Customerland

If Agents Shop For Us, Who Decides What To Trust

• mike giambattista • Season 3 • Episode 48

Share episode

Fraud doesn’t just show up as a stolen card anymore—it arrives as coordinated global operations and as “friendly” misuse from customers who otherwise look perfect on paper. We sit down with Armen Najarian, CMO at Sift, to unpack how merchants can block the bad without breaking the checkout flow for everyone else. From account takeovers and identity theft to policy abuse and return gaming, Armin explains why risk isn’t a back‑office metric but a core part of customer experience, where a single false positive can cost a loyal buyer and a long relationship. 
 
We dig into how Sift evaluates identity in real time, returning a risk score in about 200 ms, and why context across a network of merchants beats one‑off signals. It takes a network to fight a network, and that shared view turns fragmented behavior into reliable trust decisions. The conversation moves to agentic commerce—AI agents that can discover, sign up, and transact on our behalf. Convenience is huge, but so are questions: how do we authenticate agents, delegate consent, assign liability, and keep fraudsters from hiding behind machine identities? With card networks, banks, processors, and solution providers racing toward standards, the stakes rise as forecasts point to a leap from billions today to over a trillion dollars in agent‑driven transactions by 2030. 
 
We also explore why Gen Z gets phished more despite digital fluency, the practical limits of self‑sovereign identity in the private sector, and the growing dual threat: organized crime scaling up and first‑party misuse spreading inside customer bases. If you care about conversion, approval rates, and loyalty as much as chargebacks, this is a roadmap for making trust invisible when it should be, and unmistakable when it must be. If this conversation helped you see risk differently, follow, share with a colleague, and leave a quick review so more people can find the show.

SPEAKER_01:

On one hand, we're seeing highly organized fraud. Most fraud at scale is highly organized by fraud organizations based in all parts of the world with organizational structures. They're funded, they have capital, they have technology stacks, they have teams of people executing frauds of different types, account compromise, credit card theft, identity theft, right? This is that is one big source of fraud. In contrast to that, we're seeing in recent years the rise of what we call first-party misuse. First party abuse, you might hear of it as friendly fraud, right? So it's individuals like you and I that are otherwise trustworthy, kind of going rogue for a moment.

SPEAKER_00:

Today on Customer Land, Armin Nejerian, who is CMO at SIFT. And if you haven't heard of SIFT, well, shame on you. Well, we're going to talk about you, about the company and enlighten you in the process. But first, I just want to thank Armin for joining me. Really appreciate this. Thank you, Mike. Really uh appreciate the opportunity to talk. So there is an awful lot going on in your world. Funny thing about it though is it touches my world um only occasionally, but when it does, it tends to touch it violently. You know, it's always because of some giant news item that you know you have to consider these things that you otherwise kind of don't. With that, though, explain to us what SIFT is and your role there. Yes.

SPEAKER_01:

And to your first point, we've all felt as consumers the impact of digital fraud, or we know someone who has. And so, and I have too personally in a big way. Uh, and so what we do as SIFT is one of the leading providers of business to consumer fraud protection for for digital merchants worldwide. And when I say merchants, it's a broad array of merchants from your traditional online commerce sellers to travel and tourism airlines to online gaming and many, many more verticals. If there's a digital transaction, chances are SIFT is there behind the scenes scoring that transaction and assessing for risk.

SPEAKER_00:

So um I do want to get into the mechanics of how that works because there are a handful of different mechanisms out there that different vendors are using. Um, it just from the research that I've done, SIFT seems to be little everywhere, like you just said. Um, so you've got a system that a lot of people appreciate, but um I think maybe before then, can we talk a little about your personal journey to this space? Because it also seems to me like you've got a lot of history here, which means that if I'm right, you've seen the evolution of how this works. Yes.

SPEAKER_01:

So, as a matter of fact, uh last month I celebrated 10 years as a CMO in this space.

SPEAKER_00:

Okay, congratulations.

SPEAKER_01:

Yeah, it's been quite a journey. Uh by design, though, I've chosen to stay in this space and for for many reasons. So, yes, I have seen the evolution before that. Uh many years ago, I started my career off in consumer brand marketing in the grocery products business and eventually you know went to the dark side to uh to being a technology you know marketer, B2B technology marketing.

SPEAKER_00:

So, can you talk to us uh a little about the timeline and some of the changes, the approaches to uh fraud and security that you were dealing with and implementing way back when and what it kind of looks like now? Yeah.

