Selecting the Right Digital Fraud Management Partner Featuring Forrester

0:03
Greetings and thank you for joining uh today’s webinar. I’m going to uh share
0:08
my slides right now and uh we’ll launch into today’s session. So today’s session
0:12
is titled selecting the right digital fraud management partner. I am uh joined
0:18
today by uh an esteemed um representative from Forester which I’ll
0:22
introduce more formally in just a minute.
0:27
So I’m Armen Nagarian. I’m the chief marketing officer with Sift. I want to
0:32
thank you for joining today’s session. I’m joined today by Andra Chair who’s
0:37
vice president and principal analyst with Forester. Andra oversees the
0:41
effectively the digital fraud management practice for Forester and also customer
0:45
identity management. Andra, thank you for joining us today.
0:48
>> Thanks for having me. Really greatly appreciate it. Looking forward to the
0:52
session today.
0:53
>> Excellent. We we appreciate it too. So, a little bit of housekeeping here. So we
0:56
plan to uh time the prepared comments for somewhere between 30 and 45 minutes.
1:02
And today’s session will be recorded after the session. You will see the
1:05
recorded asset available on cif.com right where you registered. So go back
1:09
to the same place later on today Pacific time. Um and you will see the asset
1:14
there. Um we will conduct a live Q&A session at the end. Um and you can also
1:20
submit questions at any time using the Q&A feature in your Zoom tray. Um and I
1:27
will be able to see those and we can we can respond to those during either
1:30
during the session or following the session. Um and we will also administer
1:34
two poll questions to keep it uh interactive. So without further ado,
1:38
let’s jump into the first poll question. Poll number one, where are you where is
1:43
your company in your journey to establishing a best-in-class DFM or
1:47
digital fraud management solution? You can see we have four responses. I’ll
1:51
pause for 10 or 15 seconds and allow everyone to uh to vote.
2:08
Excellent. Thank you for participating. I do commit
2:14
to playing back the results as soon as I can see them. and uh we uh we appreciate
2:19
your participation in the interactive polls.
2:22
So today’s formal content will focus on these four subject areas. So first we
2:28
will define what is digital fraud management and what you know what the
2:32
framing is for this very important thematic space uh in the in the
2:36
technology and risk landscape. Second we’ll talk about some of the
2:40
capabilities and and important trends to note. uh we’ll roll into best practices
2:44
and then we’ll talk about some predictions how we see the future
2:47
playing out. During the session you’ll see Andra and
2:51
I um um back and forth and kind of managing it uh more interactively.
2:58
So with that let’s launch into just a little bit of context on digital fraud
3:02
management challenges from a CIF perspective. We’ve been in this business
3:07
for about 12 years. When we look broadly across the industry um it is no surprise
3:12
that there are a number of attack types um across the digital user journey and
3:17
that’s really depicted on the left in the purple and pink tiles. you can just
3:21
see some of the more common attack themes and attack types that we see and
3:26
many others see in the industry um driven through the consumer web and it’s
3:31
and these are always evolving which we’ll talk about um the sophistication
3:35
of attack types um and the proliferation of the attack types is really
3:39
astounding. Meanwhile, you can see on the lower right there are a number of
3:45
teams. There’s a lot of people within the enterprise that care about solving
3:48
this important problem and these teams don’t always work together um in in a
3:55
cohesive and coordinated way. We’ve seen some siloism um well-intended but
4:00
keeping information in separate areas and so that’s been a challenge to
4:04
adequately addressing and properly addressing the task at hand which is to
4:08
reduce the digital risk and deliver a better customer experience.
4:12
And then finally there is no lack of solutions and tools in the marketplace.
4:18
Um there are different business models um you know technologydriven services
4:23
driven um different pricing structures um different sets of capabilities and so
4:28
we appreciate that. We understand that there is a lot of choice in the
4:31
marketplace. Uh and that’s adding to certainly the confusion and the
4:36
important choices that you need to make as enterprise stewards of your risk
4:41
management and fraud management capabilities. So when you put all that
4:44
together, there are certainly um challenges to be aware of. Meanwhile,
4:49
from the data, so this data is from SIFT primary research. So every quarter we
4:54
conduct a um a a an extensive quantitative study through the digital
5:00
trust and safety report that we publish. We also supplement that with insights
5:05
from our own data network and and other polls. And so you’re seeing here results
5:09
from those bodies of work. Um, and on the far left, it’s just stunning. Like
5:14
the year-over-year for second quarter 22 versus second quarter 23, a massive
5:20
increase, 354% increase in account takeover attacks as evidenced by what we
5:25
see through our own data, through the own attacks that we’re tracking in our
5:29
own network with the uh hundreds and hundreds of customers across our
5:33
network. It’s just been astounding. Um meanwhile from the consumer research
5:38
we conducted 76% of consumers would just give up on a brand um if they have an
5:44
experience that is unfavorable. Um if there was a fraud incident that they
5:50
were a party to or or a victim of they have indicated the consumer that they
5:55
would simply walk away and try and shift to a competitor. So abandonment can be
6:01
very high and switching can be very high if there’s that bad experience from a
6:05
consumer perspective. And then finally we can see on the right 60% of consumers
6:10
admit to committing fraud and knowing someone who has. So here’s where we get
6:14
into friendly fraud, first party fraud where uh maybe the intent wasn’t
6:19
fraudulent from the beginning but someone is acting in a way that um is
6:24
not in the best interest of the business. And so this is a very
6:27
interesting insight that this preponderance of you know friendly fraud
6:30
is perhaps higher than anyone thought. So again when you kind of look at these
6:34
data points combined with the last um piece of context in the last slide the
6:39
the problem is real uh and the need to address with a proper digital fraud
6:45
management posture is more important now than ever before.
6:49
With that, let’s transition into some points of view from Andra on what is dig
6:54
digital fraud management.
