Finance & the Future of Risk | Is Fraud Messing With Your Margins?

0:00
Uh welcome everybody to the finance and the future of risk. Um I think this is a
0:05
super timely topic. Um we’re going to talk about fraud and we’re going to talk
0:10
about uh if it’s messing with your margins. So today uh we have a great
0:15
panel lined up for you. My name is Kevin. I am the SVP of customer
0:19
experience and trust and safety at SIFT. I’ve been at CIFF for almost eight years
0:24
now and prior to that led my own risk fraud trust and safety teams at
0:29
companies like Google and Meta and Square. And today um it’s my pleasure to
0:34
introduce our panel to talk about something that’s super near and dear
0:37
important to me. Uh but with that said, I’ll first pass it over to Steve.
0:42
>> Hey everybody, thanks so much for joining the webinar. I’m Steve and I’m
0:45
lucky to be the CFO of SIFT. I’ve had a uh been the CFO of a number of
0:50
business-to business enterprise SAS companies. So familiar with this
0:54
challenge and excited now to be with this company. We help solve the problem
0:58
of fraud. Ashley, over to you.
1:01
>> Yes, sir. So, I’m Ashley Johnson, director of cash and treasury management
1:04
for Texas Roadhouse. Been in the treasury department for Texas Roadhouse
1:07
for almost 11 years. Prior to my time here, I was a banker. So, I have kind of
1:12
the banking side of the business and that knowledge and then now the merchant
1:15
side as well. So, it’s nice to be here. Thanks for having me.
1:18
Andrew.
1:20
>> Hey everyone. I’m Andrew Meyers, uh, manager at Deote. Uh, my specialization
1:26
is fraud detection. Uh, we work with a lot of large financial services clients
1:31
as well as um, e-commerce merchants um, helping them from everything
1:37
uh, you know, fraud system architecture um, implementation um, system
1:42
optimization. Um, so we work on a really broad range of platforms, um, SIFT
1:48
included. Um, and, uh, you know, pretty much help clients with the end to end,
1:53
uh, fraud detection, customer experience. Um, but, uh, nice to be here
1:58
today. Thank you.
2:00
>> All right, welcome everybody. So, uh, for those of you on the call that are
2:04
not familiar with SIFT, we are an AI powered fraud platform that helps
2:09
businesses protect themselves and their consumers against various forms of
2:13
online fraud. So, we’re going to touch on a lot of that stuff today. Um, but
2:17
the biggies here include things like payment fraud, so using stolen credit
2:20
card credit cards online, account takeover, content abuse, fighting these
2:25
things called chargebacks. I know many of us are in our our Super Bowl season
2:29
here where we just had Black Friday and the Cyber Five and kind of heading into
2:33
this larger holiday season. Um, come Q1, we’re probably going to get hit with
2:37
some chargebacks here from consumers that may not recognize charges or
2:41
experience some sort of fraud. That certainly comes back to the finance team
2:44
in terms of how to deal with those as well. And so that’s just a little bit of
2:48
a background on Sift and who we are and what we do. Um, and with that, really
2:54
want to spend the bulk of the time here digging into with our kind of esteemed
2:58
panelists here around the finances, the the ones and zeros, if you will, when it
3:03
comes to these numbers. What are businesses doing in this side? How are
3:07
they protecting themselves and their businesses, their end consumers, uh,
3:11
against this type of abuse? And, um, really what is what are they looking at
3:16
either now and kind of potentially even taking a peak into 2025? um what changes
3:22
do they expect? Um I think we’ll go straight into some questions with folks.
3:26
If you uh are kind of in the audience and you do have questions for us, feel
3:30
free to put them in the Zoom chat. Um we will definitely try and get to those by
3:35
the end of the call. And so with that, why don’t we actually just jump into the
3:40
the questions straight away. I’m going to stop sharing my screen now and so we
3:43
can see our faces here. Um and let’s go. First question is um how has the
3:49
landscape of financial fraud evolved in recent years? Um and what new challenges
3:54
are finance professionals facing today? And so with that, let’s uh maybe Ashley,
3:58
you want to step in on that one?
4:00
>> Sure. So in 2019, we successfully completed as a company the acceptance of
4:04
EMV payments. So um with that liability shift of fraud, our instore fraud that
4:09
we were seeing went to the card not present environment. And we know what
4:13
2020 brought all of us, which was the lovely world of COVID. And before COVID,
4:18
we were not focused on um to-go sales, really online ordering. That wasn’t
4:23
Texas Roadhouse’s way of life. So, we quickly had to pivot. Um so, pivoting
4:28
from an instore, really focusing on that dining experience with our guests to a
4:32
car not present environment. We had to make some changes. All of our sales were
4:36
taking place there, but then we also saw fraud start to uptick in that
4:39
environment as well. you know, COVID, you know, ended after a couple of years,
4:44
but we continued to see our online sales stay pretty static. Um, so even though
4:49
we weren’t a company previously that were focused on card not present sales,
4:53
we then were still staying in that in that high volume area. So, uh, fraud
4:57
wasn’t going away. It had just shifted for us on where it was taking place. And
5:01
so, that really led us to looking for a solution that would, you know,
5:06
ultimately help our operators and help our guests.
5:09
And actually maybe just a quick followup on that one. Yeah. Like obviously co
5:13
disrupted the lives of everybody out there.
5:16
>> When it came to Texas Roadhouse and you mentioned making that that pivot pretty
5:20
hard from let’s say instore to to digital. Were there plans already in
5:25
place and you were like oh we just need to accelerate it or more was it more
5:30
like oh we weren’t expecting to do this really at all and now like given the
5:34
reality is like we need to do something right now.
5:37
>> Yeah. So, we had accepted, you know, online orders. It was 3% of our total
5:41
sales as a company before COVID. Um, then it went from 3% of total sales to
5:46
now it’s 100% of your sales are online. So, um, it was we were prepped but not
5:51
prepped. So, we had to switch our provider for online ordering in the
5:56
middle of that within the first three months because the order volumes the the
6:00
previous provider that we had couldn’t couldn’t handle the amount of orders
6:03
that were coming through from a volume perspective. Um so we had to switch and
6:06
pivot to that. Um so before that we weren’t really looking at fraud. I mean
6:12
it wasn’t really anything that um was a main driver or a factor for us. Um and
6:17
then with those changes with some of the pivots we had to make then now it’s 18
6:21
to 20% of our sales even today. So it it didn’t go back to that 3% which was a
6:27
non-issue for us at that time. Okay.
