How Retailers Are Transforming Customer Experiences with Data & AI

00:00:00:00 - 00:00:34:22

Tori (Online Retail Today)

Welcome, everyone, and thank you for joining us for How Retailers Are Transforming Customer Experiences with Data and AI. Presented by Online Retail Today and sponsored by Orium commercetools and Google Cloud. I’m Tori, the webinar coordinator at online retail today, and I'm excited to bring you this fresh session with well-earned insight on the transformative potential of AI and data analytics and shaping the future of omnichannel personalization and ecommerce.

00:00:34:24 - 00:00:55:28

I'm really looking forward to hearing from David Azoulay, Marc Stracuzza, Román Tejada and guest speaker Sucharita Kodali and what I'm sure will be a valuable discussion. Just so you know, we will be recording this webinar in case you have to leave early. If you happen to miss any part of today's session, you can go to the landing page to view the recording.

00:00:56:00 - 00:01:04:10

We’re going to send that link in the chat right now. Up next.

00:01:04:13 - 00:01:34:08

Before we go any further, I want to thank Orium commercetools and Google Cloud for sponsoring this webinar and helping us to make this happen. Or is North America's leading composable commerce consultancy and system integrator. They specialize in composable commerce, customer data and retail platform engineering, commercetools. Inventor of the Headless Commerce is a digital software provider that empowers organizations to embrace innovation and thrive by providing flexible APIs and enabling agile, customizable commerce infrastructure at scale.

00:01:34:11 - 00:01:54:07

Last but not least, Google Cloud accelerates every organization's ability to digitally transform its business. They deliver enterprise grade solutions that leverage Google's cutting edge technology, all on the cleanest cloud in the industry. So thanks again to our sponsors. Up next.

00:01:54:10 - 00:02:11:02

Let's get some technical things out of the way. Please don't hesitate to ask questions during today's event. You can do so by submitting your questions into the Q&A panel found in the bottom bar of the webinar window. Great questions make for a great webinar, so don't forget to stick around for our Q&A session to get even more fresh insight from our panelists.

00:02:11:05 - 00:02:31:01

The chat is also available for open discussion, but please know we will be using the Q&A panel for your specific questions. My wonderful colleague Tara, will be fielding your questions today. We'll be happy to answer anything technical you might have. So feel free to pull up the Q&A panel, say hello and let her know that you're listening. Closed captioning is also available for this webinar.

00:02:31:04 - 00:02:50:29

To enable this feature, select the “More” button in the bottom bar of the webinar window and select the captions option. Lastly, if you have any audio issues during today's presentation, you may choose to dial in by phone or dial-in information can be found in your webinar confirmation email from Zoom. This information is also up on the screen and Tara can relate it to you at any point if you need it.

00:02:51:01 - 00:02:53:29

Up next.

00:02:54:02 - 00:03:19:15

Again, I'm really excited to hear from our four amazing speakers. So without further ado, I'm going to pass it over to David to start off the introductions. Thank you Tony. Hello everyone. My name is David Azoulay. I'm director of product at Orium. My background spans software engineering, solution architecture, and thought leadership, especially in the composable commerce space. And I'm looking forward to our discussion today.

00:03:19:18 - 00:03:40:26

Michael. Okay, great. Hi, everyone. I'm Marc, who's the director of portfolio strategy at commercetools. I've been a product development leader for more than 25 years. Focus on product solutions, software development, architecture, product management, development management, kind of. All of it. Looking forward to this conversation today. And I'm passing it over to Román. Thank you. Marc. Hey everyone.

00:03:40:26 - 00:03:57:25

My name is Román, and I'm going to upload specialists at Google, working with our largest retail customers and helping them accelerate their existing roadmaps with AI. Before Google, I spent two years at AWS and the maker product team, and I really spent most of my career building products and bringing them to market. So I'm super excited about this conversation.

00:03:57:27 - 00:04:25:01

Now let me turn it over to Sucharita. Hello, everyone. I'm Sucharita Kodali. I'm a retail analyst at Forrester. For those of you who may not know, Forrester. Forrester is a technology research company, and I have been working with a lot of our retail and brand manufacturer clients, to support them and, their various, go-to-market strategies, of which these days, a lot of it is about data and AI.

00:04:25:17 - 00:04:49:23

So I just want to start with a framing of, AI versus generative AI, because generative AI is what we're spending a lot of time, in, you know, kind of our current, conversations talking about, but it's a subset of AI. There is AI, which encompasses a lot of different disciplines. Some of the oldest are things like machine learning.

00:04:50:08 - 00:05:20:00

Even expert networks, which really aren't even around that much, but we're sort of version one of artificial intelligence. But there are also, solutions like robotics, which absolutely fall in the AI framework, but really are outside of what we are talking about when we talk about things like generative AI. So but yet you have, you know, PhDs in AI at universities like Stanford and Carnegie Mellon, that are absolutely focused on, you know, kind of each element of that middle layer.

00:05:20:05 - 00:05:44:03

Now, there are sometimes some, nuances between what's machine learning versus a neural network. And often, a neural network, can be defined as a, as a subset of machine learning. It's tends to be a more sophisticated, version that has many more nodes. That, that, that sort of match the input to the output or connect the input to the output.

00:05:44:18 - 00:06:14:27

And a form of neural networks is deep learning. And there are many types of neural networks and deep learning. It has a specific definition that it involves a certain number of nodes. And there has to be a large number of nodes. And once you kind of hit that threshold, it's classified as deep learning. And deep learning algorithms, apply to many different things that we have seen, over the years that, that are relevant in retail.

00:06:14:27 - 00:06:38:27

Some of the most common use cases, of course, are fraud detection. Camera vision is another one. And that's going to probably be much more, used in things like facial recognition or, and identifying products like you'll see in, cashier less commerce store. And of course you have generative AI, which can be, a large language can which can be based on a large language model.

00:06:38:27 - 00:07:02:13

And, you know, typically the use cases that we're seeing are ones that respond to prompts. And, when we are looking at the landscape of generative AI, I just thought it was interesting to, share this content because it is, a number of the different types of, solutions within the generative AI landscape that have been funded by venture capitalists.

