceptance of orders.
That’s generated another $100 million or something to the bottom line without, it seems, a huge uptick in the level of fraud.
Takeaway 3 – forensic analysis and rule refinement
MICHAEL:
Yeah, that’s exactly… You’re great at anticipation.
My third takeaway was that once you’ve got the data, then it seemed to be almost a forensic level of analysis of the problem.
There was one story about how they had quite a low acceptance rate in one territory.
They went through all of the rejections, if that’s the right term, and identified whether they would have rejected or not.
They then worked with their provider to alter some of their risk rules.
That is something you can only do once you’ve got the foundational data right.
This work is really quite difficult and might seem quite tedious, but it pays off over the long term.
They were able to get that acceptance rate up and up.
So those are the three things:
– framing of the problem
– a relentless focus on collecting data
– and then this nose to the grindstone problem-solving, problem analysis phase.
All of that, interestingly, translates into the fact that they can automate some of those functions and then really get leverage on it.
I thought it was just a fantastic journey story that a lot of people could get learnings from, for the way that those problems apply to their organisation.
Stakeholder management and board engagement
COLIN:
I was really impressed by their analysis of stakeholders.
They’d done a really good job thinking through whose hand do we also need to hold.
They’d got some direct, regular contacts with the board.
Again, it’s around your framing point, positioning this work as relevant to all the different stakeholders: the immediate ones, the secondary ones, the tertiary ones, and even external stakeholders.
They put together reports that went to the board, making this group highly relevant.
As if £100 million on the bottom line doesn’t make you relevant, that soon gets you attention, I think.
MICHAEL:
Yeah, absolutely.
You can imagine taking on a challenge like that, getting those sorts of returns, and then having all those stakeholders that understand the mission and you have great communication with.
You then wonder what other problems that opens up if you have demonstrated your credibility and legitimacy in tackling particular problems.
That opens you up for people to collaborate with on other problems, like your second biggest problem.
While some of this might not sound very sexy – collecting data and grinding away at these problems – the dividend comes over the long term.
Peer reactions and the role of data
COLIN:
It was interesting to hear the reaction and the questions from the others in the group.
Your point about the data seemed to attract the most number of positive likes, if we move into the social media world.
The other retailers were blown away by the approach they’d taken to data and the amount of data they had available, which was good to see.
MICHAEL:
It seems like if you’re not building your own dashboard, then you might want to, pronto, because everyone seems to be.
We’re all at different stages but are really taking the data bit seriously.
The way that you can communicate that to stakeholders is only going to crystallise what the challenge is for any of the retailers.
Future challenges – crypto and AI-enabled scams
COLIN:
I like the way that at the end, the retailers talked to some of the future challenges.
Crypto seemed a big one.
What do we do about crypto? Do we take crypto?
That was something interesting on their mind.
Also, the number of attacks, perhaps, with the scams coming out using AI and language and modelling people’s voices, is pretty scary as well.
So there’s a lot for them to deal with going forward.
E-commerce loss prevention. Dealing with adaptive offenders and limits of automation
MICHAEL:
The other thing that I noted was that it’s great you have these systems and this automation.
I did like the flexibility of wanting the customer service agents doing less and the CRM doing more, but clearly the service agents still have some autonomy or latitude.
There were hints that the bad actors are adaptable.
There was one example where they found a fraud ring and it seemed to be email takeover was the vector.
These bad actors were taking over good customers’ accounts.
Removing the easy opportunities is great and that’s what we should be doing.
But you need to be constantly vigilant on changes in patterns and not be too reliant on any of that automation.
Offenders will adapt and the methods will change.
If the algorithms that we have built aren’t able to detect that, then it leaves you vulnerable.
So we just need to always be vigilant that some of the bad actors can adapt their methods and find new vulnerabilities in processes.
Closing remarks
COLIN:
That’s great. Well, thank you very much, Michael.
That was a really great summary and set of notes for us to reflect on.
Those people who weren’t there can pick up something there and perhaps come back to some of our future meetings.
Thank you very much for taking the time.
MICHAEL:
Yeah, no worries. Thanks, Colin. See you. Bye-bye.