6:56
>> Thanks, Armen. Yeah. So, so these problems are definitely, you know, uh
7:01
top of mind for a lot of folks, right? So we define right this is is this is
7:06
basically the text uh the threeliner here in the slide uh as digital fraud
7:12
management as an endto-end commercial offtheshelf optionally orchestrated
7:16
solution that allows for preventing detecting and intercepting account
7:19
takeover and payment and non-payment fraud detection and that provides tools
7:23
and automation for fraud investigation retail chargeback management reporting
7:27
and customer authentication. That’s quite a mouthful, right? So, so what
7:31
does this thing do? Um, basically it allows organizations to reduce financial
7:37
fraud. So, cost of fraud as Armen talked about is is basically something that you
7:42
know has been skyrocketing. Uh, number of different forms of fraud, you know,
7:47
payment and non-payment fraud, the payment fraud types uh are are
7:52
definitely there. So, there’s definitely the reducing non non-financial fraud
7:57
aspect that’s there. um reducing the fraud investigation cost, right? That’s
8:02
another interesting and important element here. You know, it’s all good
8:06
that you’re able to contain your fraud, you know, losses, right? Payment or
8:10
non-payment scams, policy abuse, but you cannot hire an army of investigators,
8:16
right? you have to contain the number of people who uh investigate and analyze
8:21
you know fraudulent activities as well as the basically the IT resources that
8:26
manage these solutions and the last point here is is improve the customer
8:31
experience right as we’ve seen you know a 70 plus percent of of people would
8:36
move to a complete different you know provider retailer e-commerce commerce
8:41
provider media channel whatever they have if they’re a victim of fraudulent
8:45
activity video, but I would even add if people cannot get their thing done, the
8:50
reason they came to the website, you know, they’re going to be just quickly
8:54
quickly moving someplace else. So, so this is definitely an important element.
8:58
And if we move to the next slide, please um it is clear right that this is not
9:05
just a an online um story, right? So uh digital fraud management is always going
9:13
to be a combination of a number of different steps and stages and protect
9:19
and covering a number of steps and stages in the customer journey. Right?
9:22
So identity verification is definitely going to be um an interesting step. the
9:28
the more an organization is sure about the authenticity and the re realness of
9:35
an identity that signs up um the lower the risk of fraud is going to be. So
9:41
basically if you’re able to do identity verification even if it’s simplistic
9:45
that this is not Mickey Mouse signing up but the by but the real Andrash or Armen
9:49
for buying products or services your your risk of fraudulent activity is
9:55
going to be much lower. So that’s number one. Number two is risk scoring, right?
9:59
This is basically looking at payment and non-payment transactions, assigning some
10:03
kind of risk score to that that transaction and being able to make a
10:08
decision whether we’re going to let that activity and and and action happen or
10:13
not, right? Or we want to maybe investigate it manually. So risk scoring
10:17
is definitely the heart of the matter here. We have you know all solutions
10:22
need to be able to provide some kind of a score some kind of an assessment of
10:26
risks around a certain transaction or activity.
10:30
Uh definitely customer authentication method and authentication method
10:35
integration is important right so you have to be able to log people in and and
10:40
basically monitor that activity that is definitely part of the process. uh case
10:47
management if there are suspicious activity uh cases or transactions that
10:52
are being detected for uh and they need to be investigated manually by a fraud
10:58
specialist at a retailer. There has to be a tool to do this, right? You and
11:02
that and a tool that is able to show uh connotations, linkages between certain
11:08
activities, entities visually and nonvisisually. And then there’s always
11:12
reporting, right? You definitely want to be able to understand um you know your
11:17
trends uh and and really uh activities risky activities risky users uh types or
11:24
or areas of of your merchandise or or the items that you sell that are more
11:29
prone to risk or more prone to fraudulent activities. So reporting and
11:34
analysis of of fraud data actually can can help uh the organization. So if we
11:41
move on to the next slide right there’s definitely you know not all um you know
11:46
products are created equal in this thing right so we definitely see that this
11:52
conversation is evolving from just a pure account takeover or atto detection
11:58
and bot management bot activities or detecting robotic activity in the
12:02
website um to basically people wanting an end toend solution and not cobble
12:09
together patchworker tools, right? You definitely have to be able to work with
12:13
a with a tool, a solution that is just not something that’s been inhouse
12:18
developed or or you know kind of been uh cobbled together, welded together from a
12:24
number of different components. Um purpose-built chargeback management for
12:28
merchants is definitely been an area where we’ve seen uh a lot of uh
12:33
different differences between offerings in our assessment of these products,
12:37
right? Um, chargebacks are a very critical part of of any fraud management
12:44
set of processes, right? Being able for the merchant to uh properly prepare for
12:50
fighting chargebacks is definitely part of the part of the conversation set of
12:54
requirements flexibility. So supporting various authentication policies such as
13:00
passwordless authentication is definitely important. As I said, you can
13:04
definitely bring in partner products here, but uh the more the solution, the
13:09
more your DFM solution supports authentication, the less you have to
13:13
reach out to, you know, yet more vendors out there. And then configurable
13:18
investigator friendly case management is definitely an important element here. Uh
13:22
don’t forget right that the investigators for fraud fraud
13:26
investigators fraud analysts are spending considerable amounts of time um
13:32
in in this whole universe right on a few screens right where they just look at
13:38
you know transactions that have been ris you know risor high by the DFM solution
13:43
and these folks are trying to kind of make sense of these transactions and be
13:48
able to investigate and understand them and make the final decision whether to
13:52
you know pass them or stop them and and report fraud. So these tools have to be
13:58
really good and usable for these fraud investigators to be able to produce good
14:03
results in in fraud fight. So as we move to the next slide, we want to talk a bit
14:08
about key capabilities right uh in our uh
14:13
in our assessment of the of of DFM solutions. And on the next slide, first
14:19
of all, we want to talk about market changes, right? So, we’ve seen a lot of
14:23
vendors investing in in workflow configuration improvements, right? This
14:27
is realizing that every retailer’s business setup and how they support a
14:33
customer journey is going to be different, right? there’s it’s it’s not
14:36
basically uh just you know wanting to score things or wanting to case manage
14:42
things or it’s it’s it’s basically a lot more uh elements a lot more data sources
14:47
that you might want to bring in. So orchestration is absolutely an an
14:51
important investment that we’re seeing vendors uh making that really allows uh
14:56
retailers to bring in a much broader variety of data points into their
15:01
decision- making and investigation processes.
