Solving Retail’s $1.7 Trillion Inventory Problem With RFID and AI

Hello, everyone.

This Retail Technology Spotlight series podcast is brought to you, of course, by the Omnitalk retail Podcast network.

I am Chris Walton, your host for today's interview and interview in which we are going to explore the topic of blind spots.

There's a lot of talk around AI, but there is one big blind spot in most of our retail operations that is still likely preventing us from harnessing AI's full power in the physical world.

And I am of course talking about inventory accuracy.

They can only be as good as the underlying data that's fed into them.

These tags now need lesser power from the RF waves coming from the reader to wake up and say, hey, I'm here.

I'm tag number so and so.

When I just step back and think about it, the implications just even on food safety are enormous.

What's happening is rfid.

And the use of RFID in the retail environments is also evolving.

Quality of your data matters.

Trying to scale what we're talking about across thousands of locations, which many retailers operate, is never easy.

I love talking about the intersection of technology and people.

Inventory distortion isn't a new problem, but it's a stubborn one.

Out of stocks, frustrate shoppers, overstocks eat margin.

And for years, the tools we've had to fight it have been slow, manual, and honestly, also a little bit of a guessing game.

IHL estimates, for example, that retailers lose an estimated $1.7 trillion annually.

That's trillion with a T to inventory distortion.

So how do we finally solve the 1.7 trillion dollar inventory distortion problem?

And what does it take to actually see every item in a store in real time?

And to help us answer those very questions, I'm pleased to introduce Gagan Luthra, the VP of Product at impinj and Goggin.

Welcome to omnitalk.

Thanks for being with us.

Hey, Chris, thank you for having me.

Happy to be here today.

Yeah.

No, so I'm curious.

So, you know, I spend more time around RFID than probably your average bear, but not nearly as much as you all do.

And there's probably a lot of people listening that aren't, you know, steeped in RFID day to day, you know, as you and I are.

So how would you sum up why inventory distortion has been such a persistent problem in the industry?

And, and, or, and also what I know was I.

Was I too hyperbolic in the open?

Or do you think I was not hyperbolic enough?

I think it's interesting, right?

I would claim that you undersold it.

I undersold it.

I did.

Okay.

Okay.

Why?

Tell me why.

You know, we hear this all the time from our customers, our partners out in the industry, especially in, you know, the retail industry.

Inventory distortion has been such a persistent problem for retailers that they are ultimately making decisions with an incomplete view of what's actually happening in their store operations.

At the gist of it, it's actually a data problem much more than an inventory problem.

Just because a shipment manifest says that this package was on this truck to be delivered to store X, it doesn't mean it made it to the store.

Or just because a store's, you know, inventory management says a product is out in the front room, doesn't mean it has, hasn't been, you know, moved, misplaced or worse yet lost.

So retailers have spent, you know, millions if not billions of dollars spending these building these IT systems, these database management systems and analytics on top of that.

But that's typically based on, you know, snapshot based data, PO based data.

And these systems then end up forecasting what they're missing.

And this is where certain retailers have stood out is they've been able to make that leap from paper based data to actual item based inventory data.

And then when you apply technologies like RFID throughout your supply chain to have true item level visibility of the physical items that, that are in your retail environment or anywhere in your network of in inventory supply chain, that data accuracy becomes a lot more important.

That data accuracy climbs up and then the, the agentic AI systems that are trying to process that data can do so much more with it.

So it's ultimately, I believe, a data problem and it is a big problem in the industry.

Right, that's so interesting.

I've never actually heard anyone say that the way you just articulated it.

You know, in my 30 years of retail now, I've never heard anyone say that.

It's interesting too as I think back to like reading through, you know, earnings reports and everything, when I hear people talk about inventory problems, it gives me a new len lens by which to evaluate those statements and say like no, you probably have a data problem on your hands and how are you thinking about your data structure and your infrastructure and your technology infrastructure to combat that data problem.

So that's a, that's a great nugget right off the get go.

Goggin.

All right, so I want to talk a little bit about Gen 2X.