00:07:02:13 - 00:07:27:04

And, the example companies on the right hand side and, the approximate amount of VC funding that each of these different line. These categories, these subcategories have captured. Not surprisingly, the biggest bucket is the one that we've all heard about, which is, which is OpenAI, as well as companies like anthropic, which have also, taken a huge chunk of financing.

00:07:27:04 - 00:07:55:17

But the vast majority of that 22 billion is, is OpenAI alone. But you'll also see a lot of other types of companies, that touch retail, things like, copy generation for Marceting purposes. You'll have companies like Jasper and others, don't want to, you know, specifically call out, you know, any particular company, but, you know, kind of just more as an illustration of, what is out there that that retailers are, are exploring and, and looking at.

00:07:56:04 - 00:08:18:05

You also have things like video generation, and there is a lot of promise and talk about things like video generation in retail because of everything from the ability to create, commercials, like television commercials or commercials that may show up on a YouTube or, it could be video that could be, you know, created for, a product detail page.

00:08:18:05 - 00:08:59:11

So, so a demonstration video or, a 360, you know, look at an image. So, so definitely a lot of different, use cases and applications. And I'm going to get to specifically where are retailers, using generative AI the most and in a moment. But what I also want to step back and, you know, kind of give you a bit of, of, you know, kind of the, the forest or worldview of retail is that when we look at retail and retail technologies, we like to break it up into three buckets that, for the most part, are, mutually exclusive and collectively exhaustive of all of the functions within a retailer.

00:08:59:12 - 00:09:22:23

The first is, that which is consumer facing and that's your that the, those elements there where you are touching the shopper. And that could be a store, it could be a call center, it could be your website, or any type of a digital touchpoint. Then you have the field worker facing, capabilities. And those are anything where you have essentially hourly employees, involved.

00:09:22:23 - 00:09:42:26

And that could be the store. So you're talking about the field workers being the store associates or a store manager, a district manager. I mean, anybody who is, connected to the field, it could be a warehouse worker. It can be a call center worker. And then you have the knowledge worker facing, functionality, and that's everything that's back office that essentially sits at headquarters.

00:09:42:26 - 00:10:08:17

It ranges, from everything from from it to Marceting to shared services like accounting and procurement. So that is, essentially every function that sits within a retail organization, regardless of you, I was like literally like going through or structures and looking at like who the executive committees of, you know, kind of various retailers are and everybody can sit somewhere, you know, kind of on this chart.

00:10:08:17 - 00:10:29:24

So, what I wanted to do is now walk through for each of those three buckets. Where is generative? I, you know, kind of most impactful now and where even the opportunities to bring it to, to, to have an impact. But even more broadly than that, what I have we've been seeing in these different, areas.

00:10:29:24 - 00:11:02:16

So if we start first with the consumer facing functions, I want to look at what do we already have that is AI or automation related. That is widely deployed. What do we have that we've been talking about that touches on AI. But, you know, kind of may not be as common. And then what's now new as a result of everything that we have seen getting funded about gen AI or, that is is now available because of large language models, that, that are and are out there.

00:11:03:03 - 00:11:30:08

So you'll see that when we look at what's already in wide deployment and can be characterized as AI, we have things like recommendation engines, and we've had early versions of chat bots that have often been pretty decent for the needs of a lot of retailers, because what it does is basically helps to answer either really basic questions or can essentially decide where to, where to channel, a particular chat.

00:11:30:23 - 00:11:49:11

And you also have, of course, you know, the phrase, which probably I've spent more time talking about than just about, you know, kind of anything in my time at Forrester is personalization, right? And data and digital experience optimization. So these are things which I would argue at this point in time are table stakes for retailers and brands.

00:11:49:11 - 00:12:15:28

And if you don't have these things, like you need to get those things, up and running and, and as excellent as they can be, first, and then you have some of these less common solutions that are out there, things like autonomous stores. Yeah. You know, kind of camera vision or biometrics for age verification. So those are, you know, obviously for, for for reasons related to cost or privacy, they, they are, not nearly as broadly deployed.

00:12:16:04 - 00:12:36:27

And then we have a lot of things that are now new and exciting now, whether or not they're going to be, you know, kind of game changers, TBD. But, but, you know, they're out there and, you know, those that are early adopters are absolutely looking to explore and potentially experiment with some of these new things, which could be everything from generative photos, a model.

00:12:36:27 - 00:13:13:05

So you've seen, probably some of the, the, the media goals that have happened around that. I think Levi's had one, where they were, you know, kind of generating, models. Or they were generating photography of items and for, for, also for the purpose of advertising. And it was a little controversial because, but, you know, there was that think there was I think some of it had to do with, you know, kind of your actually, you know, kind of appropriating somebody's IP, you know, their, their, their image, their likeness, and you're not paying them for it.

00:13:13:05 - 00:13:55:11

And these are people that, that, you know, kind of absolutely, should, should be compensated for what they are doing for, for a brand, same thing with, other things like personalized video. Greetings. We're seeing I that's an interesting one because it gives, retailers the ability to clientele like, that's like personalization at scale. This, this vision that, you know, kind of a store manager could provide a personalized greeting to every new member of a loyalty program that is in a local, you know, kind of store or, you know, kind of your you may be able to personalize, a message that is, promotional one based on

00:13:55:11 - 00:14:19:14

what you know about an individual. I generated Marceting. Also something that that people are excited about is the ability to potentially, scale, you know, Marceting copy to, to be as broad as it can be. So now I want to switch, to field worker facing solutions and go through a similar analysis. Of what's already widely and deployment.

00:14:19:14 - 00:14:52:22

And what we see there is also a lot of really, really, table stakes technologies that most retailers that that I talk to most enterprise, level retailers already have and, find essential to their operations. So it's everything from store labor optimization to fulfillment analytics. You know, kind of inventory management, as to what you're going to place in what store, IoT even and sensors like RFID are becoming increasingly mainstream now.

00:14:52:22 - 00:15:16:22

Less common are things like, in-store drones, you know, or any type of a robot that could roam around a store or to take pictures. Mainly of images, mainly off of the inventory, so that if it can identify if something's out of stock or if items are misplaced, and then the hope is that then that would get fed into something like a task management system.

00:15:16:22 - 00:15:46:07

So the store says he could get a, a notification to go fill that shelf or to move, you know, kind of this item that has been misplaced to its proper helm, that type of thing. So what's new due to Gen AI or other neural network developments? Certainly we're seeing things like camera vision. There's a company, for instance, called neural, that looks at camera vision, particularly for exceptions in a manufacturing setting.