15:04
We see artificial intelligence or AI uh for you know unsurprisingly for a lot of
15:10
things right risk scoring um so making sure that you have an assessment of of
15:15
of the risk level of a transaction. Investigation is definitely you know
15:20
what you want to investigate and how you want to guide the investigator through
15:24
the investigation process is absolutely important and even reporting right uh
15:29
how you formulate and present the data can can be impacted by AI uh we’re also
15:36
seeing improved jur jurisdictional compliance right um a lot of times
15:40
certain elements in fraud management such as device ID or IP address may not
15:45
be acceptable uh for fraud even fraud fraud management purposes in and
15:51
sometimes GDPR you know in Europe the privacy regulations actually limit the
15:55
usability of various attribute types in in certain jurisdictions. So tools have
16:01
to be able to support that and be able to tune and configure accordingly.
16:07
And then lastly the coverage of first party and friendly fraud and scams.
16:12
Right? These are activities are really uh one of you know are some of the
16:17
hardest things to detect right because they’re not necessarily related to
16:20
payments. Um I’m going to talk a little bit about these activities but uh first
16:25
party um and and friendly fraud are really mainly about things like um good
16:32
customers. So so hither to good customers starting to claim thanks and
16:36
do bad things. So, for example, somebody buys a a valuable item, they receive it,
16:42
and then they claim that somebody else did it or they never received that item
16:46
or they they return the item, but they return an empty box or full of stones or
16:52
or just just full of dry ice, right, to a wrong address that really triggers the
16:57
refund process. And so, they basically get the item for free. Um, so so this is
17:01
definitely a difficult uh uh problem to to combat. And in addition to fighting
17:08
friend fighting payment fraud, third party fraud that is perpetrated by
17:12
people that do account takeover and and collect you know stolen credit card
17:17
numbers. Uh so the fraudsters we see more and more first party and friendly
17:22
fraud and scams happening which are uh you know promote can be promotions fraud
17:27
which it could be reviews fraud such as you know people writing all sorts of fa
17:32
fake reviews on a website. So payments are just really the core uh that cause
17:37
the the you know a big component of of fraud losses but there’s these
17:43
additional elements scams policy abuse friendly fraud that really uh you know
17:48
adds to to the pain pain here. So you have to have tools that actually cover
17:52
this. So uh if we move to the next slide right I want to talk a bit about our uh
17:58
inclusion criteria. So this is our forester uh wave on digital fraud
18:03
management that was uh reflecting the third quarter of of 2023.
18:09
Uh we included tools and and DFM products that really where the vendor
18:14
showed us a complete digital fraud management offering strategy. Right? I
18:19
mean there’s a product you can ascertain there’s a product and it’s sometimes
18:23
it’s not that easy. Right? The second bullet item, you know, we have customer
18:27
valuated uh solution presence in the e-commerce and retail verticals. We
18:32
looked at vendors that have, you know, in addition to other verticals like the
18:35
financial services, uh like gaming, gambling, online marketplaces, etc. a
18:41
robust presence in the e-commerce and retail verticals. And the reason for
18:45
this is that we see the questions coming from these verticals for the most um
18:51
revenues, right? relatively simple, $30 million in product revenues annually. Uh
18:57
we we want to basically look at vendors that have a viable uh viably large
19:02
revenue stream. And then lastly, mind share, right? We we looked at vendors
19:07
that have a uh good end user as well as vendor clients mind share. So these
19:13
vendors come up in end user inquiries. customers, our customers ask about these
19:17
vendors and if on vendor briefings when from other DFM vendors, you know, they
19:23
are being mentioned as formidable and really good competitors, right? So, this
19:27
is basically wanting to make sure that we circle the the set of products here
19:32
in this wave for assessment that really are the the real important uh tools out
19:39
there. So if we go to the next slide with a bit
19:42
of a drum roll here this is uh our wave right so uh again I think it speaks for
19:49
itself right I don’t think I need to kind of go into too too much detail here
19:54
uh definitely a diverse landscape with a lot of differentiation in it um there’s
19:59
a lot of vendors they have each their strengths and weaknesses and keep in
20:04
mind we try to include vendors that really um are are falling ing into the
20:10
leader, strong performer, contender, and sometimes in the challenger category,
20:14
but but looking at tools that really are the cream of the crop. So, if a vendor
20:20
appears in this graphic, right, that that means that there’s a decent
20:24
solution at play here. We’re not looking at fly by night uh tools here. So, uh
20:30
definitely uh an important element. um good uh uh representation in in every
20:37
single category here as well. So uh I just want to talk a bit about on the
20:43
next slide some of the waiting of uh criteria. So we basically have two um um
20:50
elements here. The current offering which really looks at the bubble’s
20:55
position on the vertical axis right on the wave graphic that we’ve seen. Uh the
21:01
strategy that determines the bubble’s position on the uh horizontal axis and
21:06
the market presence that basically sets the size of the bubble. Right? So um we
21:12
looked at a number of different current offering categories. This is what the
21:16
product actually does, right? Um uh here uh we we absolutely look at user
21:24
management. So this is the administrative user management and
21:28
business user management uh capabilities. We looked at uh the whole
21:33
risk management uh set set of uh questions here that really talks about
21:39
um uh really rules uh rule uh decision rule based decisioning. So here heristic
21:46
rules we looked at statistical decisioning out there as well that
21:50
really looks at AI models, machine learning models, you know, various ways
21:55
of algorithms like evasion or supervised unsupervised uh learning algorithms as
22:01
well. Then we also looked at customer authentication policies which are really
22:06
how you lock people in right passwordless, biometrics and other based
22:10
other types of of capabilities. case management. This is basically Q
22:14
definitions, investigator screens, how people uh how administrator, how how uh
22:20
analysts and investigators within a retailer or or a third party can
22:24
actually in a day-to-day process can look at the outsorted uh transaction
22:30
cases and look at them, you know, manually and make a call whether it’s
22:34
real fraud or not. uh we looked at channels and transaction types which
22:39
really boils down to the fact that it’s not just online web but we see mobile
22:44
apps, phones, phone based transactions and others. So this is definitely uh not
22:49
u just a a uh different not just like an online credit card payment activity
22:57
non-payment fraud and policy abuse was very interesting. These are the areas
23:01
where we looked at things cryptocurrency payments, wallet payments, peer-to-peer
23:06
payment type of uh transaction rescoring as well as policy abuse. Right? So, this
23:12
is the bracketing uh the the hering the um the you know item not received types
23:19
of all kinds of scams and policy abuses um types of activities. The next thing
23:25
is chargeback management. that talked about this in in in in some detail how
23:29
the vendor has tools in the fraud fraud management DFM solution to be able to do
23:34
the uh the chargebacks you know uh fight those chargebacks and and win as much as
23:40
as possible um ROI and reporting it’s definitely important to kind of look at
23:46
the return on investment um of the tool itself right so and and today’s tools
23:52
actually are able to show to the retailer whether the tool is worth it
23:58
keep you know or not right so it’s it’s definitely uh really an important
24:03
element and then there’s traditional reporting like dashboarding and ad hoc
24:07
reporting types of capabilities integration uh how you tie the solution
24:12
into various other types of platforms such as identity verification
24:16
authentication customer identity management etc tools as well and then
24:22
lastly scalability right this is the the amount of of of transactions that the
24:28
solution protects at large customers. On the strategy side, right, that’s that’s
24:33
definitely important to mention. Uh we looked at the vision uh of the vendor.