6:32
>> Anyone else here? Yeah, I can jump in here. Um, a big
6:39
change that we’re seeing kind of just across the landscape both on the
6:43
merchant side and the, you know, the financial institution side is, um, the
6:48
increase in scams and identity theft. Um, a lot of this is bolstered by kind
6:54
of the new tools that are available for fraudsters. those different kind of
6:58
generative AI excuse me AI based tools um deep fake tools um I mean one thing
7:04
just as simple as chat GPT which I think everybody has heard of um over the past
7:10
year and this was an article um in the Wall Street Journal um there was a 14x
7:16
increase in the volume of fishing emails that have gone out um to people um so
7:22
we’re seeing a lot more compromised credentials compromised PII which kind
7:28
of enhances the fraudster’s ability to make those bad transactions. They’re not
7:33
just using the compromised card now um to put through a transaction. They’re
7:37
also able to take over the entire account um if they get declined by a
7:42
fraud detection strategy. They’re able to go in and authorize um that it was in
7:46
fact them who made the uh the transaction, you know, posing as the
7:50
good customer. Um, but what we’re seeing from a technology standpoint, and that’s
7:56
kind of the lens that I’m probably gonna speak on a lot today because that’s
8:00
that’s mainly what I do, um, is, um, work with the different technologies
8:04
that are out there. Um but bringing in those different cyber detection signals
8:10
um what you know the person was doing on the device when they made the
8:14
transaction, your IP geoloccation, your device ID and using different consortium
8:20
tools to look at you know bad device history, bad IP range history and a
8:25
broad range of other things. bringing those signals into your transaction
8:30
detection actually really improves um you know kind of the the performance
8:36
that you get out out of those strategies. Um but it also gives you
8:40
different signals to fight chargebacks with. Um so for example you you have
8:46
somebody who made an account you know a couple months ago they used a specific
8:50
device a specific IP address. Um, so now they’re trying to claim a transaction as
8:56
fraud. And you know, another fraud type we’re seeing a huge increase with is
9:01
firstparty fraud. Um, those bogus claims, those bogus chargebacks that you
9:05
see. Um, you’re able to provide more information just beyond, you know, we
9:10
sent this to the the customer’s good shipping address. Um, we now know the
9:15
device that was used. This customer has been using that device for a long time.
9:19
The IP geoloccation is right by their house. Um but providing that different
9:24
data to the networks can get you those um kind of better decisions on the
9:29
network side on uh the different liability decisions.
9:34
>> Plus one to that I think one PSA for those fraud professional out
9:38
professionals out there when you’re dealing with charge tax and going
9:41
through that representment process having those additional signals that
9:44
Andrew called out is super beneficial. That’s going to up your win rate
9:48
essentially. That’s more money back in your pocket. Great. Um, let’s move on
9:54
here in terms of the next question. Wanted to focus a little bit more on
9:59
broad impact. Um, and kind of how does that specifically impact company
10:03
investments and overall financial health? Uh, Steve, you want to jump in
10:07
on that one?
10:07
>> Yeah. Oh, I’d love to. And it’s top of mind for many of us as we go through our
10:12
2025 budgeting. Um, I know we’re not the only one. All of you financial
10:16
professionals are going through that as well. So, yeah. So the so the impacts on
10:20
financial health, it it lowers revenue and it’s and I’d say it’s worse than
10:25
that. Not only does it lower your revenue, but as as accounting pros know,
10:29
you need to make an assumption around what that amount will be in the future.
10:34
So it lowers your revenue and it drives complexity within your business when it
10:38
comes to recording the results because you have to estimate for what that is in
10:42
the future. Accounting is more complex as a result. it may also present as as
10:48
higher bad debt. Uh so not only do you have that topline problem and bad debt
10:53
issue but you also drive additional costs you know in tracking and analyzing
10:57
and reporting and attempting to address and reduce the fraud. So huge financial
11:02
impact that impact your revenue growth your expense efficiency which is more
11:06
important than any time during during my several you know three decades of
11:10
working than operating profit and ultimately valuation. So big impacts uh
11:15
to an organization particularly when it’s increasing and it’s you’re trying
11:20
to figure out like you know how can I get my arms around this increasing
11:24
metric. So now going back to the investments again perfect time to talk
11:28
about it with uh the planning season underway the the big buckets as as we
11:33
look at it are uh you know prevention detection and mitigation. So those are
11:38
the the the activities you’re trying to do. Now as far as the specific expense
11:41
type items um many ways cyber security measures lots of stuff there right you
11:47
have to have to have a secure network end toend encryption endpoint protection
11:52
multifactor authentication uh and then my favorite I think for many
11:56
of you the zero trust security models you know yeah it’s it’s just an overhead
12:01
on the business that IT teams have to implement and we all have to follow but
12:06
that’s the environment we’re in right so unfortunately All these things we have
12:10
to do they drive a little friction for us particularly inside the business but
12:14
they’re are huge important measures and wanted to call that one out first. So on
12:18
top of that there’s more investments or ultimately expense that’s made. You got
12:23
a tech stack and storage for the data gathering reporting analysis. If it’s a
12:28
machine learning based model that adds expense if you do it internally or you
12:33
have to go get someone outside to help with your tech stack. You have people to
12:36
run all these processes and this impacts every department everywhere. You know
12:41
this is not just it does this or platform group does that or finance does
12:46
that other thing. These are investments made across across the entire
12:49
organization.
12:52
>> Awesome. Um anyone else?
12:55
>> And I feel like you have to make those decisions Steve based on fraud based on
12:59
the area of the store. So with us, I mean, we’re across the country, so
13:02
there’s different hot spots of stores that have more fraud than others. You
13:06
know, we see specifically California, Michigan, Florida, they have more fraud
13:10
than some of our rural areas like here in Kentucky where I’m based. So, you
13:14
know, you want to be sure that you are looking at fraud in um a holistic
13:18
approach kind of because you have to make sure that you can drill down and
13:21
prevent fraud in those hot pockets without affecting the sales of your
13:25
stores in Kentucky that may not be dealing with fraud. So, you know, that
13:28
can also affect the financial growth and the health of the company based on where
13:32
that fraud is happening.