You know, Gen 2X.

What is first of all, what is Gen 2X in plates, plain terms?

And what does it change about how the dynamics of how RFID reader and a tag talk to each other.

At its simplest level, these RFID tags are wireless item identifiers used to identify these items and then eventually track them.

One of the best things about RFID is that it's based on a standards based technology put forth by the RAIN Alliance.

And that standard is currently on Generation 2, version 2, or as the industry commonly calls it, Gen 2v2.

You can think of this Gen 2v2 standard as the foundation of how these tags and scanners talk to each other over the air wirelessly.

And that foundation has been great for the industry because it's allowed for interoperability, it's allowed for larger scale deployments at an industry scale and so on.

But the enterprise retailers want more.

They want to get more out of their wireless trackers and identification labels.

And that's where Gen 2X comes in.

Gen 2X essentially builds upon this foundation of Gen 2 to do two things between that wireless link, between the tags and the readers.

It makes the link more reliable, which, which you know, accrues to better accuracy in your, in your read results.

And then it makes the link more efficient, which means you can do reading operations faster.

Right.

And both these things, when they work together, make these much larger scale enterprise deployments a lot more resilient.

As retailers are putting RFID tags in the thousands and tens of thousands into their stores and RFID scanners all over these deployments, their environments are also getting more challenging.

They're tagging smaller items, the density is getting increased.

So this is where gen 2x comes in and helps at that fundamental radio link to make that communication stronger, more reliable, faster and more efficient.

So essentially it's a physics problem that you're talking about.

Right?

So like, explain that more to me like in terms of like, you know, what you were able to do given the physics limitations prior to now with Gen 2X, what does it enable the retailer to do as they're thinking and listening to this?

Yeah, you know, and it's a great question because as much as we can talk about physics, ultimately the retailer doesn't care.

Right.

They do have an appreciation for the physics and the technology that they're using.

But at the end of the day they want to know whether it's going to work for them.

As I was just saying, as these environments have gotten more complex, as you've deployed more tags out in these retail environments, ultimately what you want out of your systems if to give you better data, so, so that you have better inventory accuracy.

Right.

And that's exactly where Gen2X comes in at the technology layer.

What Gen2X does is it applies a more advanced modulation technique between the tag and the reader.

Right.

And because of that advanced modulation technique, the physics is better, the radio communication link is better, and what's happening is rfid.

And the use of RFID in the retail environments is also evolving.

It's evolving from these snapshots, these breadcrumbs at moments in time, small bits of data that would tell you where things were last seen to.

Now a more continuous stream of data so that you're making real time operational decisions based on these data streams.

Gen 2X, with its improved radio link, its improved connection accuracy, its improved tag link efficiency, allows for those continuous data streams that unlocks new use cases beyond inventory management.

Got it.

So it makes the signal more reliable, the reading of the signal more reliable at the end of the day.

Okay, I want to go into something you said too.

I want to follow up on something you said because you said that you said retailers are making more of an effort to tag smaller items.

And I can remember that back from my target days doing some work, trying to figure out how to tag like little things in beauty, like little vials of lipstick, which was very complicated.

I don't even know if vials the right word goggin.

But you know, you know what I'm saying.

The little tubes of lip tubes of lipstick, I guess.

And so like I want to understand that better.

Can.

What are all the, what are all the challenges that come?

I mean, you gotta, the tags have to be a certain size, right?

But you know, like what are all the challenges that come with trying to get RFID to work on smaller sized items?

That's a great question.

And there's actually two things working here that are making it harder to read.

Tags, future generation tags, smaller tags and so on.

And how the industry is responding to the technology that needs to service it.

The first thing that's happening is, as you said, the tags are getting smaller for most, let's say retail apparel, where you've got hand tags, you've got the physical real estate to put in a large antenna on top of your RFID tag.

A larger antenna means it can communicate further, it can be heard easier.

The basics of RF and physics communication.

Now retailers want to put tags on smaller items.

You can't always have a large hang tag on a cosmetic like your vial of nail polish or lipstick, if you would.