24:38
We look at the roadmap which are the actually investment areas of uh of the
24:43
solution. What’s coming down the pike? communities, the size of the community
24:48
of the of the vendors, user group, innovation, this is typically staffing
24:53
levels, developers, professional services and and support services,
24:57
staffing levels. Uh we look at partner ecosystem, the percentage of revenues
25:02
that come from the top partners as well as uh you know how the vendor maintains
25:08
their their partner ecosystem and and evolves. adoption. What are the the
25:13
measures that the vendor actually uses to boost the adoption of a product? And
25:18
lastly, pricing, flexibility, and transparency. How easy it is for
25:21
somebody who’s never been exposed to this product to actually make the
25:25
retailer to make a call, you know, whether to buy or not buy the solution
25:29
and and basically come to a decent return investment. And lastly in the
25:34
market presence category uh the revenue from digital fraud management and the
25:40
number of installed customers right so it’s basically size of the install base
25:45
so well fairly straightforward uh elements here on the next slide right uh
25:52
uh definitely just you know if we move on to the next next one as well um
25:57
actually uh we seen the waiting right of criteria right so That’s definitely uh
26:04
part of of that as well. So there’s we have a uh uh display of these weights
26:11
once more. Just go to the next slide and
26:14
>> uh yes.
26:16
>> Yeah. Okay. Good. So that I think we we covered this over to you to talk about
26:21
some of the buyer outcomes that that you see guys uh folks are seeking.
26:26
>> Yeah, appreciate it. Thank you for like all that context around the you know the
26:31
capabilities and um and certainly how the wave was conducted. I think that
26:35
context will be very very helpful for uh for our audience. So really appreciate
26:39
that. Yeah. I just wanted to share just a little bit of context um you know from
26:43
our point of view here at SIFT having having kind of been in the trenches
26:46
around what buyers are ultimately seeking you know and the you know
26:49
hundreds and hundreds of merchants and e-commerce providers that Sift interacts
26:54
with uh on a daily basis. Um the if we were to distill down the business uh
27:00
outcomes that they are seeking, I I believe these five capture it really
27:04
well. Um ultimately it comes down to great decision-m right-time decision-m
27:09
to uh make a yes, no, or maybe even a maybe call in the moment. Uh but to do
27:14
that with confidence and have confidence that the incidence of false positives is
27:19
very very low. So that’s that’s really first and foremost. The visibility and
27:22
control is super important. uh and being able to even with an MLbased um solution
27:28
to have some degree of control over policy setting um in tuning for your
27:33
specific vertical is also a really important outcome u and requirement that
27:37
we’re seeing from customers. Doing all this and making these massive
27:41
investments that have real implications on your topline and bottom line, doing
27:45
that efficiently is very very important. As we talked about earlier, um there are
27:49
a number of business models in the marketplace. There’s software ccentric
27:52
business models, there’s servicesbased business models, um, and there’s
27:56
different efficiency ratios when interacting with vendors, you know, on
27:59
one end of the spectrum versus the other. And so capital efficiency is
28:02
certainly something that, um, every customer we talk to, this is a top-of-
28:06
mind outcome with that is the people efficiency. And I think that’s very much
28:10
a byproduct of the business model of the solution provider you engage with. Uh
28:14
but being able to have hands-on controls and drive the outcomes that you need um
28:20
with um with with a core team uh without getting too deep into a team or too rich
28:26
of a team since cost especially in these recent quarters is becoming very very
28:30
important. That’s an important outcome that our our buyers are seeking. Um and
28:34
finally getting to value fast. Um especially where there’s science
28:39
involved, there’s models that need to be trained and learned on your own data. um
28:43
and getting to value fast is certainly an outcome that um many many of our
28:47
customers if not all seek in their journey.
28:51
I want to pair that last slide with uh another point of view and we again we’re
28:56
hearing this time after time from our customers and and really touched upon
29:01
this. It’s not just about account opening. It’s not just about login. It’s
29:06
not just about payment transactions. It’s the full digital user journey where
29:11
risk is evident and where right decisions need to be make need to be
29:17
made at those moments in time and to make those decisions with confidence um
29:22
leveraging science and data and to do it rapidly. The worst thing that can happen
29:28
is to disrupt a good user with a bad call. Um, and so avoiding those bad user
29:35
experiences when it’s in fact one of your most valued customers is so
29:39
important. It’s one of the reasons why full digital user journey visibility is
29:44
so important. So the context and insight from one moment in time in the journey
29:49
can be leveraged further on down the line for that same customer.
29:57
So with all that, let’s transition to the next topic around trends. what we
30:02
are seeing in the marketplace and we would like to start it off with Andrush.