13:34
>> Yeah, for sure. Again, more expense, right? It’s not just monolithic to your
13:38
point. It’s like now you have to have the visibility into those different
13:41
hotspots.
13:42
>> Super interesting.
13:45
>> Makes sense. Steve, you reminded me of a story in terms of when we were talking
13:48
about being able to estimate volumes. Uh se several years ago when I was working
13:52
at Square, I was having a conversation with our CFO at the time and I was I
13:57
managed the fraud team and uh I went into our one-on-one thinking, “Oh, this
14:02
is going to be great. Our losses are super low. Um they’re actually lower
14:06
than expected. Um she’s going to be very happy with this.” And during that time,
14:10
we were prepping to go public as a company. And as you go public as a
14:15
company, what is very good to have is essentially um clean balance sheets and
14:21
you want predictability quarter over quarter, month over month, etc. And I
14:25
went into her and I said, hey, look, our fraud losses are lower than expected.
14:28
Like this is great. Uh and she actually said to me, Kevin, as we prep to go
14:33
public here, we want to make sure that we are in line. Not only like we have a
14:37
band, like our losses, we don’t want to exceed X and we don’t actually don’t
14:40
want to go below Y. And so she wasn’t upset per se, but it was it was a good
14:45
lesson for me like, oh, we want to have a very smooth kind of kind of guard
14:50
rails here on the high and the low end when it comes to this type of
14:53
predictability. So that was a good lesson for me.
14:56
>> So interesting.
14:57
>> Yep.
14:58
>> Yeah.
15:00
>> You think you’re great news and it’s like, hey, but you know, it’s to some
15:04
degree uncertainty. It’s in that banner of uncertainty. Okay. Well, now where
15:08
will it be next quarter? Yeah. you know, as a as a CFO, you know, whether I’m
15:12
delivering the news to to my boss or I’m receiving the news, that that is a super
15:17
interesting dynamic.
15:18
>> Yeah. Um, so let’s kind of move on to kind of what key features should finance
15:24
leaders look for in an enterprise fraud solution to ensure a quick time to value
15:30
and and really a strong ROI at the end of the day.
15:35
>> Okay. Yeah, I can take this one. Um so a couple of things and you know a lot of
15:40
this comes out of lessons learned on working on these types of
15:45
implementations. Um but the first one is kind of ensuring
15:50
that you know what the vendor the platform that uh you’re looking to go
15:55
with has a lot of experience and has a lot of capability that’s um very
15:59
relevant to your use case and the type of organization that you are. A lot of
16:04
vendors kind of want to be everything to everyone. Um but you know at the core of
16:10
it usually they have um you know a set of clients who are within a certain kind
16:17
of you know whether it’s merchant whether it’s financial services. Um it’s
16:21
important that you know your vendor has that experience of working with a client
16:25
that’s similar to you and give you the the type of risk signals that um you’re
16:30
looking for for the type of you know products or the type of organization
16:34
that you are. Um and another is um the endtoend solutions that a lot of
16:42
vendors provide. Um there’s a ton of value in getting a solution that can
16:48
give you, you know, what you need in a box. Um you know, it doesn’t require
16:52
those really complex data integrations across, you know, your cyber detection,
16:56
your transaction detection, um your case management. Um I I see a lot of
17:02
organizations get stuck with um tech debt um where they’re piecing together
17:07
these different systems where you know they know this one is the best for this,
17:11
this is the best for this, this is the best for this. Um but getting those all
17:17
into one kind of platform provides a ton of value. Um, even if you know you you
17:25
have systems that you’re already using for case management, it’s important when
17:29
you’re looking for new detection vendors for example to consider the kind of case
17:33
management solution that they have and it may kind of you know reduce the um
17:39
the lifts of that implementation as you’re not having to do that kind of
17:43
back-end integration across systems and it can reduce your kind of ongoing
17:48
operating costs um with that system. Um, and I guess one last thing, and this
17:55
goes back to your your vendor understanding your use case is when
18:01
you’re using these different risk decisioning platforms, um, oftentimes
18:05
you have to call that platform to get the signals um, for a specific risk
18:11
moment. So, a customer is opening an account, a customer is making a
18:15
transaction, or they’re clicking through, you know, kind of different
18:18
steps of your platform along that like general transaction journey or customer
18:23
interaction journey. Um, but it costs money to call those different systems to
18:29
bring all of those different data points together into your risk decisioning. So
18:34
it’s important at implementation and even on an ongoing basis to you know be
18:40
really strong with where you’re leveraging those systems where those um
18:45
different systems are pulling data from those customer interaction points and
18:50
where you’re bringing in those signals together uh in order to make that final
18:55
risk decision because those those system API calls can be a pretty big cost
19:00
driver. So you want to be sure at implementation and on an ongoing basis
19:06
that you’re really getting the ROI that you want off of those system calls
19:11
because certain risk moments, you know, don’t have the same risk as others.
19:15
Transaction and onboarding obviously are two big ones, but once you get in other
19:19
parts of the customer journey, um you want to make sure that um you know
19:24
you’re really using those systems to their full potential.
19:29
>> Makes sense. I’d say one one PSA for folks out there um yeah certainly when
19:34
you look for a fraud tool and and I did this when I was looking at my own
19:38
solution providers is they should have a very clear understanding and so should
19:42
you about what are the positive business outcomes that you want to achieve and
19:46
how does let’s say an internal tool or an external tool that you bring on serve
19:51
that purpose and how much is it going to to move the needle um Andrew out of
19:56
curiosity of the clients that you work with how many systems or how many tools
20:01
does a typical client have when it comes to to fraud prevention? Like sure, we we
20:07
might have we might desire to have the one-stop shop as a client where it’s
20:10
like, oh, it’s only a single pane of glass. We have one tool and it does
20:13
everything, but that’s a a panacea to some extent where like you probably
20:18
aren’t just depending on a single vendor for everything and there’s even risk
20:22
exposure in that. But was curious like do you have a take on how many tools or
20:26
systems that a client might use here? Well, I could probably go on for 10
20:33
minutes on this. It varies so widely um depending on industry and the type of
20:37
client. I mean, some they’ll give us their architecture and it’s a big
20:41
spaghetti chart of, you know, 15 different systems that vary by product,
20:46
they vary by customer journey type. If we talk about financial institutions,
20:50
they have in branch, they have contact center, you know, you can, you know,
20:55
submit certain kinds of transactions like a wire over the phone talking to an
20:59
agent. Sometimes detection is completely blind to that transaction
21:05
um versus a customer doing it in branch or doing it um through their online
21:10
banking app. Um what we see a lot is fintexs kind of have an advantage here
21:15
because they’re often a lot newer and technology has progressed a lot even
21:19
over the past 10 years once fintex became kind of a bigger and bigger
21:24
industry. Um but a lot of the consolidation of systems are using one
21:29
platform for many different things. We see that a lot with fintex um merchants
21:35
as well. a lot of um larger merchants uh merchants kind of have the benefit
21:41
sometimes of that agility because you know unlike a financial institution they
21:45
don’t have 10 different kinds of products. Um they have a platform that
21:50
they’re in control of. They can get really good cyber signals from customers
21:54
interacting with that platform. Um and they kind of you know put all those
22:00
different data points together through a single solution. um and you know make
22:06
more effective risk decisions. That’s why we often see you know much better
22:10
fraud performance um from merchants uh than we do with financial institutions.