00:15:46:27 - 00:16:20:03

So it's looking for things like that that could be safety aberrations or, you know, problems that, you know, kind of an item, didn't quite, you know, kind of meet the specifications of whatever quality control standards that, that you've set, facial recognition for sentiment analysis. This has actually, been around for some time, but, the idea and the hope is, is that, you could potentially maybe tie this to, you know, kind of, and a message that you share with somebody.

00:16:20:16 - 00:16:58:04

I'm also, though hearing that this is, probably one of the things that, while possible, not one of the things that, anybody's going to eagerly embrace because of privacy concerns and, you know, kind of potential legislation that could shut it down. Anyway. There's also things like the ability to use AI to identify sounds or smells, even that could signal, you know, kind of something that that could be, a danger to, to workers or could signal, a problem that may be happening, knowledge worker facing solutions.

00:16:58:04 - 00:17:24:07

And this is where I tend to this is the chart that I, you know, kind of if I were to give a full like, you know, kind of workshop on this, it's usually most of our time is spent on this particular slide. Because this is where most of the value that we've observed in AI to be. And a lot of it is in just machine learning or sophisticated, uses of machine learning.

00:17:24:07 - 00:17:55:04

So leveraging deep learning to, manage things like fraud or manage things like, anomaly detection, which is another big one, particularly for website performance or even understanding, you know, things like, site deliverability. These are all things that, are incredibly valuable, very much standard now in ecommerce and hugely that, you know, kind of they have a huge ROI.

00:17:55:18 - 00:18:28:26

So, you know, some of those other things that I didn't reference already are, solutions, like, of course, Marceting and ad tech optimization. There is a tremendous amount of AI that goes into everything from programmatic advertising to, search engine optimization. So, so those are, that I have yet to meet a retailer, honestly, that that doesn't have some solution that they're partnering with on that, on that, that aspect, pricing optimization, returns optimization and even of course, trends.

00:18:29:09 - 00:18:50:18

And looking at competitive intelligence and I don't see a ton of retailers having competitive intelligence teams, to be honest. But it is something that, those that do find it really valuable and it can often unearth, opportunities for them because they see that, you know, kind of this is, you know, something that you may not have known is is like the secret to success at one of your biggest competitors.

00:18:50:18 - 00:19:16:11

But, you know, you don't have anything that is, is even, you know, kind of vaguely competitive. And it's it offers you an opportunity to think of something new. Less common solutions. I still don't see robotic process automation as deployed as it should be. And if there is anything that should be more widely deployed in retail, it's probably that because there are so many repetitive processes, that just delay things.

00:19:16:11 - 00:19:54:29

And when systems are not synchronized and those processes have not been automated yet, they're still manual. And when they're manual, they just probably take longer than they should. So, and that's something to consider. And then what's new now are, things like generative code. Forrester has a term for it. We call it Turing bots. And, those are, pretty valuable for a lot of the strongest developers, because the idea is that if you can create on average ECS lines of code in a year, you know, kind of I mean, I can enable you to potentially double your output.

00:19:54:29 - 00:20:16:12

And, you know, kind of it's almost like a first draft of something that gets written that you can edit versus having to write something in the first place. Also a lot of, talk about robo buying, robo merchandizing. It's sort of like hands off the wheel. Very similar to what, Amazon does with its inventory management.

00:20:17:09 - 00:20:38:16

So those are some of the things that, that we think are most interesting and have promise. Now, if I were to map all of these things on a two by two matrix, it would look like this. And, on the bottom, on the x axis, it is the maturity of, these various technologies that I referenced on the previous slides.

00:20:39:02 - 00:21:16:13

And then the y axis has the impact. It is the ROI essentially, that retailers have shared that they have seen or expected. Often it's what they've seen if they've deployed it. And, the things that I will say and the upper right hand side are those things that I would characterize as table stakes. And on all of the previous slides, they're essentially pulling from that top bucket of, that top list, like the things that we've had that are already in deployment, that have been proven to be valuable, the recommendation engines, the personalization, the store analytics, etc..

00:21:16:28 - 00:21:47:13

The things that we think have the most promise are the ones in the upper left hand bucket. And these are areas that we, we do think are, where there will be the most likely to have a benefit that is tangible. And that's things like AI generated Marceting. The generative code, anomaly detection is, is one that we're, we we encourage people to, get an anomaly detection solution if they don't already have one.

00:21:48:00 - 00:22:10:07

And then there are things that, you know, kind of they're less mature, and it's still unclear whether or not it's going to have a tremendous impact. It's things like, you know, whether or not even generative video is going to have a huge impact, like some of those examples that I shared with you about the personalized greeting from a store manager or to a person, TBD we haven't really seen a lot of it.

00:22:10:07 - 00:22:46:08

And, you know, it'll be interesting to see if, if people just tune that out because they know it's, you know, kind of auto generated or whether, they're surprised and delighted by the content that, that it can offer. And there may be certain types of applications. It can be generative video that may be created for, you know, kind of certain use cases on demand for things like, in installation as an example, you know, or like, you know, kind of a product, how to manual, you know, maybe that's where the, the greater value is versus in some of the Marceting use cases that we've spent time talking about.

00:22:47:03 - 00:23:11:17

So that, with that and this is, I think my last slide is, just stepping back because, on the create. Well, let me actually go back, you know, kind of, of everything here on this chart, there are only a few that are truly things that could be characterized as gen AI. And, I put those in asterisks, on, on that slide.

00:23:12:06 - 00:23:50:24

But we also what we've also done is looked at a lot of the case studies and examples of retailers and brands that have, experimented or are experimenting with some type of a gen AI pilot and, and looking at what are these people doing and what are different companies doing? I thought it'd be useful to share what most people seem to be doing and, a significant percent the top use case is, is actually when, retailers and brands are doing text summarization, transcriptions, and essentially insights generation.

00:23:50:24 - 00:24:21:27

So they are pushing huge bodies of, you know, kind of essentially files. It could be anything from, you know, I think like Kroger and Estée Lauder. We'll take like a lots of customer service transcripts from, call center calls or they'll take, a lot of Marcet research files and put it into, some type of, of a, you know, private chat, GPT type of, tool and, summarize what those insights are.