Retailers are also embedding the tags into the packaging itself as opposed to having hang tags.

So you start getting limited by These form factors and that shrinkage is happening, that shrinkage is causing the antennas and the inlays to get smaller, which means they become harder to read, right?

They become harder to hear in a noisy environment.

So this is happening in the industry.

It's because the retailers are demanding it and the industry is pushing this innovation out there.

But what it means is that communication technology that Gen 2 and Gen 2X that we were just talking about again needs to be able to respond to that change, right?

So that's one change happening on the physical side of the tags with them getting smaller.

The other thing that's happening with the tags are they're also getting more efficient right?

Now recall, these are passive battery free tags that get powered on by the radio waves that are looking for them from the readers, right?

So the generation over generation, these tags now need lesser power from the RF waves coming from the reader to wake up and say, hey, I'm here, I'm tag number so and so, which means they're getting more efficient.

But typically how these RFID systems work is, you know, imagine a crowded room and I'm a reader and I'm shouting from across the room, hey, tags.

Anyone here speak up.

And the tag essentially whispers back, I'm here, it's me.

The reader's got to be able to pick that up.

Now you can imagine as the tag gets smaller or as the tag gets more power efficient, that whisper coming back across the room is even more faint.

The reader has to do a much harder job to be able to pick up that link, that signal and hear these texts.

This is again where technologies like Gen 2X come in, where they help with that radio link while also meeting the emerging needs of the industry with folks wanting to go to smaller tags or putting tags onto the items in itself which require more efficiency in the silicon.

Wow, wow, that's a great, great explanation, Gogit.

Like, man, you're very, very articulate and very do a great job of explaining it in like, just very easy to understand terms for me particularly, and hopefully for my audience too.

All right, so, but you know, the one thing I would ask you though, because talk is cheap at the end of the day, you know, and I think our listeners appreciate the candor with which we approach these conversations.

So like, I want to know too, you know, do you have a real life case study that you can share where, you know, you've seen the benefits of what you're talking about in action?

I think one of the poster childs, especially in the retail apparel industry of, of of an enterprise that's adopted RFID and, you know, done wonderful things with it and become an exemplary in the industry is really the Inditex group, you know, with the Zara brands more popularly known here in the U.S. but you know, globally, really, they've got malls, you know, retail outlets across the, across the globe.

But what they did, you know, beyond just putting RFID into every apparel item they sell as that individual item level tracker, that was just the start.

But what they did was they put it at the very onset of the item being manufactured.

Right.

And then they've set up the infrastructure in such a way that it tracks the lifecycle of that item all the way from manufacturing through transport warehouses and distribution centers into the retail stores, all the way through point of sale.

Right.

And this has allowed them to do things like reduce their inventory counting time.

You know, this is, which is the labor savings from several days to a few hours.

It's allowed them to take their inventory accuracy from 65% to 95, 98%.

And one of the most interesting metrics for me, which really stood out, they made a public statement on this just recently.

Their unsold inventory levels were sub 1% and if I recall, closer to half a percent, which is astounding if you think about it, because the rest of the peers in the industry, when you look at global fashion apparel retail, they average upwards of 15 to 20% of unsold inventory.

The Inditex group, with the benefit of item level identification and tracking, with the deployment of a system throughout the supply chain, and with the building of IT systems and data management systems that can organize this information in an easy to consume manner for their enterprise business operations, has allowed them to get down to the sub 1% of unsold inventory.

And it's really fascinating to see that proof out there.

And the company crediting RFID as one of the key reasons it got them to this stage.

And it all, all of this just translates into operational excellence and then ultimately business outcomes and revenue.

Yeah, no, that's a, that's a great proof point.

Especially considering just how fast fashion they are in terms of their business model.

Being able to hit those numbers is pretty incredible.

Yeah, I mean, I, I personally saw, I went to their, their flagship store in Madrid a few years ago and was blown away by how they were utilizing RFID within that store, even to the point of like knowing what items are going into and out of the fitting room, you know, like which items people are leaving behind, which is really cool.