22:16
>> Makes sense.
22:17
>> Yeah. you know I and I you know as you were you were commenting Andrew you know
22:22
I thought how I’ve been lucky I’ve made a so far a long career of of working for
22:27
software companies and I’ve bought a lot of software and so I was thinking about
22:32
it like almost from a checklist perspective like how do I think about it
22:36
uh and thinking also for specifically this kind of platform so it’s in the
22:41
flow of money like it’s in the flow of the transaction so it has to be real
22:44
time predictive self-learning because you can’t at this volume this
22:50
scale you can’t go in and fine-tune real time and it has to have an adaptive risk
22:55
scoring mechanism and all of that I’d suggest requires an AI based platform
23:01
that uses machine learning. This is just at a volume that is it can’t can’t be
23:05
managed real time in an optimal sense. So that’s like number one and then like
23:11
like many pieces of software you got to have robust integrations and you were
23:14
talking about that Andrew and here it’s payment platform CRM your legacy
23:18
financial systems probably you have to support authentication tools it’s a it’s
23:24
a required item uh you also critically many software systems but this one as
23:29
well you have to have great visualization and reporting right so
23:33
that means drill drill down capabilities audit trail generation and I’ll mention
23:38
the last one which is also general. Get great references. There’s maybe multiple
23:43
providers of what you’re looking for, but get great references that are
23:47
relevant to you. If you’re a quick service restaurant, don’t talk to an eye
23:50
gaming company and vice versa, right? Demand and obtain and follow up with
23:55
great references. Uh so more of a checklist type approach, but I wanted to
24:01
wanted to mention that.
24:03
>> Makes sense. All right. Uh let’s go with uh can you share some specific examples
24:09
of how fraud prevention technologies have helped your organization reduce
24:14
costs and improve margins?
24:16
>> So I can talk about that our structure is a little different than most. You
24:20
know majority of our restaurants are company owned but that doesn’t mean that
24:23
we run their P&Ls. So we have what’s called managing partners which own their
24:27
each store is their business and so they have their own P&L. They’re looking at
24:31
expenses. they’re looking at a P&L on a on a granular scheme like we’re doing
24:35
it, you know, companywide. Um, and so those chargebacks hit their bottom line.
24:40
I mean, that’s an uncollectible that is affecting their bonus, that’s affecting
24:43
their pay. Um, you know, that’s that was a huge hit to them. And when that
24:48
continued to uptick, we continued to hear more noise about how what are we
24:51
doing to prevent this? And Steve, to your point, references how we
24:55
implemented the solution that we have was because we talked to someone who was
24:58
in a restaurant environment and said, who are you using? how does it work? And
25:02
that’s what led us to you guys, you know, and that goes back to, you know,
25:06
to SIFT and to um the reference that we had from one of our actually
25:11
competitors, but you know, we have a great relationship with them. Um and so
25:15
our structure is different in the fact that those MPs own that P&L. So, how can
25:19
we help them eliminate those chargebacks in their store and get that money back
25:23
in their pocket? Um and so that really led us to looking for a solution. um
25:28
just we went live honestly in P9 with all of our stores. So we’re a slow and
25:32
steady type uh company. We don’t we don’t just go out the gate with
25:37
everybody. So we started piloting stores in P4. Um we went with our top 10 that
25:42
had the highest chargebacks across our company. Um implemented it. We also
25:46
wanted something that um didn’t affect operations. So we wanted it to where the
25:50
guests didn’t fill it, our operators didn’t fill it. So how does that look?
25:54
Um and you know SIP was the best solution for that. So, um, piloted in
25:58
April for a couple months and then went live full across our 650 locations in
26:03
P9. Um, I will am pleased to tell you that as of P11, which ended last week,
26:08
we um saw the lowest number of card notresent chargebacks that we had seen
26:12
since P1 of 2023. Um, so and all that was, you know, of course that took
26:18
integration and it took, you know, um,
26:20
>> a small lift from our IT team, but back to Andrew’s point, like we found a
26:25
solution that was already built in with our platforms. Like it wasn’t something
26:28
that was a huge lift for us or for our IT partners. And that was um, how can we
26:33
get our results quickly? How can we get this money back in the pocket of our
26:36
operators and our MPs who is the reason that I have a job? um and what can we do
26:41
to keep sales where they where they currently are. So we you know it was a
26:45
kind of a complex structure of we don’t want to affect sales but we want to
26:49
decline the sales that are brought.
26:54
>> Awesome. Thank you for that Ashley and great to see those those initial results
26:57
come out.
26:58
>> Yeah, we’re really excited to see 2025. So our little internal team is like okay
27:02
let’s get a full year under our belt and see what that looks like. For sure.
27:06
>> Excellent. Anyone else?