00:24:21:27 - 00:25:05:14

Companies like Newegg and Amazon, are summarizing lots of customer reviews in that way. So some of this stuff is internally facing for the benefit of a company's employees. Some of it is externally facing for consumers as well. There's a lot of chat bot, experimentation out there. But one thing I will point out is the slight nuance with chat bots is that they're not always consumer facing some of the chat bots, like P&G for instance, has something called the chat PG tool, which is internally facing, that's for their own associates to ask you know, questions about, products or even it's stitching together, different databases within the company to, to

00:25:05:14 - 00:25:28:22

be able to answer questions for or certain databases, to be able to ask certain tech to answer certain types of questions. Image generation is being experimented with. And then interestingly, what showed up the least but probably has some of the biggest benefit is some of that Marceting content. So, that's an interesting one, probably because, well, you know, I don't really have a good answer as to why Marceting content.

00:25:28:22 - 00:25:52:02

I would think that it would be more deployed. But, you know, maybe people just haven't been as, you know, open with sharing what some of this case studies are and it's actually happening but less common, less commonly seen. So, these are, you know, kind of for your own edification, a bunch of, examples of companies, and across a variety of different sectors.

00:25:52:02 - 00:26:15:18

As you can see, it ranges from grocery to brand manufacturers to furniture to quick service restaurants and what they're doing. And as you can see, a lot of these examples, tie back to that previous page of, you know, kind of falling somewhere within image generation or text summarization or chat bots. So with that, I'm going to hand the virtual mic over to David.

00:26:15:29 - 00:26:36:20

If anybody has any questions or you needed more clarification on these slides, you can reach out to me. It's first initial last name at Forrester. Thank you so much. So to recap, I think this is a great segue into discussing a new whitepaper that we've created around retail data and AI, and specifically in a composable commerce context.

00:26:37:03 - 00:27:07:19

This white paper and we'll share a link in the chat, right now and throughout, this white paper explores how important data in AI is for setting brands apart and driving the growth. The paper presents two main strategies that retailers can use to effectively use their data with AI. The first strategy focuses on leveraging all of the new AI capabilities being introduced in all the various analyst services at play.

00:27:07:21 - 00:27:40:21

Usually, these are customer facing for example, we've seen search providers introduce better performing AI algorithms to improve product search results and performance. We've also seen content management systems introduce AI writing assistants to help, you know, content creators produce more, content quickly. And while maintaining the the brand's voice and guidelines. And the second strategy that this white paper explores is a more comprehensive approach.

00:27:40:23 - 00:28:20:03

It involves harnessing the extensive data, found in all of the services that are part of a composable commerce stack. And which we, you know, which are often scattered and fragmented on their own. But using tools like Google BigQuery, looker Studio, vertex AI, this this strategy aims to consolidate all of the critical data points from these services into a centralized location, which then enables retailers to generate actionable insights that can improve, business outcomes.

00:28:20:05 - 00:28:51:05

And on this, on the next slide, we cover a few, real world examples in the white paper. Organizations that are effectively leveraging their, I, their data with AI. The first example, is Hanes Australasia. Originally they were a wholesale business. They relied heavily on spreadsheets and, you know, inefficient tools. And as many companies, do still today.

00:28:51:19 - 00:29:13:18

But when they started to move into the retail sector, they took the opportunity to modernize their data strategy and embrace tools like Google BigQuery, vertex AI. And using these tools, they were able to successfully enter the retail Marcet and even achieve double digit growth in revenue per session. That was a metric that was important to them. Another example is Macy's.

00:29:14:16 - 00:29:42:11

They have a huge catalog, huge customer base, and they not only needed to meet the current demands, but also prepare for future scalability. And they turned to vertex AI, to enhance their product search and recommendations. And this this move led them to increase their customer engagement higher click through rates. Also boosted their revenue per visit metrics six times is another great example.

00:29:42:11 - 00:30:10:16

They took a unique approach to streamline guest services using vertex AI conversation to create a chatbot. But this chatbot draws it does a lot more than just chatting. It takes, it draws from all of the Six Flags data, and it is available to answer guests and queries for everything from park information to ticketing to suggesting rides and attractions based on the guests personal preferences.

00:30:10:19 - 00:30:41:03

And by implementing this chat bots, they were able to automate a third about, about a third of all their guest inquiries. And that reduced, you know, wait times enhance customer satisfaction. Of course. So all of this underscores that data and AI technologies are ushering a new era of innovation and customer interaction. Retail brands that adopt new strategies today are positioning themselves to dominate the Marcet in the future.

00:30:41:06 - 00:31:20:05

And there are so much more in in this white paper. Please take a look, at the link that's being shared. You can get early access to it and you know, explore these topics further. But that said, I will pass it along to Marc. Awesome. Next slide please. So yeah. So and this is great what we're talking about here because I think what we're seeing and what we're what we're level setting here is within the composable commerce ecosystem, there's this wealth of data that can be leveraged to provide insights and drive decisions and enhance our customer experiences and higher levels of efficiencies and all those amazing things that Sucharita was talking about,

00:31:20:05 - 00:31:39:10

both that are happening now, but are also kind of on the horizon a little bit. And we've talked about that. Right. Predictive AI has been providing targeted help in retail for a long time. For years, solutions like dynamic pricing, payments and auto routing has been around for a while. Search is all leveraged predictive analytics and machine traditional machine learning outside of the scope of any kind of generative context.

00:31:39:13 - 00:31:57:05

And what we've seen with the composable commerce movement is it's opening up access to all of this information, all of this data. So now more than ever, all of your valuable data, all that data that you've got that is being used in all the components is available to you in a usable and structured format, so you can query and aggregate these this information into mixed data lakes.

00:31:57:08 - 00:32:18:03

And these data lakes are going to enable this combination of data to be analyzed as a whole, as opposed to individually, also within your own domain as opposed to the individual products. Doing it for you. Because and the reason for that is because in the last year, we've seen this increasing democratization of access to AI. So in the past, it took this team of data scientists and engineers to build all these analytics efforts.

00:32:18:03 - 00:32:36:06

It took specialized skills. But now, with the support of partners like Google, AI tools are available and becoming more and more usable by everybody. All of us, just yesterday I was playing around with, the Google's new AI builder, which is super cool and, trying to figure it out. But, you know, I haven't developed for, for a decade and, it's super cool to kind of get back and tinker.