30:08
>> Thanks Armen. So definitely an important element as as you said you know a few
30:13
seconds earlier it’s it’s a complete journey right and the interesting kind
30:18
of aspect of covering the entire journey is basically a policy right so
30:25
there is payment fraud which is really obvious right I mean if if you have a
30:29
fraudster using a stolen credit card number I mean that there’s not much you
30:33
know kind of debate around how good or bad that that’s bad stuff right if you
30:37
have somebody who’s taking over accounts of good customers and making payments uh
30:43
and buying goods on on the victim’s behalf, that’s bad. But when it when it
30:48
really start when the lines really start to blur here, um it’s basically
30:53
policies, right? So, uh if you look at policy, right, a policy and its abuse,
30:58
right, are problems. So, first of all, what is a policy? A policy is a codified
31:02
intent to shape business, right? It it it is t typically something that
31:06
provides streamline easy experiences to good customers and limit fraudulent
31:11
experiences for bad customers and obviously they’re victims, right? So
31:16
what are some of these policies, right? Um you know obviously payment fraud is
31:22
not allowed, right? That that’s a policy you have. But you know if you if you go
31:25
beyond that to non-payment areas, right? Is reselling goods and merchandise
31:31
allowed or not allowed? Right? If you’re a shoe manufacturer, right, so say Nike,
31:36
and you’re you’re issuing a limited release or limited batch of of say
31:41
50,000 pairs of collector’s item Nike shoes, right? What’s the intent here? Do
31:47
you does the organization, the shoe manufacturer want to allow those shoes
31:51
to be bought up by, you know, five different, you know, gangs groups of of
31:56
resellers, right? And resell them or not, right? There’s there’s not one good
32:01
answer to this, right? Are freight forwarders allowed or not
32:05
not allowed? Right? Can you know can somebody from who represents another
32:09
country, right? Where these shoes are even worth a lot more money even if
32:13
they’re not special edition buy a thousand pair of shoes and ship it to an
32:18
address that’s a freight forwarder to that country, right? Is this allowed or
32:22
not allowed? Right? So there’s there’s definitely good uh you know there’s not
32:26
a good not not a single answer to this question. Is wardrobing or bracketing
32:30
allowed or not allowed? So wardrobing is basically uh the idea that somebody
32:35
buys, you know, a pair of shoes and then returns it a year later, you know, worn,
32:42
right, with the intent of getting their money back. So basically renting a pair
32:45
of shoes almost at zero cost, right, for a year. Bracketing is basically, you
32:51
know, buying five pairs of shoes, different sizes, trying them all on and
32:56
return them, right? Uh some brands right may allow one and not not the other.
33:02
Some brands disallow both or allow both right. It’s it’s really uh policy
33:07
intent. There’s no good or bad here. Um sometimes the items are of such high
33:13
value that you an organization and retailer wants to allow some of these
33:17
activities to happen. Return windows and conditions. How can you return? How long
33:21
can you return the item? for how long um um you know the customer can return and
33:27
get the refund for that item. What does it mean that you know if if the if the
33:32
customer has never received that item what are the replacement policies and so
33:36
on right so these are all parts of policy that that all play a role here.
33:41
So if we move to the next slide, right, if you have fraudsters abusing these
33:47
policies, right, that is typically a bad thing. Um, the reason it’s a bad thing
33:52
because it is it’s somebody’s got to pay for, you know, the guy who bought a bad
33:58
pair of shoes, worn them for, you know, for a year, and then returned them,
34:02
right? And that that those pairs of shoes are are really unusable, and you
34:06
have to ride those off, right? Um so so that’s one thing that these types of
34:12
policy abuses really cut profitability uh and really are are are causing you
34:20
know prices actually to go up for the legitimate uh customers.
34:26
Typically the challenge is is uh the perpetrator has may have been a
34:31
legitimate customer, right? So it’s not like it’s all committed by fraudsters,
34:35
right? You know, some people just are not aware of these types of uh of rules,
34:40
right? It’s a violation that can be unintentional, right? Hey, I I never
34:45
knew that you don’t allow, you know, freight forwarders. I never knew you
34:48
don’t allow, you know, bracketing of items, right? Um and and there’s a ton
34:53
of you know kind of potentially uh problematic
34:58
policies right that if if they’re being violated then customers and then then
35:03
the customers are treated unfavorably afterwards they can actually come back
35:08
at the organization and say hey you know you you discriminate against me right
35:12
which can definitely be a a um a problem. So, like I said, the the the
35:19
typical scams or policy abuse types of activities are uh promotion abuse,
35:24
right? Like there there’s definitely the coupon abuse and and really various ways
35:28
of buying more items at a discount than, you know, normal user should be loyalty
35:33
and sign up abuse, right? U basically redeeming loyalty points uh
35:39
that you know that you never accumulated. Um item not received,
35:44
right? it by big guy then big screen TV uh that comes in fine you plug in the TV
35:49
and then you claim that you never received the television the FedEx guy
35:52
stole it or you know basically UPS guy stole it and then there’s a fraudulent
35:57
returns right basically you get the item and then you return the box uh basically
36:03
empty to the wrong address or or full of dry ice or or rocks and and and
36:08
basically you know how that triggers the uh merchants uh
36:14
the merchants re refund process, right? So, these are all um elements that um uh
36:21
you know really cut into profits. So, so we understand why it’s a what they what
36:27
policy abuse and and scams are, why they’re a problem because they cut into
36:32
uh customer acquisition cost and and and cost of items, right? And that limit the
36:36
competitiveness of of the retailer. Now let’s move on to the next slide and and
36:41
try to see why policy abuse is hard to combat right so the biggest challenge
36:46
right as I said is that hi or two good customers start doing bad things right u
36:51
and if you start cracking down on these activities this may actually backfire
36:56
and can hurt your brand image really bad right so you know you really need to
37:00
know what you’re trying to do here a unified and document policy is typically
37:05
going to be hard to create and understand and usually easy the game for
37:09
the people, the fosters that know what they’re doing. And if you start, if you
37:14
don’t do this right, you may even in kind of expose your organization to
37:18
legal challenges, class action lawsuits, and all these other kinds of things that
37:22
hey, you know, there’s been a number of good people, legitimate customers that
37:26
got basically discriminated because of a bad policy or or bad segmentation or bad
37:31
rule, bad uh decision-making process by the retailer. and then they start filing
37:37
class law class action lawsuits against the company. So these are things you
37:41
generally you know want to avoid right. Uh Armen if you want to you know speak
37:47
about frauds interconnected nature. I think you had you were thinking of doing
37:51
that.