SPEAKER_01:

So 10 years ago, the summer of 2025, 2015, I should say, the um, you know, I'd say we were in the aggressive growth phase of digital commerce, right? This was the moment in time where most, you know, most holiday shopping was transitioning to online or a good chunk of it. Um, every traditional business was becoming a digital business and digitally transforming. Uh and you know, and with that, the exploitation of fraud became very real. And so um 10 years ago, the landscape of solution providers like SIFT was probably 25% of what it is today. Um, and also the sophistication of the fraud attacks were not nearly as sophisticated. Yeah, fast forward 10 years, uh, the proliferation of fraud, both organized fraud and what we call friendly fraud, people like you and I that just decide to go to go rogue, uh, but not in a really nefarious way, uh, exists in a big, big way. And fraud, I would say from a merchant perspective, people we sell to, uh, they're viewing fraud decisioning not as a cost of doing business, but really as a customer experience investment because they want to get the they want to make the right decision for every single customer, every single transaction that they're scoring. That's been the big shift in the last 10 years.

SPEAKER_00:

So I think I'm stating the absolute obvious here, but you know, trust has been baked into the process and has been assumed to be there since day one. But with what SIFT is all about, it doesn't necessarily become important until you're in either uh you've been violated somehow or you're in kind of a vulnerable context, in which case you're you're really thinking about it. Otherwise, SIFT kind of operates in the background just making sure things go well, correct? 100%.

SPEAKER_01:

So SIFT is not visible to the consumer, but we are always present and always on. And so what we effectively do, let's just say you're making a transaction on an airline, you're buying an airline ticket, you hit the buy button, SIFT is called in real time. We come back with a risk assessment of effectively the user behind you know who hit the buy button, and we'll provide a real-time score in 200 milliseconds of should you trust this transaction, right? Is this a high risk or a low risk transaction or somewhere in the middle? And that's what we do, and it's all behind the scenes.

SPEAKER_00:

Interesting. I was looking through some of your research prior to this conversation, trying to kind of just frankly, just because it's really interesting, finding out how the fraudsters are operating and how people like SIFT are kind of countering that. But you know, one of the the big questions that occurred to me is not only wondering how SIFT is going about understanding what the potential fraud models might look like and how you what you're doing to mitigate that, but but how businesses are thinking about fraud right now. Because on the one hand, you probably, maybe even you personally, have a really clear view of how this all works. My sense though is that the businesses that would buy SIFT services probably aren't quite as sophisticated there, probably don't have a view to how what the real dangers are, uh, the real risks and the and the size of that liability.

SPEAKER_01:

That has changed a lot in the last 10 years. And you know, SIFT is primarily selling to you know mid-market all the way up to the enterprise, where they're, I'd say, more aware than maybe an earlier stage business of risk management, right? And so who we are selling to, you know, typically like the titles involve like a head of digital fraud, a head of payments, uh even the treasury departments, you know, within the company, the finance and the CFO persona, where risk management um is is becoming core to how they operate their businesses. And so I'd say the level of awareness, unlike 10 years ago, is pretty substantial. And I'll take it to the next level. You're starting to see certain segments of customers where risk decisioning is in their DNA. And so I look at some of our fintech customers where understanding risk at an intimate level is core to what they do. And SIFT is riding along there with, you know, to augment a lot of the in-house capabilities that we have. It's really um the level of awareness from a buyer perspective has grown tremendously over the last 10 years.

SPEAKER_00:

Um, it brings to mind a conversation I had recently with a CEO of a company who broke it down like this. He said, you know, my my my job is um cost, revenue, and risk. He said, I I look at that daily, and I frankly, I don't know if I've ever heard anybody break it down in those particular terms. I never heard anybody break it down saying risk was one of the top three, and it's clearly kind of shifted positions there. And that person, um that person operated in a digital retail space. He was an e-commerce provider. So yeah, I can see why that's becoming such a big deal.

SPEAKER_01:

Well, here's why it matters, and I alluded to this earlier on is the inverse of risk is delivering an amazing customer experience. So if you get it wrong, like if SIF delivers a wrong score and we issue what's called a false positive, and we declare the user behind the transaction as being high risk and they're not, they're actually a great customer, what happens? You're going to seriously disrupt a good customer flow, perhaps cause them enough insult to want to go to the competition. So it risk management is critical, and the inverse of that is getting and treating all your great customers right. Because the reality is of all the transactions that SIFT is scoring, more than 95% of them are trustworthy. It's that 5% that we've got to get right, and we don't want to get any of those 95% wrong.