00:32:36:07 - 00:32:54:22

I think it's a lot of fun. So the intersection of these composable commerce, the global commerce ecosystem and these accessible APIs, and the data is coming out of the APIs is where this opportunity is, because you can combine all this data from all these sources, and then you could just set these easily accessible APIs on top of it.

00:32:54:28 - 00:33:14:08

Sorry. Easy, accessible AI is on top of it. And this is going to really help you get that deep, deeper level of understanding. So when you take data like orders and carts and product data and pricing and customers things, things that come out of commercetools, for instance, and you and you merge that with your content from your CMS, your content, from your promotions engines, your front end clickstream traffic.

00:33:14:15 - 00:33:40:01

All of this data is going to come together and provide really interesting insights that we're not even going to be able to see. The AI is going to be able to introspect these things that there maybe we're not, that we wouldn't even have a visibility into. So I think what you're going to see is this these insights is going to get turned into, again, the extra actions of personalization and optimizations and and of course, conversational chat bots, which I think is just is amazing that you can talk, talk to your eyes.

00:33:40:01 - 00:33:58:18

I think we saw that it was in 2023 that that, conversational AI was able to, to bypass the average human's ability to do common sense completion sentences, which is why everything's blowing up right now. It's able to do, what? Previously, it was never able to do. So the horizons here, it's all coming into fold, and it's just going to get more and more interesting as time goes on.

00:33:58:20 - 00:34:16:06

And, I think the next slide is for you, Raymond. Thank you, Marc, and thank you for calling out to some of the Google tools here. So I don't have to go through a detailed intro again. So I'm going to go through what an AI infused journey, consumer journey looks like. But first we should look at why this is important.

00:34:16:09 - 00:34:36:19

So while this journey stages are not new, the ecommerce landscape is constantly evolving, right? So personalization and a seamless customer journey are the keys to success. But we've known this for a number of years. What's really changed is the power and sophistication of the tools at our disposal. And that's where the AIs making a real difference. So let's look at how AI enhances the customer journey in retail.

00:34:36:22 - 00:34:55:09

On the first box you see awareness. We previously relied on the consumer having a very precise idea of what they want or otherwise. They would have to wrangle filters and try different keywords for a few minutes. But AI powered personalization helps you cut through that noise and place your products front and center. For those who are most likely to be interested.

00:34:55:11 - 00:35:26:02

And so with tools like we're mentioned previously, like vertex AI search for retail, we are now able to understand subtle nuance of someone who knows exactly what they want versus someone who's just browsing. We can analyze preferences and browsing behavior to provide super relevant recommendations and ensure that basically the right customers discover your brand. The next one on consideration, you've probably experienced this, but a static product, images and descriptions don't always give a full picture of the product, especially if you're a tech savvy customer and need to see the specs and features.

00:35:26:04 - 00:35:47:01

So one of my favorite recent developments into is the rise of just multimodal gen AI, not only for things like visual product search, but also generating and enhancing product images or inferring attributes from the images themselves. So this technology lets customers ultimately understand products in a whole new way, right? Because it eliminates the guesswork and it ultimately gives them the confidence to make a purchase.

00:35:47:04 - 00:36:11:04

And speaking of purchase, who hasn't abandoned this is the next stage, right? A shopping cart, due to a confusing checkout process. So AI powered chatbots and virtual agents are setting a new standard here, and I've personally helped customers implement these with tools that were mentioned again earlier, like vertex AI Agent Builder, where you can build an always on friendly assistant who's an expert, on your product, providing instant support and guidance.

00:36:11:04 - 00:36:38:16

And that ultimately reduces friction. It makes the buying experience just better, and just as enjoyable as the browsing experience. Now, with post-purchase building, a loyal customer base isn't just about making the sale, it's basically what happens after they hit buy, right? So we have technologies like Contact Center AI that completely revamp your customer support capabilities. You can analyze millions of customer interactions to anticipate customer issues, and then you can empower your agents with the right solutions.

00:36:38:16 - 00:37:04:00

Now, being proactive with your customer issues, with your preferences, with your trends. You know, it's only one of the few ways to ensure a positive brand experience, but it's actually one of the most crucial after the purchase process. And then finally just loyalty, right? And here's the thing. You know, loyal customers aren't just repeat buyers. In my view, they're your most valuable asset that feeds directly into organic growth and understanding changing preferences.

00:37:04:03 - 00:37:25:27

So I can help you foster those long term relationships by learning from the billions of data points. And yes, it's billions, at least from the Google side, that are generated throughout the journey by analyzing purchase history, browsing behavior, and even customer service interactions, I can service, essentially valuable insights that generate unique audience groups. We talked to about Marcet Marceting, audience generation.

00:37:25:29 - 00:37:46:21

And then you can automate your Marceting campaigns. You can automate loyalty programs and just have a deeper understanding of your most valuable customer segments. So I know I just went through a lot and you probably got the gist of this, but I isn't just about optimizing one metric. Maybe you care about cart abandonment, maybe another retail enterprise cares about urgently improving customer service response times.

00:37:46:23 - 00:38:05:22

So my advice is always to establish what your top business priorities are, and then apply it to create an excellent ecommerce experience. So it's about empowering every interaction, from the first discovery to becoming a long term supporter of the brand. Now, the last thing I'll say is a Google. We have managed solutions and capabilities to drive value with AI at every single one of these stages.

00:38:05:24 - 00:38:26:27

So if you want to get started, I'm actually going to place one of one example here. So for vertex I switch for retail because you want to learn more. So thank you. And I'll hand it over to Tori. Thank you. Okay. Let's transition to the fireside chat for a deeper discussion on today's topic. This first question is for David.

00:38:27:00 - 00:38:59:11

What challenges and opportunities are present with data activation in a composable commerce context? That's a very good question. In the context of composable commerce, activating data comes with its own set of challenges and opportunities. One major challenge is dealing with multiple services. And because of that, dealing with what we call data fragmentation, because we're a composable commerce stack, integrates multiple services and platforms.