And you think about it from From a merchandising standpoint, to help you understand like, you know, what's, what's catching people's eye, but what's not fitting their bodies as they're having that experience, ultimately having.

That, that technology identifier allows them to convert, you know, them or anyone else in their position to convert their physical retail operations and have a lens on it similar to an online store.

Yep.

When you and I shop online, they know exactly which item I spent how many minutes looking at, what I added to my cart, what I removed, and all these behavioral patterns about a customer's shopping behaviors, if you will.

And with rfid, you can translate that at a store level.

You know, what people picked up, what they took to the fitting room, what they purchased, what they walked out of the store with as well.

So it's really fascinating what you can do.

Again, going back to data.

Right.

You talked about inventory as the opening problem statement.

Data accuracy fixes that.

And all this work about agentic AI systems, which is fascinating.

They can only be as good as the underlying data that's fed into them.

Right, right, right.

Which is actually no different than the underlying problems we've had in digital commerce.

As you think about retailers trying to light up different supply chain capabilities, like shipping from store, pickup in store, all those same things get exacerbated by what we're talking about.

Okay, so I want to shift gears a little bit to close us up because I've got a real expert here in you goggin.

And I want to know, like, where are things going next?

Like, what problems are still out there that have not yet been solved but are on the cusp of being solved?

Like, I'm curious to get your take on that.

Like, how do we continue to fight this battle and win it, so to speak.

From my perspective, you know, this, this item level visibility really becomes a foundational and fundamental capacity that all retailers will have in the future.

Right.

100%.

We'll already seeing this well deployed in retail apparel.

Right.

As I talked about this segment of the industry being the poster child for adoption, but we're seeing adoption because it's still very early in this, in this innings in some ways.

Right.

So we're seeing adoption across, you know, health care, across food and grocery, especially as it relates to traceability, recalls, authenticity.

We're seeing great adoption in supply chain and logistics because it's not just about what's in my store, it's about how it got there and all the places it had to get through in that complex network of supply chain.

That's one trend happening, expanding to different categories, different industries, and they all come with their own set of challenges.

Different items to track, different business processes to work with.

The other thing that's happening is RFID has historically been a replacement for barcode and thereby a replacement for counting my items and getting inventory as the technology curve moves.

And we talked about Gen 2X and a lot of things you can do with Gen 2X, retailers that have already adopted and done the baseline inventory management with RFID are getting into the next unlock of applications.

Right.

Not only can they breadcrumb attack to know that it was last seen here, you can now accurately talk about which specific zone a item is present in.

You can talk about an item's movement, you can talk about finding items, you can talk about items transitioning through certain areas or process workflows.

So this is really putting a lot of demand on what that radio technology can do and what information you can get out of it to do unlock some of these additional use cases.

And Gen 2X is helping with that.

Others in the industry are working on it.

Ultimately, this is creating.

It is evolving from going from those data snapshots coming from the breadcrumbs to a continuous stream of trusted data.

And this is why at impinch, what we are working on is bringing all of this technology together as part of the impinch platform.

And we're calling it physical intelligence.

The intelligence that your physical items in the stores need.

The data that helps process that information that ultimately feeds your agentic AI systems is the next unlock.

And we're working towards that with the impinja physical intelligence platform.

And there's a lot of innovation still to be had and to be made.

But the industry is ready for it.

We see a lot of excitement and, and a lot of momentum towards it as well.

Got it.

So, so set.

So set another way.

And it's kind of funny because we're 25 minutes into this conversation.

We haven't really talked about agentic AI yet, except in the tease in the beginning.

But so said another way you're saying like what's next is the application of AI principles into the data collection of this to then decide what the best courses of action are that can be handled however a retailer decides to handle them.

Exactly.

I think then, you know, the industry talks about AI as, as a, as a general term, AI means a lot of things.

But you know, there's intelligence at the edge.

So knowing what, walk to your exit doors as a stolen item or knowing what's in Your point of sale, self checkout basket.

So there's intelligence being applied at the edge to make sure those use cases are 100% accurate, reliable accurate and so on and so forth.