27:10
Yeah, I guess just a quick response to this one. And you know the thing with
27:15
new fraud prevention technologies um implementing something like that um
27:21
there’s a lot of costs involved. There’s a lot of internal time that has to get
27:25
spent um on on those types of projects. Um taking people away from their day
27:30
jobs, bringing in new people who are able to work with those platforms. Um,
27:35
but it’s also important to consider on the other side kind of the operational
27:40
efficiencies that’ll be gained with those platforms. If you’re getting a new
27:44
case management system, for example, you know, the level of time that um is going
27:49
to be saved um for agents to be working alerts um to be conducting customer
27:55
outreach. Um and then when it comes to the more sophisticated detection
28:00
platforms and you know the AI models kind of even the self-learning fraud
28:05
strategies that um some platforms offer you know allow you to to test strategies
28:11
in real time that can actually learn based on the patterns that they’re
28:15
seeing in the data. you can actually run, you know, a much lighter data
28:19
science team, a much lighter strategy team uh when working with those systems
28:24
and especially with platforms that have that different um you know the different
28:31
signals that exist usually in the past would be in a silo like your cyber
28:37
detection systems and your transaction detection systems. you’d have a whole
28:40
cyber team, you would have a whole transactional
28:44
um strategy team. But now with newer platforms, you’re able to bring those
28:49
teams together and have a much lighter staff that um you know can kind of do
28:55
both of those things and um manage the different strategies that are behind
28:59
those platforms.
29:02
>> Yeah, I think that’s great advice. Um, I know like everybody it’s a team or a
29:08
company that’s trying to be financially prudent, you always want to kind of trim
29:13
the fat and consolidate where you can, whether it’s tools and and frankly teams
29:16
too. Like one of our most important assets are our people and they can be
29:20
quite expensive uh to retain and grow and where you can how can technology
29:25
step in uh and either take the place of some human teams but also just from an
29:30
efficiency standpoint um enable them to work on I’ll call it the grayest of the
29:35
gray cases here and on the fringe whether that you know the vast majority
29:39
just to be realistic here the vast majority of transactions or accounts
29:43
going through the system are legit like 99 plus% of them are legit But you just
29:47
have that sub 1% that is trying to do a disproportionate amount of damage in
29:53
that ecosystem. So even Ashley mentioned maybe it’s in particular markets. It’s
29:57
not just it’s not Kentucky. It’s it’s some store in uh Nevada that that might
30:01
be a hot spot. Uh and so having technology and tools and teams being
30:06
able to take a scalpel to the problem um is very effective, but you don’t you
30:11
can’t necessarily deploy a team to every state or every store. Um, so being able
30:15
to consolidate that with technology ideally, um, is certainly critical.
30:20
>> Well, and Kevin, to that point, if you’re like us, we weren’t familiar with
30:23
fraud.
30:24
>> So, we weren’t the experts. So, how do you leverage, you know, a team that can
30:28
help educate you to be an expert in fraud when you didn’t, you know, have
30:32
that experience previously? That’s a great question because and
30:36
that’s a great thought because a lot of the companies certainly that I’ve worked
30:40
at or uh work with it does run the gamut here in terms of do they have a
30:46
dedicated fraud team with data scientists engineers operations etc to
30:52
that some of some large companies hey if they believe fraud is part of their core
30:56
DNA yes they’re going to invest in that that core team or maybe if they have
31:00
multiple big products each big product will have their their separate team. Um,
31:05
but on the other end of the spectrum, hey, if risk and fraud isn’t core to
31:09
your business, from an operations efficiency standpoint, why would you
31:12
build that team completely internally? Is there an opportunity to rely on kind
31:17
of thirdparty professionals to do it and then kind of like not quite set it and
31:22
forget it, but give them an opportunity to to run with it so you can focus on
31:25
more of your key objectives for your business
31:29
>> and help guide you to make those decisions, you know, through that that
31:31
lens of education as well. Yeah. Yeah, Andrew, question for you on this one. In
31:37
terms of the clients that you work with, what would you say their
31:42
broad maturity is? Um, and is it differ by industry or by kind of different
31:48
parts of the world or region? Um, do you have a particular take on that one?
31:53
>> Um, it’s usually a function of volume actually. So I mean we we work with you
32:01
know institutions or organizations that are more regional. We work with you know
32:05
much larger ones that are global and organizations that are in between. Um,
32:11
so yeah, like I said, it’s it’s usually a function of volume because the volume
32:16
kind of dictates the amount of fraud that you’re seeing and the amount of
32:20
fraud that you’re seeing kind of dictates the amount of money that uh the
32:24
organization is willing to invest in the uh the fraud program. Um, so yeah,
32:30
generally that would be I’m trying to think to of exceptions to that rule. Um,
32:35
I could bring up FinTech again because, you know, they kind of have that agility
32:41
starting out, um, you know, not that long ago for a lot of them and they’re
32:45
able to bring in those new really mature technologies and get all of those
32:49
systems really talking to each other well. Um, but whenever we talk about,
32:53
you know, old older organizations, usually it is a function of um, of how
32:58
much um, volume that they’re getting transaction wise. And then I guess
33:04
downstream how much pain that’s causing in terms of financial loss.
33:08
>> Yeah, especially.
33:09
>> All right. Um let’s move on here and want to talk a little bit about um kind
33:14
of we’re talking about AI. Let’s talk a little bit more about AI powered risk
33:18
decisioning and how does it play in enhancing operational efficiency um and
33:23
customer retention.
33:25
>> Uh yeah, I’ll jump in first. Um, we think about operational efficiency. The
33:31
scale of of what a human can can handle effectively, accurately, efficiently is
33:39
dwarfed by what a platform can do. Not again the word panacea, one of my
33:44
favorites. Someone used it earlier. Not that you drop in a AI platform and
33:47
everything’s fine. But that’s part of how you became become operationally
33:51
efficient. It’s just the scale and volume. It’s just it’s just not possible
33:56
for a human or a team of humans to do this in the real time fashion in an
34:00
effective manner needed. And so while that’s not free, the cost of humans
34:06
really adds up as you attempt to address it with humans. So AI just allows the
34:10
use of many more signals than possible for humans. You know, thousands, you
34:14
know, tens of thousands more, not just the types of signals, but then the
34:17
volume of the signals. AI is the only way to run real time risk decisioning
34:23
and and so from an operational efficiency perspective once you achieve
34:27
scale or get anywhere close to it or planning to a platform AI enabled to
34:32
achieve the benefits makes much more sense and then also you can I think the
34:38
data is clear you’re able to prevent false positives false negatives you can
34:42
then transition not just from payments but transition that over to uh account
34:47
takeover and being able to prevent that and then that gets to customer
34:50
retention. If they if customers are not if they are subject to account takeover
34:55
or ret or or you know a false positive you know that negatively impacts their
35:00
experience. If they aren’t they’re more likely to stay. And it’s clear also
35:04
friction is a is is plays a role as well. But if you’re able to provide a
35:08
better customer experience for the consumer, like in the case of we we
35:12
offer a platform, the consumer has a better experience that enhances the
35:16
customer retention for parties who are our customer. And so you want to be able
35:20
to cap capture both the operational efficiency as well as the great
35:25
experience for the customer through all those means to drive attention of the of
35:29
the consumer.