00:38:59:11 - 00:39:26:26

The data collected is spread out, it's siloed. It's always contained within each service. And that makes it hard to get a unified view of all the data. And that is crucial if you want to make, you know, informed business decisions that are, that take into account everything. So the opportunity here is that every service in a composable commerce ecosystem is also highly specialized.

00:39:27:00 - 00:40:04:29

And what that means is that the data that it collects is very rich and diverse. So when a business manages to pull together all of this data or the important data points and centralize it, they can analyze it and more deeply discover new and interesting relationships between data sets. And even better, actionable business insights. And, you know, better insights lead to better decision making, you know, more personalized customer experiences along with, you know, all the other benefits we've discussed today.

00:40:05:01 - 00:40:26:18

And, you know, just to compare, when we look at a monolithic or homegrown platform, you may not even have that same level of access to the data. And then even if you do extract that data, you may not have the quality and the richness of the data that that you get in a composable commerce stack. So, I think I think that's, the opportunity is right there.

00:40:26:18 - 00:40:50:15

It's it's just about getting the right data in the right place. And, yeah, that's the answer. I don't know for anybody else. Yeah. I'm just going back to the to the group here. Yeah, I know I'm I'm with you, David. I think I think there's there's the yin and the yang a little bit. Right. There's, there's multiple components that are all coming together in a composable commerce environment.

00:40:50:15 - 00:41:14:28

But, but then of course, these components are built to be integrated with or they're built for accessibility. That's that's the whole kind of tenets of composability. And so everything's just right there at the fingertips for the, for the data activation strategies that be put in place. So, yeah, there, there, it may seem more ominous because there's multiple pieces you have to put together, but it's so much easier because they're meant to be put together and they're meant to be interacted with.

00:41:15:00 - 00:41:40:07

So I think maybe it's a maybe the barrier to entry is more on the mental side than the actual side. I totally agree. Okay, this next question, I'm going to throw it back to Marc. What role should ethics play in data activation in the use of AI? Yeah, it's a it's a good question. And I think for me ethics falls into the three main categories with it when we're talking about AI, and the first of which is transparency.

00:41:40:07 - 00:41:58:25

And the question there is is do your customers know where you're using AI? And why is that important? Well, surveys have noted that 90% of customers think retailers should be required to disclose how they use their data in AI applications, and an 80% to 87% of customers believe they have a right to know, to right to review their own data.

00:41:58:27 - 00:42:25:00

Right? So customers care how their information and data is being used. So are you being transparent to accountability? Do your customers know how you're using AI? And this gets into making sure that the data is being used in a fair and non-biased manner. We want to make sure that the information is being used appropriately and respectfully. And then three, which gets into security and privacy, which is are you focused on ensuring the vital customer data is protected.

00:42:25:02 - 00:42:46:21

And we're seeing studies show that data privacy is the number one concern of consumers surveyed. And we're talking about Gemini specifically. And I think it's a trade. I think you mentioned earlier, you know, we weren't sure why, maybe Marceting ma tech isn't higher up in the adoption list. And I'm thinking maybe because it's externalized and people are really, really concerned about, you know, that privacy aspect and doing it respectfully in privacy.

00:42:46:21 - 00:43:01:19

And maybe they're still even though I think the tools are very strong in this regard, they're still maybe uncertain about about that aspect of it. So I think we need to build trust with the tools, and we need to build trust. And and as brands and as customers, we need to make sure that we're doing this respectfully and appropriately with with respect to data.

00:43:01:19 - 00:43:17:17

So, really what it comes down to is when you're building out your AI strategy, build it out with these three tenants in mind, make sure you're doing it Cross-functionally bringing in all the different perspectives within the organization, because everyone's going to see it differently. And I'll virtually have one perspective legal. Of course, you can have another your IT team.

00:43:17:19 - 00:43:32:18

And I think all those coming together is going to ensure that, that ethics is going to be, going to be considered properly. But again, it has to be intentional. You have to take the attempts to do it.

00:43:32:20 - 00:43:51:06

I'll, I'll just add that I absolutely agree with the framework and the strategy around transparency, accountability, privacy, and like having the customer understand exactly how their data is being used and having them consent to that. And then also while you're providing answers, having those be cited to actual, reliable sources of information is one of the most crucial things you can do.

00:43:51:09 - 00:44:14:21

Great answer, Marc. The only thing I would add is, I think that we are going to inevitably face legislation around this, and I think, Marc's point about transparency is going to be a big one. And where I think it will eventually go is that you will have to disclose who you're working with and what, you know, what you're actually doing.

00:44:14:28 - 00:44:34:27

So don't do anything that you would be ashamed or embarrassed to admit or, don't do don't partner with any partner that, you know, kind of could create a liability for you. So just keep that in mind.

00:44:35:00 - 00:45:01:24

Yeah. If this next question is for online AI models have been known to hallucinate or make up information, how can you mitigate this? To improve the reliability of AI powered solutions? Yeah, certainly. So great question, by the way. And hallucinations happened a lot last year because of how standalone LLMs work. Right. They basically don't understand anything, anything outside the information that they were trained on.

00:45:01:24 - 00:45:22:26

Right. And more recently there's lookup capabilities and guardrails. But the reality is those nations will continue to be a thing. This year because the model's ability to produce a correct answer basically diverges exponentially with the answer the length and the complexity of the underlying data. And so I don't want to go too deeply into the rabbit hole. But to mitigate this, you know, you have to start looking at things like grounding.

00:45:22:28 - 00:45:51:14

So it'd be great to integrate your models with your data sources, your APIs, and any other enterprise systems that you may have so that it uses your data and that data to answer any questions about products, customer support, etc.. And in addition to the obvious advantage of mitigating hallucinations with this approach, and being able to cite sources like I mentioned earlier, like think about product manuals, policies, etc. if you set it up correctly, you also get the ability to keep that knowledge base up to date, right?

00:45:51:14 - 00:46:22:21

So you have new products, new documents. You can just add those and you know, it'll be part of like the broader workflow to ensure that generative responses are accurate and relevant through time. Yeah. If I can jump in here, I think another another methodology that we've seen successful is is is retrieval augmented generation, commonly referred to as RAG, which is essentially is you're using a search, layer in between the prompt and the response to find those pieces of information that contexts, that data to put into the context.