37:51
>> Yes absolutely and I before I do that I just want to um make one quick comment
37:55
on policy abuse. Just uh to tie back to one of the points I made up front based
37:59
on the consumer research, right? 16% of consumers based on our own research have
38:04
admitted to engaging in some form of firstparty fraud or friendly fraud and
38:08
and certainly involving policy abuse. So um these two slides Andra are you know
38:13
very real and uh you know maybe even underestimated and underappreciated in
38:17
the world of digital fraud management. So excellent uh excellent clarification
38:22
there. Um what you are seeing here is a depiction of of actual fraud patterns on
38:29
the SIFT global data network. Um the the visual to the right which um hopefully
38:35
you can see the animation. It’s an actual fraud ring that we uncovered
38:39
based upon our own forensics, our own research. And the point we want to make
38:43
here is there is a complex web of players and actors um and elements and
38:51
ingredients in an organized fraud ring. And that’s depicted here with the
38:55
different colored uh nodes of uh transactions, credentials, uh payment
39:00
authentication um components, um pieces of infrastructure. And it’s just
39:05
fascinating to see, you know, to contrast the first party fraud or
39:09
friendly fraud examples with the other end of the spectrum of far more
39:13
organized transactional fraud um and the sophistication that takes place. Uh we
39:18
felt this was just a very interesting point of view to share um with this
39:22
audience. And the point is in the world of digital fraud management having the
39:25
capabilities to accommodate and detect and then act upon all forms of fraud
39:31
organized and disorganized or for a first party is uh is very important from
39:35
our perspective and from the perspective of our customers.
39:38
Um the second point I want to make on this slide is that you know a an attack
39:44
in one industry certainly can have ripple effects into other industries.
39:47
Right? This is a network effect uh where certainly the attack types the patterns
39:52
um the testing might take place in the marketplace space and then might move
39:56
for example into uh a buy now pay later transaction. And so we’re seeing those
40:02
types of um testing, learning, especially from the organized fraud
40:06
community and and then rapidly, you know, iterating and then deploying in a
40:11
different space where they have that advantage, that split-second advantage
40:14
of time uh where maybe those uh attack patterns haven’t yet been detected by
40:18
the broader fraud management community. And then finally, this notion of the
40:23
democratization of fraud. It’s a message that we’ve been projecting from a SIFT
40:28
perspective for uh many quarters right now. And the notion here is that the
40:32
tools um and processes are now more broadly available than ever before for
40:38
fraud actors to play in the fraud economy and benefit the the cost of
40:44
success or the the benefit of success if you’re a fraud actor is significantly
40:47
high um if you have some techniques and some innovation. Um and certainly more
40:53
fraud actors can participate in that fraud economy. And so that’s something
40:56
that we should all recognize in our quest as fraud fighters to um to put our
41:02
best foot forward and and help really stop the proliferation of fraud as best
41:08
we can as a community. So with that, let’s transition into the
41:15
the final chapter here on best practices. And is uh going to lead us
41:19
off.
41:20
>> Sure. Um so real quick, right, the risk scoring is the heart of it all, right?
41:24
So you have to be able to risk score by payment, non-payment types of
41:28
transactions and and be able to cover you know policy abuse and scams as well
41:34
as payments. Next thing obviously everybody uses rules heristics. These
41:39
are for like if the the item is this SKU and the the amount of dollars paid for
41:45
them is higher than this than then add to the risk score. So this is basically
41:49
your rules and heristics help with tactical you know stop gap measures you
41:54
know uh in in fraud risk scoring and sometimes they make life a lot easier
41:58
compared to machine learning and AI based models which then in turn help uh
42:04
protect against un un you know so far not seen behaviors and be able to add a
42:09
lot more uh in essence clustering and and understanding of of activities and
42:16
detect behavioral anomalies. uh the reporting should be capturing
42:21
return on investment and value of a customer. So these are areas where
42:25
business people actually can see the value of fraud management. Uh and then
42:30
checking efficiency of fraud detection at least weekly is very important right
42:34
you cannot rest on your laurels as Armen showed with with these moving graphs
42:40
right things are changing fast right sometimes within matters of hours right
42:45
the fraudsters find new ways of of uh gaining uh monetary uh benefits out of
42:53
fraudulent activities and then basically expecting discussions with marketing and
42:58
business stakeholders Fraud is not standing in on its own. It has impacts
43:02
on customer acquisition, retention and the experience. So it’s always a
43:06
balance, right? How much you want to you know you want to sacrifice in terms of
43:12
the customer experience towards you know stronger uh fraud management. Good tools
43:17
can actually help you do both. So that’s my perspective. Armen, what what are
43:22
your areas of excellent practices?
43:25
>> Thank you. Yeah, very good context. So just a few more to to pile on. Again,
43:28
this is an amalgamation of I would say insights and best practices from the
43:32
customers we work with at SIFT every day and you know we talked about point one a
43:36
few times but just to reinforce it decisioning across the full digital user
43:40
journey right it’s not one moment in time it’s not just loginins it’s not
43:43
just payments as important as those are it’s having full visibility and be able
43:46
to provide a real-time risk assessment with confidence at those moments in time
43:52
um consortiums and consortium data is becoming more and more important across
43:56
the industry this is not unique to sift necessarily. We’ve been doing it for a
44:00
long time. Uh and and really the notion is it takes a network to fight a network
44:04
is the phrase I like to say. And so consortium data um where the collective
44:09
can benefit from the insights um and the decisioning from the rest of the
44:14
community is a very very important measure uh that we believe in and we we
44:19
are proponents of and certainly our customers are too. Um the third bullet
44:24
I’ve referenced a couple of times and this revolves around business models. Um
44:28
here at Sift we are a software company. We have built a digital fraud management
44:33
platform. Um and we lead with that in our our empowering our customers to
44:38
fight fraud. That said, uh we know that every environment is a little bit
44:42
different and there are cases where a managed service offering is important um
44:48
to supplement or augment the the fraud solution the software solution that’s
44:52
being delivered. So having the um as a as an organization as a fraud fighting
44:57
organization understanding where you fall and where your needs are is very
45:01
very important and looking for vendors that provide that uh degree of
45:04
flexibility is also very important. with that. Um, on the issue of science, um,
45:11
SIFT is by definition and by background a we’ve led and we lead with our machine
45:16
learning, we lead with the data science, but we know that every environment’s
45:21
unique, every environment is different. So while we can get you probably about
45:25
80% of the way there with unsupervised machine learning models that can be very
45:30
applicable to your space, we know that there’s that incremental, you know, 20
45:34
points that is unique to your space somewhere between 10 and 20 points. And
45:38
so applying those policies and rules uh for your business and fine-tuning that
45:42
is going to be important. So again uh that is a best practice that we’ve seen
45:46
is really having that balance of machine learning le with the ability to
45:50
fine-tune policies accordingly. um embracing the ecosystem um not just the
45:55
consortiums but you know working closely with your solution providers and the the
46:01
domain and subject matter experts that are there um participating in peer
46:05
industry groups for example participates quite heavily and the MRC the merchant
46:10
risk council it’s a tremendous um industry group or consortium where
46:15
there’s a tremendous knowledge sharing and we’re very proud um to be members we
46:19
actually have someone on our team who’s on the board of adviserss for the MRC
46:22
and that’s been very valuable I think mutually for all parties.