SPEAKER_00:

I'm actually surprised to hear that it's that it's uh as large as 5%, but that's how naive I am. I don't play in the space. I want to, I want to, because I need to talk about AI for a little bit because I feel like there's been an outsize impact in fraud. I mean, we're we now as consumers are using AI-based tools for all kinds of things, but it seems to me like fraudsters, people who are intent on doing the wrong thing, they have those same tools at their disposal. So it seems to me as if we're fighting an escalated battle here. Do you have any perspective on that? You know, what has AI done there or or what can it do?

SPEAKER_01:

So we're seeing the rise of what's called agentic commerce right in front of us. What does that mean? Is shopping agents that a consumer like you or I can commission to go out and perform tasks for you, including discovery, registration, transaction, lights out without any of your involvement. So that has changed the game uh from a consumer perspective on how we shop, but also from a risk perspective, right? Because that brings up a whole series of, okay, as a solution provider that's in the business of assessing risk, you're now dealing with a machine, not a human. And how do you understand the identity of that machine? Is that a trustworthy agent or not a trustworthy agent? Is it nefarious or not? So, yes, this is the big wave that's that's right in front of us that is helpful to consumers, but also presents a different type of risk to the merchants.

SPEAKER_00:

Would you frame this as consumer false confidence in a gentic AI at the moment? Or are we not quite there yet? Or am I am I trying to put a label on something that doesn't quite exist yet?

SPEAKER_01:

There's been some recent research that where where the uh the awareness by consumers, consumer research, um 70% of consumers are aware of the opportunity to invoke an agent to go out and and do their shopping for them. Um and 30% of them are concerned about trying it. So I'd say it's early adopter phase still. Okay. There are risks, right? There's certainly risks, and the industry, the merchant industry has not yet, I'd say, um, arrived at a framework or a set of standards for how agent agentic commerce should be conducted. And therefore, that you know, when with that ambiguity presents risk and opportunity for exploitation by fraud actors. So I would say the early adopter consumers are are certainly experimenting with, you know, trying, um, but the majority of the mainstream have not yet gone there. I saw some research yesterday by Edgar Dunn that says by 2030, agentic commerce-driven transactions will be over$1.3 trillion. This is worldwide, um, you know, and versus where it is today in calendar year 2025, somewhere at about 30 billion. So significant growth is expected, you know, over the next five years in agentic transactions.

SPEAKER_00:

I'm trying to equate that kind of adoption to, and by the way, that doesn't really work, but I'm trying, the the adoption of say chat GPT or any of those kinds of uh GPT level assistants. Um, at first they were just, I'm gonna toy with this and see what works. And we've seen just a just a giant spike in adoption and usage for all kinds of things, and you to the point where you know, in many aspects of life, it's ubiquitous, it's just there, it's doing things that you know people, humans were doing uh a year ago, and it's doing it in a big way. And agentic shopping, though, is different. I don't think it really equates directly because there are higher stakes there, somebody's playing with your money. So I think what that means is there's going to be a slower adoption curve, and people are gonna be a lot more sensitive to ideas like fraud and security along the way, which to me means that people like SIFT are gonna play a larger role in that kind of broad adoption in a gentic shopping.

SPEAKER_01:

I I I agree with you, I absolutely agree with you. I would say yes to your first point, that the you know, as a consumer using chat GPT, the risk is relatively low as a consumer, right? You might get information that's too generic. Um, you're not going to use it. If you're getting medical advice, maybe there's some risk there, right? If you're following what ChatGPT is saying, but the risk, generally speaking, is low and the ubiquity is pretty high, right? It's very easy to go out and try ChatGPT. In the world of agenti commerce, yes, the perceived and actual risk is higher, right? Because you're dealing with your own finances, your own effectively, you know, approval of a transaction that you're not actually involved in. Um, and therefore, yeah, the risk is the risk is much higher. And I think what what it's going to take to get the mainstream going is I think a lot of good industry cooperation. We're seeing the card networks like Visa and MasterCard really going hard to try to develop standards and lead the way that work with the banks, that work with the payment processors, that work with the solution providers like SIFT. So this is one of those industry level waves that will require a lot of coordination and some standardization in order to get this right.