00:46:22:21 - 00:46:41:13

So whether that's a website link or specific, specific information, and then the, the massive context windows that are coming along, the ability for these AIs to really have this bigger memory, I think is also really helping to minimize the hallucination front. So, again, I think there's still a trust factor that has to be built. It's kind of like a little bit of a private moment with AI.

00:46:41:13 - 00:47:05:04

Like, I got to see it before I put it out into the field, but I think the the abilities are there. It's more of the the groundswell of information and the consumers trust in the, in the, in the eyes. Yeah. And if I may, I add something to that? I been very impressed with Google's ability to ground, their chat bots, the ability to ground data to a BigQuery table to, to a storage system.

00:47:05:06 - 00:47:37:27

When you combine that with, you know, piping and data from a possible commerce, ecosystem, you can bring data from your content management system, you know, all your product information, all your SKUs that your company information. You can easily bring that in in real time and have your chat bots, you know, grounded with your data. So I think there are great tools in the Google ecosystem to ensure that, you know, your models are grounded, correctly and in real time.

00:47:38:00 - 00:48:04:20

So, Katrina, I want to pass this next question on to you. Are you seeing anything unique from an organization structure standpoint that has helped to enable the usage of data or the integration of AI within the retail industry? The main thing that we're seeing is, just, some evolution of org structures really to, to, to help, accelerate experimentation.

00:48:05:10 - 00:48:36:29

So we'll see. Whether it is, a data team or whether it is an innovation lead, or whether it is somebody who's entrusted, with, with trying to figure out how to integrate an LM into a business. It's more sort of who's the owner? Is there an owner? And, is there the ability to, enable that team to work quickly and nimbly?

00:48:37:18 - 00:49:16:21

If anything, it's more of, you know, kind of, modeling of the Amazon two pizza team, that we're that we're seeing more companies trying to embrace is just, you know, rather than have everything be a decision by committee because, you know, that would just mean that none of this would ever happen, is to, you know, kind of enable a smaller team to be able to go run with, with a project, and, experiment and, you know, kind of test and learn really as, as Triton, as that may sound.

00:49:16:23 - 00:49:34:19

I totally agree with that. Yeah, yeah. Sorry—, Mike, now you've go ahead. Please. No, I was gonna say what I love about what I love about AI and what's going on with AI right now is you don't know where the skills are going to come from. Any organization. It's just it's it's in the past, it was so.

00:49:34:20 - 00:49:57:01

It was so specialized, the abilities. But now, because everything's so accessible, what we're seeing is, is the talent for being able to to approach AI and to and to implement it, whether it's from the prompt engineering perspective or from leveraging the tools that are coming out, is really coming from everywhere. And I think to your point, there was as well valid is just enable your people to do it and, and then get out of their way.

00:49:57:04 - 00:50:19:27

You can see what see what they come up with. At some level. So yeah, that that's kind of a spark to my brain. David. Sorry. Go ahead. Yeah. I mean, I was just going to cut the same thing, we heard or, have done that ourselves. We've, you know, we've, we've provided AI tools to all our employees, and everyone has been just running wild with it and coming up with great and interesting use cases and applications and tools.

00:50:20:10 - 00:50:35:26

So, you know, my my advice is, you know, promote this internally in your orgs. Again, give everyone the tools, give them some training on how to use them and, see what they come up with.

00:50:35:28 - 00:50:54:03

Thank you all for your insight here. I want to transition to our next slide. So now I would like for you all to share, one key takeaway from today's session that you'd like our audience to hone in on, starting with David. Yeah, sure. I, I'm looking forward. You know, we're we're at the start of something great.

00:50:54:03 - 00:51:22:10

I think, all the tools are coming together. They're getting really easily accessible, for everyone. And to me, it's apparent that the strategic use of data and AI is fundamental for retail brands to to thrive in the future. So I my advice is for retail brands to embrace data strategies today and, and start getting alignment internally.

00:51:22:26 - 00:51:48:24

Align on specific metrics you care about, work your way to implementing. You know, your, your data flows to, to to measure those metrics and to work off of those metrics. And if you're interested to learn more. Once again, I'll, mentioned the our AI and data white paper. The links in the chat, you know, have a look at that over there and I'll pass it to Mike.

00:51:48:27 - 00:52:09:14

Yeah. I think my key point is, is, AI is dependent upon data. So make sure that you have great access to your data. Put it into a place that's like that. You need your tools to be able to to to make your, your data available to you. And they needed a really good place to put this data that can aggregate and scale with this massive amount information you're going to be leveraging.

00:52:10:04 - 00:52:30:02

I think that's going to be the biggest pivot point here is across brands and companies who are, who are thinking about their data and or putting their data into aggregate, their data now and building good data strategies are going to have the advantage to take, to have the ability to take advantage of all these amazing AI improvements that are coming down the pike and, rolling.

00:52:30:05 - 00:52:57:21

Yeah. Thank you. Marc. I think the takeaway for me would be a lot of ideation was done in 2023. And this year I'm seeing a lot more urgency with customers in removing some of these. I use cases to production within retail. And in doing that I think I isn't just about efficiency. But as I alluded to earlier, it's it's more about reimagining the customer journey from hyper personalized product discovery all the way to proactive customer support.

00:52:57:21 - 00:53:16:05

Right? And I basically helps retailers anticipate, personalize and enhance customer needs. And that's really, the way to be, creating lasting relationships that drive sustainable growth in the future. So that'll be my takeaway.

00:53:16:08 - 00:53:36:20

Oh, no. Security. Oh. All right. So, well, there are only three bullet, so I wasn't sure if I was going to be allowed to chime in, but, Yeah. So, Raymond, I think that you had said it earlier, and I, I, I hope that everybody, embraces the, the importance of what's your business goal first.

00:53:36:20 - 00:54:01:11

And, the number of conversations I'll have with companies who are like, let's talk about AI. And it's like, we can talk about AI, but, you know, kind of yeah, and it's AI, you know, it'll be great. It'll be like talking about chemistry or biology or, you know, or some subject, but that it has to come back to ultimately, what is it going to do for your business and what are your business goals?

00:54:01:11 - 00:54:24:16

What do you, you know, are you looking to do something as specific as effect cart abandonment? Are you looking to improve your conversion rate? Are you, you know, kind of looking to grow sales and improve loyalty? And, you know, kind of there are so many different ways to achieve those. And AI is going to be a tool that can help empower each of those answers.