There's AI being applied at the aggregation of all this data at a store level, at an enterprise level on how supply chain is flowing.

And typically the industry has done a lot of analytics on that data.

Now they go into orchestration based on that data.

I was just going to say that word orchestration 100%.

That's the next big unlock.

You already see my supply chain.

Now you trust the data because it's based on items, physical presence.

What decisions can you make for me?

What orchestration can you do from system A to system B and the handoff in between so that I can do automated replenishment so that I can meet my customer promise on BOPIs and things of that sort?

Yeah, right.

Yeah.

God.

The implication when I just step back and think about the implications just even on food safety are enormous.

As you think, think about how you're talking, as you think about how what you're talking about in practice.

There was a good example of that recently, I think in the, in the press and it was the Chipotle Group, right.

When recent outbreaks happened in the, in the United States on, on food safety, the Chipotle Group was one of the enterprises that was able to trace, track and stop those items from being further distributed to customers much before anyone else could.

Right.

And it was fascinating again attached to item level visibility as opposed to paper based visibility on your supply chain.

Right.

100%.

All right, so before I let you go, let me get you out here on this one last question.

So you know, I mean, I've been covering, I've been covering this topic pretty religiously for the past ten years almost now.

And you know, the one thing that I found about it is that trying to scale what we're talking about across thousands of locations, which many retailers operate is never easy.

So I'm curious, as the VP of Product at impinj, what are the biggest lessons you've learned over the years in terms of getting it right with the retailers with whom you deploy your solutions the first time out?

I think one of the.

There's two lessons that I've maybe call out here.

The first thing is to always start with the business problem.

Right.

Many of us often in the technology world, start with the technology and see where you could apply the technology, how you could get the most out of it, the best out of it.

You know, Adam Pinch and myself personally as well, always thinking about starting with the business problem first and then applying technology or a combination of technologies to solve that business outcome.

And often the case, you know, with rfid, that is, that is how things happen.

RFID becomes a complementary technology to, let's say vision or barcodes and other sensing based technologies that all work well together.

When you've started with the business problem that you want to solve that problem in this case as an example, could be loss prevention or inventory management.

The second lesson that I've learned, and I'd say we've talked about it a bunch today, is the quality of your data matters.

Right.

Ultimately, where the world is going today with AI based agentic systems, enterprises are making decisions based on their enterprise data that they have available to them.

So the quality of that data is of the, you know, utmost interest to enterprises to make sure that it's accurate, to make sure it's authentic.

And they all have a good baseline to work off of.

I think keeping these two things in mind is a recipe for success, or at least has been for me and for Infinch.

And yeah, that'd be my words of wisdom on it, if you will.

Awesome.

Awesome.

Well, great stuff, man.

Great stuff, man.

This is the exact type of content that I love doing on Omnitalk.

It's just so great to have you and so great to pick your brain on a topic that is near and dear to my heart.

A topic I'm really passionate about too.

So, um, I'm guessing, you know, some of our listeners may want to reach out to you Goggin.

What's the best way for them to do that?

Yeah, hey, I'm on LinkedIn, you know, look me up, shoot me a message connection intro, Gagan Luthra and I'll be happy to engage with you folks.

Yeah, I heard you're pretty active on LinkedIn too.

I'm going to have to give you a follow and check you out.

Like you, you, you, what do you.

What, what all.

What types of content are you putting out there?

No, I love talking about technology.

I love talking about technology and the intersection of technology and people.

You know, at the end of the day it's, that's what motivates me and drives me.

So yeah, I love the LinkedIn is a, is a great place to just, you know, stay abreast with what's going on in the world, in the tech, tech world.

So I love sharing my thoughts on that and engaging with others as well.

Yeah, darn right.

Darn right.

I second that.

For sure.

All right, well, Goggin, thanks so much for joining us.

For everyone out there, today's podcast was produced, of course, with the help and support of our fabulous producer, Ella Sirjord.

And as always, on behalf of all of us here at Omnitok Retail, be careful out there.