35:32
>> Makes sense. Anyone else have a take on this one?
35:38
Yeah, I can I can add in here and I think uh for fear of just mirroring what
35:44
Steve said in a different way because I think he nailed it. Um but I’ll break
35:49
this up. You know, talking about enhancing operational efficiency and
35:54
customer retention. I mean there’s a ton of value that you can get um you know
35:59
operationwise um with you know being smarter about um
36:04
your fraud detection. You know what actually meets that threshold to make a
36:09
decision to create an alert that is going to have to be handled by
36:14
operations staff. Um because that’s a lot of that is what leads to customer
36:19
friction and has a strong impact on customer retention. Um, but you know, we
36:24
think of things like even just soft alerts, you know, where you’re not
36:28
actually impacting the customer at the transaction, but you’re still um, you
36:33
know, reaching out to them. Um, letting them know that something suspicious,
36:37
excuse me, suspicious happened on their account. Um, that leads to more inbound
36:42
call volume. Um, and then that can lead to, you know, customers using your
36:48
product or your platform less because, you know, once once they hear the word
36:53
fraud, it often scares off customers. A lot of people don’t really experience
36:58
fraud very much in their lives. Um, and so whenever, you know, they’re getting
37:03
reached out to by their financial institution or or by a merchant, um, it
37:09
kind of makes them second guessess, you know, using that product or that
37:12
platform. Even though you know everything is there for their protection
37:16
um some customers can be encouraged by it. Um but some customers you know kind
37:22
of get scared off and they think okay is this platform not very secure? You know
37:26
why are they so confirmed or so concerned about my transaction um being
37:32
fraud? Um and then you know customer retention
37:36
just drilling in a little bit more there. um kind of the level of friction
37:42
that you introduce um for things like 3DS. Um there are a lot of AI based
37:48
models that kind of determine um what transactions get sent through 3DS. Um
37:54
and then even on the financial institution side, the um the AI based
37:59
models that are making the decision on what gets challenged. Um the challenges
38:04
um add extra friction for the customer because they have to enter in a onetime
38:08
passcode. Um, so it it creates a lot of um different pain points for the
38:15
customer. Um, some of them are, you know, they never seen a 3DS challenge
38:20
before. So to them, it looks like a popup screen that, you know, is my
38:24
computer being hacked. This is I’ve never had to enter a one-time passcode
38:28
for a transaction before. Um, especially in the US. I mean, we see that a lot in
38:32
Europe. 3DS has been around for a very long time, very widely used. Um but on
38:37
the US side it um you know is a little bit um more nient. A lot of big
38:44
merchants are slowly moving towards it. Um so anyway friction is a big part of
38:49
customer retention and um it’s really important to to consider where you’re
38:54
introducing friction because it may um affect your customer um you know using
39:01
your uh products or your platforms. Again,
39:04
>> Andrew, I think you bring up a great point there where when it comes to in
39:09
and many businesses, even within the same industry that I’ve worked either
39:13
worked at or spoken to, they have different takes on when to introduce
39:19
friction and how much of it is for educational purposes. Like we’re doing
39:23
this for the end consumer’s safety and security. like we want to educate them
39:27
on how to uh use the the product uh properly, but then on the same maybe on
39:33
the flip side here um we don’t want to freak them out like and if we freak them
39:37
out that means they’re going to use our product less and they think something’s
39:40
wrong. And so there is that fine balance between freaking out and education of
39:45
like oh this is actually a security blanket for you.
39:48
um some end consumers uh actually feel assured like oh I’m glad my financial
39:53
institution is asking me about um more security or even when let’s say you take
39:58
a lift or an Uber around like in the app now uh for safety reasons like they’ll
40:03
they’ll notify you of like oh looks like you got dropped off like further than
40:07
the expected endpoint here is this correct and those are all safety issues
40:11
um that ultimately they made the decision like we think this is a net
40:15
positive for end consumers it’s going to make feel uh more safe on the platform.
40:20
>> Well, and to that point, Kevin, preventing fraud is brand protection,
40:24
too. I mean, with that’s what we’re talking about with if you’re preventing
40:26
that customer’s fraud from happening, then you’re protecting your brand. And
40:30
so, you know, your common customer doesn’t know that their account
40:34
credentials were sold on the dark deep web and they just happened to use it at
40:37
Texas Roadhouse to buy a steak. you know, in their opinion, if they had a
40:40
fraudulent transaction at Texas Roadhouse, then Texas Roadhouse was
40:43
hacked or, you know, their information was stolen by Texas Roadhouse. And so,
40:48
by preventing that fraud or preventing that chargeback from even occurring on
40:51
their account, you’re also protecting your brand and the reputation, which
40:54
then in results allows for that customer loyalty too to continue.
40:58
>> Great point. Yeah, I cannot tell you how many times
41:01
back in the day I used when I worked at Google, I took phone calls uh and emails
41:06
from people saying, “Hey, I see a $500 charge on my credit card for Google
41:11
advertising. What is that?” And I’d have to explain to them like, “Looks like
41:14
your credit card was compromised. It was used on our platform uh to do, you know,
41:18
AB and C things.” And I hated taking those calls. Um which actually drove me
41:23
to want to take care of it. and won an opportunity on the fraud team like came
41:27
up. I ended up taking that role and you know two decades later I’m I’m still
41:32
still doing it.
41:33
>> Um awesome.
41:35
>> That’s back to that education piece too of you know your your common customer
41:38
they hear fraud and it’s scary. They don’t understand that it didn’t just
41:41
happen by dipping or swiping your card somewhere like your information is in
41:45
other places.
41:47
>> Exactly. And that that can be hard to say hard for someone to stomach because
41:51
in that education vein
41:54
>> often times it’s hard for consumers to point the finger at themselves of like,
41:57
oh yeah, I probably shouldn’t use the same password across every single
42:01
account or maybe I shouldn’t share my Netflix account across like all these
42:04
different people uh and and have it be the same password for my
42:08
>> bank account. Yeah. Exactly.