00:54:24:22 - 00:54:50:18

But, you know, kind of AI alone is not the right way to approach it. And I, you know, I hope that, you know, kind of everybody is, is in agreement on that. Or to start with AI versus, you know, start with the business problem first and then layer in, you know, kind of however, I can fit.

00:54:50:21 - 00:55:09:26

Thank you all for your responses here. Before we do a session, I want to remind everyone that now's the time to get straight from the source, insight from our panelist. Great questions. Contribute to a great webinar. Don't forget to submit your. Additionally, we understand if you need to run, just know that today's session is being recorded and will be emailed to you within the next 48 hours.

00:55:09:28 - 00:55:32:28

If we don't get to your question today, you can connect with our panelists via the social media channels that will be on the screen shortly, and it'll also be in the chat. Okay, so let's dive into a a shouldn't this first question comes from Stephanie. What what are the prerequisites, if any, for implementing AI in an organization? Let's start with Marc.

00:55:32:28 - 00:56:03:29

What are your thoughts on this? Curiosity? You know, I think when we talk about framing the problem correctly, it's interesting you're dead on there, right? Like, let's not build, AI solutions for the sake of building them. Let's solve a real problem. Is certainly, probably the first thing, but, from a from an actual implementation perspective, you need to be able to understand you have access to, to a high quality, tool, tooling, whether that's if you're doing generative, there's lots of options.

00:56:03:29 - 00:56:33:08

Obviously Google Vertex Studio is amazing in that regard. You know, if it's predictive. There's also a lots of options out there. And again, Google's got amazing tools. All of them there. So I think having access to the tools is probably one of the harder things that we've seen. And vetting them properly. Not because they're not available, but again, is because you have to build organizational alignment behind using the tools and making sure that, that everyone's that everyone's approved them properly because there's an approval process that we've seen organizationally with a lot of people picking up AI tools.

00:56:33:13 - 00:56:45:03

So I think figuring out what you want to do and then figuring out which tools you want to start with and then be curious.

00:56:45:05 - 00:57:06:20

Anyone else want to provide any input on that, or would you like to move to the next question? I think Marc covered it is really good. Perfect. This next question comes from Aldo. Do you think generative AI is creating new jobs across the retail industry, or do you feel that only training can fulfill the gaps in the future?

00:57:06:23 - 00:57:10:25

So to read, I want to pass this one off to you first.

00:57:10:28 - 00:57:31:07

There are, this is, it's not a new answer. It's an answer that I think we've all heard is that, it's hard to predict what jobs will change and evolve, but we can be rest assured that there will be new jobs. And they will be, you know, kind of based on, on these new technologies.

00:57:31:07 - 00:57:54:06

There's no question about that. I mean, who would have predicted that, you know, you'd have paid search experts, you know, 30 years ago, you know, or email Marceting experts. So, there there's a lot that will change and absolutely evolve. And, just go back to do you need to be as nimble as you can be. And, one thing I've heard consistent.

00:57:54:06 - 00:58:12:13

So it's, it's interesting that, code, you know, kind of we had this, this huge and Marc may, may have a point of view to, to weigh in on this, but, you know, there was this huge push to get, you know, young people, you know, up to speed on coding and then, you know, kind of people were like, oh, with ChatGPT, we don't need code anymore.

00:58:12:13 - 00:58:32:12

But it's like, no, no, no, no, no, nothing could be further from the truth because you need to know you. It's like, you know, you can't you can't just expect that, you know, the code is going to be right. You know, you need to know enough to know how to audit it and rewrite it if it needs to be rewritten and to, you know, kind of edit it so, so that it absolutely does not.

00:58:32:12 - 00:58:50:04

You know, it's like saying just because there's a calculator, you don't need to know math. Like nothing could be further from the truth. And in fact, it's it's probably even more crucial now because it is going to be so embedded that you need to have the critical thinking skills to be able to go and and check your check the work.

00:58:50:06 - 00:59:20:28

Yeah. No, I'm with you here. I think that critical thinking skill is exactly right. It's it's these tools are going to enhance our abilities. They're going to enhance our, accelerate our abilities to deliver things faster. Going to your development example, you know, having the ability to, to generate code, for the simple use cases is going to speed up things that, you know, things speed up certain things, but it's not going to replace the developers, need, I think it's just going to make us more and developers more efficient as they go through their day, at least in the short term.

00:59:21:03 - 00:59:42:17

Yeah. As we go longer term, like you're saying, we don't know what's going to happen five, six, ten years. Nobody knows. But in the short term, I think it's going to enhance jobs. It's going to enhance abilities and jobs rather than replace. And to jump on this one. There's a quote it's been going around that says, you know, I won't take your job at somebody using I, will take your job.

00:59:42:19 - 01:00:05:16

So my advice is, if you're not using AI in your day to day operations, you should start, you should start learning these tools, understand how they work and see how they can you can leverage them to to do your job better or, you know, to improve, you know, outcomes for your business. That's my $0.02 on that.

01:00:05:19 - 01:00:26:04

I love that quote, David. I'm absolutely agree. And echoing what you said, I, a more recent one that sort of piggybacks on that is that, I saw in a news article, employees are not waiting for their employees to, to get their stuff ready with gen AI. And, honestly, I'm happy to see that because there's curiosity to adopt.

01:00:26:04 - 01:00:46:24

And then there's also the you see enterprises taking the time to do it correctly and so forth. On both of those sides. I'm optimistic. And, you know, luckily, Google is one of those, enterprise ready gen AI platforms that can help you accelerate that. So yeah, just adding that to the answer, but really echoing everything that you said about accessibility.

01:00:46:27 - 01:01:03:13

Thank you all. And it looks like that's all the time we have for today. I hope you all learned half as much as I did. If we didn't get to your question today, you can connect with our panelists via the social channels on the screen. And they're also in the chat. Before we go, we'd like to encourage you all to participate in the survey that will pop up in your browser after the webinar.

01:01:03:16 - 01:01:26:04

This feedback helps us to continue to produce webinars you all enjoy. Thank you to our sponsors, Orium, commercetools and Google Cloud and a big special thank you to our incredible speakers for providing us with such a valuable presentation today, and a huge thank you to you all for attending. I’m Tori, and I hope you have a great rest of your day and you.