SPEAKER_00:

Um, I'm going in prognostication mode, which is never, never good. I always look like an idiot after saying stuff like this, but but here goes. You know, um, as agentic shopping starts to take hold, as people like you and me start to kind of feel our way into that world, we are, I think, out of necessity, because we're smart, prudent people, we're thinking about how safe is this agent, not only with my money, but how safe is this agent in terms of getting me the right thing? You know, what when the box arrives at my door, what am I really going to see here? Because we're no longer afforded any of those gut level sensitivities in, you know, looking at somebody at the checkout counter, and you know, if they're not looking at you in the eye, well, that's a signal, you know, and you you don't register those things conscientiously, but or rather consciously, but you you do register them. And um, all of those ways that even right now in e-commerce, we see, you know, the merchant has been rated, you know, star ratings and reviews and those kinds of things. An agentic shopper, right now, it's just a really cool tool, but but how will how will we understand how we can trust these things?

SPEAKER_01:

Yep. And that's where companies like Sift and others play a role. And so back to the bigger topic, again, where there's an opportunity, a fraud actor will exploit, right? We saw that even before the dawn of digital commerce, right? Or loss prevention, still a thing, right? But there was loss prevention and that was managed in the world of digital, different types of fraud, right? Credit card theft, account takeovers, policy abuse. There's all different types of fraud. And the word of in the underlying way that providers like SIFT do what they do is to really understand the identity of the user behind the transaction. Not just for that transaction that's taking place right now, but over a long period of time across a network of merchants that that we work with. That's what we do. We run a global data consortium. I often use the fra the phrase, it takes a network to fight a network. So assessing a transaction, be it being it from an agent or even an authentic user in the moment, you only have limited context. But when you have the benefit of looking over the last seven years across a diverse set of industries, as a solution provider, and we can we have that data, we can make a better, more informed decision about the trustworthiness of the consumer behind that transaction. And so you apply that into the world of agents, you think about agents will have identities. There'll be an identity associated with the agent that will be commissioned by a consumer, right? So the consumer associated with the agent, there's a relationship there. But the agent on its own will start performing transactions and will start to build that longitudinal view of that agent or derivatives of that agent and have the same framework, the same risk framework rooted in identity to establish that trust over a period of time.

SPEAKER_00:

Which sounds like it's a heck of a lot more uh valuable and reliable than simple star rating ratings and reviews, which can be gamed. So you're you're applying, I think what I'm hearing is you'll be applying the same technology and thinking to uh to individuals as than you will to um agentic shoppers.

SPEAKER_01:

That's right. The framework will be similar. Some of the techniques and data sets will be unique, but the framework for assessing effectively what is an identity will not be uh will not be dissimilar.

SPEAKER_00:

That's really interesting. Um so I I want to talk a little bit about some of the shopping cohorts. I'm cherry picking right out of your your research here, but I found this one fascinating. Why do Gen Z and millennials who are supposedly digital natives fall for fishing attempts at higher rates than older generations, say like me? Why is that? I mean, you know, these are the people who are supposed to have the innate understanding of how that world works. Why are they falling for these things? Right.

SPEAKER_01:

I I think people like you and I of our of our you know age cohort, I think we're a little bit more you know conservative in our you know in our trust, right? Um, and so I would say just you know, I have I have kids of that age, you know, of the you know the Gen Z age. And my my explanation for the data, which we did commission, this is consumer level research, um, is they're guards down. They've just they haven't been burned before. Their propensity to want to trust in a digital transaction, even without a face-to-face um you know interaction is is just higher. And so that's I think that's my they're not as jaded, I guess would be my okay. Let's go with that.

SPEAKER_00:

Yeah. Interesting. Yeah, they just don't have the the scars to be a little bit more guarded that yeah, and and you and I, I'm I'm painting a rather broad brush here, but you and I, I don't believe our digital natives. I'm certainly not, but I think I'm way older than you. So I am not a digital native, yes. Okay, all right, so we have that. So yeah, I I think we're coming at it from a different a different lens altogether. Yeah, um let me ask you something. If you could wave a wand and reinvent how digital trust works, what would that look like? And I'm we've touched on some of this already, but the world that you operate on, frankly, that I do too, is changing so radically, so fast. Gen tech shoppers and AI based this and that, and consumer expectations for quality and privacy and security um are all being ratcheted up wildly right now, and I think they will be for a while. Like, how how does a company like SIFT navigate that? How do you prepare for it? How what do you tell people like me at street level, hey, here's what's gonna happen?