42:11
>> Right.
42:12
>> All right. So, let’s kind of move forward here. um want to talk a little
42:15
bit about can from a finance leader perspective like how can leaders turn
42:20
risk into revenue um by addressing kind of I’ll call it hidden leakage and
42:26
reducing fraud response costs I think we talked about that you know
42:30
just I mean that’s a natural segue into what we just talked about and that’s
42:33
creating that customer loyalty and you know uh our mission statement is
42:37
legendary food legendary service preventing that friction for them on
42:40
that service piece which is buying our food preventing friction on their online
42:45
purchases but also at the same time preventing that fraud for the guest so
42:49
that they will come back and dine with us again whether it be online or in
42:52
store you know so I don’t think you realize that yeah it may be a charge
42:56
back to their account and it hits their store operator’s P&L but in the endgame
43:01
are they going to come back and dine with you again so how do you get those
43:03
people continually dining with you even if they’ve experienced a fraudulent
43:08
transaction at your institution
43:12
>> absolutely
43:13
>> and you Also, this this is the cross functional game, right? It’s not like
43:16
one person’s responsible for this across your company, but you can remember as
43:20
you work with your team helping to put together and and optimize your fraud
43:24
platform. Make sure those chargebacks get back into the system. That
43:28
self-learning aspect, uh you get your actual chargebacks, make sure they go
43:32
back in, it results in a better answer going forward. So, that’s part of the
43:36
playbook is a self-learning piece and that’s a key piece of data to feed back
43:39
in to the to the platform. And that’s yeah eliminating those silo silos across
43:44
your organization too. I mean one of our strongest partners in the treasury team
43:48
is it and you wouldn’t think that you wouldn’t think we go together but we do
43:52
you know and it’s you know relying on their education and their information as
43:55
well as you know what we can give from the finance side to help you know break
43:59
those silos down so that we are attacking fraud to the best of our
44:02
abilities.
44:03
>> Yeah. Amen.
44:04
>> Makes sense.
44:05
>> Yeah. And sharing that fraud truth data across silos too. I mean, you see a bad
44:10
transaction, was there an account opening that happened two weeks before
44:14
from, you know, the same device or whatever it is, and tying those
44:19
together. So now when you’re detecting bad transaction openings or bad account
44:24
openings, you know, you know, what accounts had bad transactions further
44:28
downstream. So sharing that fraud truth data across your different kind of risk
44:33
points is um really important, too.
44:36
>> Yeah. I mean, I’d say one kind of key learner that I’ve had is when it comes
44:41
to I call it high volume high volume digestible data ingestion. Kind of a
44:46
mouthful there, but what I mean by that is we talked about silos a little bit
44:50
earlier. So many teams I talk with don’t have access to data within their own
44:57
company. So, it’s not about even a third party, but let’s say you’re on the
45:00
treasury team and you want to understand the the lifetime value of a customer. Um
45:05
because it’s one thing to get, let’s say, one a $100 chargeback and you let’s
45:09
say you lose it, that’s a hundred bucks. Um but if you are wrong, let’s say, and
45:13
you create a false positive and you insult the end user and say we’re
45:16
blocking you, but they in fact are a legitimate user.
45:19
>> What’s the customer lifetime value there? And oh gez, has this person dined
45:23
at our business 10 times? And their their lifetime value might be $1,000.
45:29
Um, maybe that $100 uh charge app that came
45:33
in, maybe there was something wrong with the order or maybe there’s something
45:37
like should we take the the $100 hit or the lifetime value gain of retaining
45:42
that customer. And so those are tricky questions and from a data perspective,
45:47
many teams because they can work in silos don’t always have that visibility
45:51
to make the best decision possible. Yeah, I could go on forever about that,
45:55
Kevin, because we’ve recently created a CDP or a customer data profile based on
45:59
the tokenization data that we’re getting from from the treasury perspective and
46:03
it is using it to track the longevity and the life of our customers and their
46:08
their purchasing habits and their frequency. So, it’s internal information
46:11
if we know if they’re going across brands within our, you know, our three
46:14
brands that we have or if they’re purchasing online or in store. But that
46:18
was data that they were able to utilize that’s provided to us. So without that,
46:22
you know, if we were in that silo, that CDP wouldn’t have been as successful as
46:26
it could be because they wouldn’t have the transactional data that was coming
46:29
from the Treasury team.
46:31
>> Makes sense. And going off topic a little bit and we can definitely come
46:36
back to this question as well, but I’m curious to understand like as finance
46:40
professionals in this space, what’s something that you know now that you
46:45
wish you knew let’s say two years ago or like even six months ago um when it
46:51
comes to our business and fraud and like if we had just set up the system this
46:57
way or if we did this other thing early on and we didn’t make that trade-off per
47:00
se, um we’d be in much better spot. Um, as a result,
47:06
>> I think it goes back to co for us. Um, if we had known what that would have
47:10
brought and that we did have that market out there for to go and online sales
47:14
that we weren’t really tapping into, we could have prevented fraud from the
47:17
beginning. Um, you know, it’s but we didn’t know. We didn’t know what we
47:20
didn’t know at that point. Um, so that would be, you know, that’s like what
47:24
keeps you up at night? What’s your biggest regret? You know, all those
47:26
questions. Okay, a credit card breach, a hack. Oh, we should have been preventing
47:30
fraud four years ago. But you know we we weren’t in that space then you know so
47:35
if you know I could have seen the future in 2019 we would have implemented a
47:38
solution before before co
47:41
>> well even back then it sounds like it was a 3% space and now it’s a
47:44
significantly bigger space and so it was a market we didn’t even know we had we
47:49
had room in yeah for sure
47:52
>> yeah and you mentioned technical debt earlier a common a common problem or
47:57
mistake is people just don’t think broad as deeply and broadly about a solution
48:01
they’re putting is think about the data points you want to gather. What are
48:05
those data points? Where do they sit? Get them into your platform. Whatever
48:09
your central repository is. We use Snowflake. Many companies do. What are
48:13
the data points? How do you get them into a central place consistently,
48:17
quickly, and accurately? So, you spend less time gathering and scrubbing, more
48:21
time on the analysis. And it’s it’s easy to say, think about all your data points
48:25
and get them all up front. But the more thoughtful you are about that, it
48:29
doesn’t mean you won’t need to go identify de more data points later, but
48:32
it leads to less uh redoing of work or maybe even in in some ca some cases
48:39
building that after you already start the initiative. So think about that more
48:44
as deeply as you can upfront. Be reasonable about it. It doesn’t mean
48:47
it’s going to be perfect, but you’re going to get a better answer with more
48:50
thought on that upfront.