46:26
Finally, just good monitoring of the data and adjusting accordingly. Um it’s
46:32
never one and done right when you’re implementing as Andra has shared
46:36
revisiting the tuning the policy setting frequently if not if not daily weekly um
46:43
but certainly not monthly uh because the patterns are changing so rapidly and
46:47
it’s just having that operational rigor and discipline to stay on top of it. Um
46:52
and then finally the the communication right fraud is not just managed by a
46:56
fraud organization or an anti-fraud organization within your company. Uh but
47:00
certainly there are stakeholders across the business across your business that
47:03
need to be pulled in um both from the information security side of the
47:07
business to legal and compliance um and certainly like your customer operations
47:11
teams. So with that let’s transition into poll
47:16
number two. So hopefully you can see the uh the poll appear in your Zoom console.
47:20
I’ll just read the question. Um, which team is driving the decision um to
47:27
invest in your digital fraud management strategy? And you can see we have five
47:31
choices here. And I’ll just pause for for 10 seconds or so while you answer.
47:44
Great. Thank you for participating there. Let’s um
47:50
move into our predictions. Andra, you’re back up here. I know this is the the
47:54
final segment here before break into
47:56
>> That’s always uh an interesting question. I’ll just cover this real
47:59
quick. So, additional vendor supplied models for you know things like new
48:03
payment types. Uh these can be peer-to-peer, digital, zel,
48:08
cryptocurrencies, uh all sorts of payments and they vary
48:12
quite significantly globally. It’s not just the US but there’s there’s a number
48:15
of these in Europe as well as in in in you know India uh Middle East as well as
48:21
in Asia. So there’s a lot of these types a pretty deep area scams and policy
48:26
abuse. So nonpayment types of fraudulent activity is definitely important. That’s
48:30
that’s something that is really difficult to combat because of the
48:33
reasons we talked about earlier. Generative AI, right? Um using so
48:38
generative AI is basically a tool that these tools for digital fraud management
48:43
can use and will likely use for things like providing an interface to data
48:48
scientists and business people who are not that savvy yet that you know the
48:52
data science right or all the in intricate details about model management
48:56
tuning etc. So basically you can use generative AI to provide a natural
49:01
language query to the tool. Hey, what are the most recent types of fraudulent
49:05
activities or patterns or what are the most uh defrauded items, right? Or what
49:12
are the uh the fraud rings that are most worth knowing about, right?
49:17
Investigation, right? How you produce a narrative over events that or cases or
49:22
or activities that are form a chain that is human readable and definitely
49:27
reporting, right? So being able to generate reports that not just look nice
49:31
but really hang together and have a of a a story behind them. Uh consortium data
49:37
expansion is yet another trend. Um as uh Armen talked about uh and I fully agree
49:43
with him. There’s definitely a need for learning from more data elements such as
49:49
IP addresses, device fingerprints, uh hot list, so basically allowing block
49:54
lists, models, right? So the the more models that are available from a vendor
50:01
the better. So basically the more data that that’s out there the better quality
50:05
decisions you can actually make. And then lastly we see more customization
50:10
and being exposed for risk scoring uh models to the end user retailer and
50:15
organization. It it is not an activity that falls uh strictly and purely on the
50:21
on the shoulders of the vendor, but more and more retailers and organizations
50:26
want to be able to fine-tune these models to their liking, to their needs.
50:31
>> Thank you, Andra. Uh we’re going to round this out. I’m just going to um m
50:36
really make one statement here um about Syft um and our position in the
50:41
marketplace. Um, SIFT was very grateful to have been named um, a leader in this
50:45
most recent wave for digital fraud management. This was a rigorous process.
50:49
What you see here is a depiction of the SIFT solution on one slide. Um, and we
50:55
talked about many of these issues. At the very bottom, it’s our global data
50:58
network where we score more than one trillion transactions per year. The next
51:02
layer up is really our machine learning and science and all the policy setting
51:07
driving to rules and ultimately decisions. And that’s ultimately what’s
51:11
driving your digital user experience. And at the top of this layer cake are
51:15
the solutions that cut across that digital user journey which we’ve talked
51:18
about many times on today’s call. I do want to highlight what’s on the right.
51:22
This is the ecosystem as we’ve defined it. It’s the domain expertise. It’s the
51:27
community of peers. We have a community called sifters where you can interact
51:30
with your your peers, other merchants, other digital native businesses. And so
51:35
when you look at this entirety um of this this story, this cap set of
51:40
capabilities um think about this when you think about Zift. We understand
51:43
there’s a lot of choice in the marketplace and again we’re very proud
51:46
to have been named a leader in the Forester wave. With that, let’s
51:50
transition into some questions. I’m actually going to stop sharing
51:56
and go to our full screen mode. I do see that there’s been some questions
52:01
submitted. I will uh I will pull them up and I will just quickly work through
52:09
and let’s see uh administrative question. Armen, can you please share
52:12
the Forester report please? One I see in your background. So the report is
52:16
available uh we have purchased the reprint rights from Forester. So if you
52:20
go to cif.com on the homepage you will see a link to um to to download the
52:26
report and so please feel free have at it. it’s available. Um, and we thank you
52:30
for engaging that way. Uh, next question. What do you mean by
52:35
orchestration and how is that different from workflow? Andra, could you please
52:39
take that?