SPEAKER_01:

Well, you you asked a question a little early, you know, in the beginning of your question, you know, the utopian view of how the digital ecosystem and digital universe should work or could work from from some people's perspective, is that consumers like you and I have control over our digital identities. It's what's called self-sovereign identity. We own, we maintain, we curate a digital identity that is then usable ubiquitously everywhere. And there's been attempts to do that. Uh, and it's really, really hard, right? Because that requires standardization. It requires massive consumer adoption of using a very specific framework. And I I don't know if we'll ever see that outside of uh outside of where it's forced, like you see that in certain government services, right? Where you must maintain an identity and you can use that identity across other government services. Right. But in the commercial and the private sector, I we just to get that level of cooperation. I say it takes a network to fight a network, getting that level of collaboration right now is a bridge too far, which leads us to okay, what is the then the next best way to protect consumers and to protect businesses? And that's really to put the responsibility into the hands of the merchants of assessing risk. And so that's where we've been largely right since the dawn of the digital commerce era, where merchants of all shapes and flavors, including banks, including fintechs, including merchants, are now responsible for assessing risk and making decisions in real time. And so consumers should know that every merchant again of all types are investing significantly. And it's not just to stop the bad behaviors, or someone who's purporting to be you, Mike, who might have taken over your account on an airline and and is attempting to do a transaction, or who has compromised your account and is attempting to move loyalty points from your account to somewhere else. That that is all being protected right now. But but the the positive tone is that as a consumer, because there are these controls in place with deep data, is I have a better chance as a good consumer, and most of us are good consumers, we have a better chance of being treated well and not you know avoiding customer insult, right? Not being stepped up with an extra you know factor of authentication when we're clearly we clearly have a track record of being a good customer. So the experience should be better as time progresses because the data and the tools are so uh prevalent across the industry.

SPEAKER_00:

Two to three years out from now, what do you think will be the defining challenges in consumer fraud? And that's too far to look out. I know that it's not even an I could never answer that question, but maybe you can.

SPEAKER_01:

There is a um duality that we're seeing in in the fraud landscape. On one hand, we're seeing highly organized fraud. Most fraud at scale is highly organized by fraud organizations based in all parts of the world with organizational structures. They're funded, they have capital, they have technology stacks, they have teams of people executing frauds of different types, account compromise, credit card theft, identity theft, right? This is that is one big source of fraud. In contrast to that, we're seeing in recent years the rise of what we call first-party misuse. First party abuse, you might hear of it as friendly fraud, right? And so it's individuals like you and I that are otherwise trustworthy, kind of going rogue for a moment, right? And that might mean we're abusing a sign-up bonus policy. Let's just say it's a gaming company. And if you create an account, you get$50 of gaming credit deposited to your account. Like that's intended to be used once. We're seeing consumers that again that are otherwise good attempting to abuse that policy. Interesting. Or a return policy, right? There's there's massive abuse of return policy. Again, otherwise good consumers that are abusing it, they're qualifying for the refund, they shouldn't, and many other types of what I'd call first-party misuse, or as the industry calls first-party misuse. So it's this duality of highly organized crime and then the rise of first-party misuse. And um, and I think that first party misuse is growing very quickly, from what we're seeing. And so it's catching that, it's catching that because you know, these are consumers like United. We're otherwise good, trustworthy, high volume, high-velocity transaction users that will occasionally do something we shouldn't be doing.

SPEAKER_00:

Interesting. And that that's very different from someone who just kind of mistakenly falls down the wrong rabbit hole, and you know, you've compromised the password or something like that. It's a totally different scenario there. Yes, that that makes me like I'm just not smart enough, I think, to be thinking in those terms, which is my safety net. You know, I have that going for me. But um, you know, interesting the way people can think about gaming the system. Yes, it it happens, and it's happening more frequently. Wow. Well, Armin, I really appreciate this. It's it's enlightening. Um, this is this is fascinating, and I think it's really important. And anybody who listens to this podcast or reads what we put out on customer land is touched by these ideas in one form or another. So I would just encourage you. Um, we'll have a link to SIFT's site, we'll have uh we'll uh put links up to some of the more recent research, but it's great stuff. And again, uh Armin Negerian, who is CMO at SIFT, um, thanks for the conversation. Thank you so much, Mike. I appreciate it.

Head Shot

Mike Giambattista

Host