48:54
>> Awesome. Um, I know we’re coming up on the top of the hour, so maybe last one
48:58
or two questions on on our end here. Um, this one’s a little bit more in the
49:02
weeds. We talked like Andrew, you talked a little bit about 3DS. Maybe it’s more
49:06
popular in other regions of the world or not and rather than the US. Um, but do
49:10
you have a take on how we should leverage 3DS or how different companies
49:14
should leverage 3DS?
49:18
>> Yeah. Um so we’ve seen different approaches um in the market but and I
49:26
guess when I say different approaches you know there are a lot of companies
49:29
that send nothing through 3DS. There are some that once they get you know their
49:34
3DS capabilities up and running they just send everything through 3DS because
49:39
they’re sick of chargebacks. They’re sick of you know the operational costs
49:42
that are involved with dealing with chargebacks. So they’ll just send
49:46
everything through. Um and for folks you know listening 3DS removes the liability
49:52
from the merchant and puts it on the um the financial institution. So it’s you
49:58
know transaction type that you spend a little bit more money to send it through
50:01
the gateway. Um and in return you have no liability in case of fraud. Um but
50:09
the institutions that are more mature have that dynamic decisioning when it
50:14
comes to what they’re routing through 3DS. So basically they’re using AI or ML
50:21
models um to decision on what transactions they want to send through
50:27
the gateway based on risk. um you know obviously you want to send your highest
50:32
risk transactions through um because you won’t have liability for that kind of
50:37
fraud um but you’re also introducing that customer friction that’s inherent
50:44
with 3DS that I talked about before. So with your dynamic decisioning, you’re
50:49
really just trying to find that balance between risk and between customer
50:53
friction and thirdly between the costs involved with sending transactions
50:59
through 3DS.
51:02
>> And from our perspective, Kevin, I mean, I feel like you have to, like we said,
51:05
have those partnerships with it, talk about customer friction. What’s that
51:09
going to look like if you implement 3DS? if your guests aren’t used to seeing
51:12
that and you go all of a sudden from oh, you just placed an order to now you’ve
51:15
got three different stops or one different stop or multiffactor in
51:18
between there like what’s that going to do um to the guest that’s not used to
51:21
seeing it? Um and really leveraging those external partners. So, you know,
51:25
like we said before, we weren’t educated in fraud. That wasn’t our strong suit.
51:29
So using those thirdparty partners to help educate you on what that looks like
51:33
for your business, having someone that deep dives into your organization and
51:37
your culture to understand what that looks like from, you know, a company
51:40
structure to see how this would impact your guest in the endgame.
51:46
>> Makes sense. All right. So last question on on my end um kind of for the group.
51:52
Where do you think broad mention you me and Ashley mentioned structure should
51:56
live within an organization? that this is as crossf functional as it
52:01
gets. You know, um uh it’s it can’t just be one spot. Um it’s super it’s it’s
52:08
just incredibly crossunctional. So everybody’s part of the solution. Um the
52:12
the breath of supporting the customer and also fighting fraud means just about
52:16
everybody’s involved. Uh so there’s owners of key components of of what
52:21
you’re trying to do. And it’s probably more of a working group. There could be
52:25
someone in charge of the working group but a working group uh that helps set
52:29
that up. Um so if someone says I’m taking care of it or says hey that’s a
52:33
responsibility of it finance data platform whatever I I think they need to
52:39
to adjust that view to thinking it’s all of us. We’re all part of the solution.
52:44
>> That’s funny Steve. So we have a task force. We jokingly call oursel the
52:48
fraudbusters. We were presenting to our uh you know
52:52
CTO at the time and he was like well you guys are just busting fraud. You’re just
52:55
fud busters. So we we made t-shirts. It’s a whole thing. But that’s Texas
52:58
Roadhouse culture. But you know it is it is that cross functional team. It’s
53:02
finance. It’s treasury. It’s it you know they can help us see what we’re not
53:06
seeing through their lens and we can help them see what they’re not seeing
53:09
through our finance lens. You know it’s really having that partnership with them
53:12
that you know if they’re making decisions that they think could impact
53:15
us, we’re talking about it. if we’re making decisions that can impact them,
53:19
we’re talking about it, you know, and so um I’m honestly closer with some of
53:23
those people than I am people in accounting and I work in treasury, you
53:25
know, but it’s spend a lot of time together. Um and so it’s really leaning
53:29
in to the knowledge of each other and that crossf functional opportunity
53:33
because like you said, we all own fraud, you know, we all, you know, down to our
53:37
operators and how they’re preventing it from happening in their store up to us
53:40
running the numbers at the end of the month. I mean, we all have to take
53:43
ownership of it.
53:44
>> I love it. Plus, it’s a great opportunity for some swag, right? Like,
53:48
you know, little Ghostbusters type theme or there’s many others, I’m sure. But
53:51
yeah, I love it.
53:52
>> Exactly.
53:54
>> Awesome. Well, yeah. I can tell you like from my own personal experience like as
54:00
part of running different broad teams at different companies at this point I’ve
54:04
reported up to operations, finance, legal, infosc, engineering for a time
54:13
like it runs the gamut and to me the the end story is it’s got to be
54:18
crossunctional and the reason why you have different reorgs and these things
54:21
is because it touches so many parts of the business and so everybody wants to
54:25
have at least a say or kind of a seat at the table when it comes to these things
54:29
because as Steve mentioned, we’re we’re in the flow of money here and we are in
54:33
the spotlight and sometimes you can get burned, sometimes you can shine. Um, but
54:37
that is the the the nature of the beast uh uh to many extent. All right, so
54:43
we’re right about time. Ashley Andrew Se, thank you. Thank you so much for
54:48
taking the time out dropping some knowledge with the group here. Um, I
54:51
wish you all happy holidays. Thank you so much and take care everybody. Bye
54:56
now.
54:57
>> Thanks everybody.