52:40
>> Sure. I mean, orchestration is being being the ability to be able to do, you
52:44
know, two or three things. One is being able to programmatically get data into
52:49
the DFM platform like seamlessly and ingest data in here. to be able to from
52:55
the DFM platform call you know certain other systems transactional other fraud
53:00
systems or just non-fraud tools right payment systems you know to kind of make
53:06
certain things happen like block a payment etc and and the workflow can be
53:11
part of orchestration as well so the workflow can be defined as as basically
53:15
code right of actions how things need to be happen if this is true then do this
53:20
if this is you know this other next thing is not true then not do this and
53:24
so on, right? So, typically orchestration can be a part of of the of
53:29
uh uh so so I’m sorry, workflow can be a part of orchestration or it can be a
53:33
separate thing depending on on the product itself.
53:36
>> Excellent. Thank you. We have another question. Um I think we’ve answered what
53:40
is friendly fraud again my point of view here. Um it’s it’s effectively good
53:44
customers that are bending the rules or pushing the rules around policy abuse um
53:48
or you know return policy etc. We’ve covered that with a few examples. Andra,
53:52
is there anything else you’d like to share of kind of the definition of
53:55
friendly fraud?
53:56
>> Yeah, I mean, it’s basically all the all the kinds of things that don’t
54:00
necessarily involve a professional third party fraudster, right? It’s it’s
54:04
basically first party good guys who’s been who who have been you know as I
54:08
said so far been behaving okay starting to act up quote unquote and and doing
54:14
you know causing problems causing you know financial losses or just abusing
54:19
policies or participating in scams.
54:23
>> Agreed. Okay. Next question I see here is do we have a deep dive into each of
54:27
these categories? I’d say you know two answers. one in the wave. I think
54:30
Andrush did a great job at really decomposing u the categories and the
54:35
scoring by vendor with each of those. Uh so that’s one way to answer the
54:38
question. Um um as far as a vendor point of view, SIFT is of course happy to
54:44
speak with you and and you know share our point of view on these categories as
54:48
well. Is there anything else you’d like to share? address.
54:50
>> We actually do. I I actually wrote a report on this that you know I’m going
54:54
to paste the link to uh in the in the chat as well for you guys and then you
54:59
can the safe team can actually share this this uh link.
55:04
>> Excellent. Uh next question. Do you have a fraud mana a management maturity
55:09
model?
55:10
>> Um we actually are working on one. So
55:15
that’s the answer to this. We do have some ports and parts and and parcels of
55:20
it. It’s work in progress as as we speak.
55:23
>> Okay. So, we know Gen AI is a big hot topic. You you touched upon it a little
55:27
bit. There’s a question specifically here is what place does generative AI
55:31
have in digital fraud management in the future. So, do you know I know you
55:35
touched upon this Andra, but any other thoughts? IP generator right I’d say is
55:39
the is is use case number one but beyond that is basically formulating I think
55:45
the first phase formulating uh queries uh against the system against a fraud
55:50
management tools tool in a natural language such as English and getting
55:54
natural language such as English responses back um that’s one thing I
55:59
would also say um moving forward being able to kind of even detect more
56:05
behavioral anomalies could another you know use case for this.
56:09
>> Yeah, I I would echo that. I would say from a from a solution provider
56:13
perspective um injecting AI into the forensics for
56:19
you know our identity graph for example um and understanding relationships and
56:23
connections um we’re starting to experiment with that and certainly from
56:27
a an attack perspective um you know we’re seeing genai used um with you know
56:32
creating synthetic identities AI generated synthetic identities um so you
56:37
know artificial intelligence in many forms will continue to play a role in
56:41
the broader DFM space. I see another question here. At what point in a
56:46
business’s maturity should it work with an external DFM vendor? You know, Andra,
56:50
please share your thoughts on that.
56:52
>> Um, I I would say, you know, you can, you know, any any maturity, you know,
56:58
organization can benefit from working with a tool, right? You know, typically
57:03
products are going to be much less brittle. You know, a lot easier to
57:08
upgrade, maintain, and operate. And anytime you have something that’s
57:11
in-house built, in my opinion, in our client’s opinion, it’s it’s becoming a
57:16
marriage to the developers who build that thing. And there’s a lot of kind of
57:19
one-off things that you have to do as an organization that a vendor would
57:23
basically take on, you know, in a much lower cost, right, over a long over a
57:28
long period of time.
57:30
>> Excellent. So I want to allow you know if there are any other live questions
57:33
that wish to be voiced um we can certainly make that available or you can
57:38
just again simply submit your questions through the Q&A and we can we can
57:41
address them live. So I’ll just pause there for a moment and just take another
57:45
final scan here um of the later submissions. So I see
57:50
one that just came in um very specific. So in the dispute management, can you do
57:57
analysis on dispute tags? We have so many friendly fraud um incidents that
58:04
card holders dispute as fraud. So we want a way to decipher. And do you have
58:09
a point of view on this?
58:10
>> Uh yeah. So I mean definitely being able to decipher and classify activities is
58:15
is an important um kind of capability. So the creating a fraud the pattern
58:20
catalog catalog is is a good function in DFM tool.
58:26
Excellent. Okay, last call for questions. Some some good ones here.
58:32
We appreciate it. Okay. Well, with that, Andra, I would
58:43
like to thank you and thank Forester for uh for lending you to us for today’s
58:47
session. Thank you so much for your point of view and um you know, I
58:52
personally found this to be very enlightening and rich. I would like to
58:54
thank the audience for sticking with us uh for a full hour as promised and uh
59:00
hopefully you got something out of this. Again, the Forester wave report um will
59:04
always be available for in the near term at least from the sift.com website. And
59:08
did share a link in the in the chat here and um we’re always happy to speak with
59:13
you one-on-one if you have any further questions. So with that, I’d like to
59:16
bring today’s session to a close. Thanks very much for joining and have a great
59:21
day. Thank you.