x
READY TO BECOME A MEMBER?
Stay up to date on the digital shelf.
x
THANK YOU!
READY TO BECOME A MEMBER?
THANK YOU!
Transcript
Our transcripts are generated by AI. Please excuse any typos and if you have any specific questions please email info@digitalshelfinstitute.org.
Peter Crosby (00:00):
Welcome to unpacking the digital shelf where we explore brand manufacturing in the digital age.
Peter Crosby (00:16):
Hey everyone. Peter Crosby here from the Digital Shelf Institute. David Feinleib has been at the forefront of content and commerce throughout the last decade of the digital shelf, and he is now pouring that knowledge and experience into a new venture focused on the next great unlock for commerce, which it will not surprise you, is AI at ItsRapid.ai. His new technology and wisdom on the process and people side of content creation and optimization is powering the AI test and learn to production timeline for some of the world's largest brands. Dave joined Lauren Livak Gilbert and me to bring that timeline and its lessons and benefits to life. Dave, welcome to the podcast. Today we are talking about, wait for IT, ai, and since your company name ends in it, it'srapid.ai. I'm hoping you know something about this.
David Feinleib (01:10):
Peter, thanks for having me on. We do know a thing or two mainly from our customers about AI and content, and so I'm really excited to be on the show. Thanks for having me and excited to talk about those topics
Peter Crosby (01:24):
Today. It's a great combo. And specifically when we're thinking of ai, how do you use it to scale content for the digital shelf? And that can span from the PDP through to retail media and even in store. So as you said, you talk with brands every day about this. How are you seeing them adapting to introducing AI and I imagine testing and learning what's going on out there in the real world?
David Feinleib (01:52):
So much is going on. Let me take that in a few parts. First of all, it's got all the buzz. As we all know from reading the news, the brands we work with, the PepsiCos, the new Mondelez folks like that are very excited about AI and in many ways I think they do see us as a way to get started and try out practical applications of AI with a copy or imagery or even video in some cases. And what we see brands doing is really dipping their toe in the water, stepping in, trying it out, and then they gradually adopt it and see where the technology and the regulatory intersect to allow them to really adopt AI and their content workflows.
Peter Crosby (02:44):
Yeah, because those names, I've heard of them, those are big brands and so it's natural that there would be some push and pull between the excitement of what's possible and the fear of what might happen. And how are you finding that going in your customers that are say in their second or third round of testing or learning with you, is that starting to, do you ever get in contact with the legal or compliance people and what's the tone coming from their side?
David Feinleib (03:18):
Yeah, it's a great question. Listen, I think there's excitement and I think there's a note of caution because it is new and with any new technology, as we've seen over the ages, there's always some question about how it's going to work and how we can adopt it and things like that. The area where we're seeing ai, the most comfort with AI is around text. So when we think about copy optimization for the product detail page or title optimization or retailer specific content, I want this content for Amazon, but that content for Walmart and this for Target and Kroger and so on, that's a really exciting place for brands that are doing practical operational production level work. And we've got a lot of workflows to help 'em navigate that from getting the content in to proofing it and reviewing it and making sure that it is signed off on. And when it comes to imagery, we tend to use what we think of more as automation. And so there we're pulling various elements together and then, and if we can we apply AI to the image creation so we can unpack that a little more, but there's kind of steps on the path. Maybe start with copy, then you work your way to imagery and then you get to video.
Lauren Livak Gilbert (04:41):
Do you find that brands ask you to engage with their legal or their regulatory teams before kind of signing off on something like this to give more context around how to use AI or how to be comfortable with ai? Are you having those conversations?
David Feinleib (04:56):
Yeah, a hundred percent. And I think with larger brands, just like you're going to have, let's say an ISO questionnaire or a SOC two questionnaire, you're also now going to have a set of AI questions and those are going to be things around what technologies do you use, what are the areas that we're going to be using AI in? What is the workflow to ensure that regardless of what technology we're using, we have a process in place for the brand owner, legal, whoever needs to be involved to sign off on the final content that's going live. So we definitely do see that now, and I think that's a great sign. That's a sign of increasing maturity for the adoption. It says, Hey, as a brand, we're thinking about how do we move this into production and think about using it for actual products that are live.
Peter Crosby (05:54):
Dave, I generally and maybe Ineptly or Ineptly, think of AI as having two major benefits to the organizations. There's the efficiency of things and then there's, well, I guess three maybe efficiency scale and then performance. And you mentioned optimization earlier and you're talking about workflows and things, and your company has rapid in the name. I just wasn't sure where in your mind and then maybe in your customer's mind, A, does that sort of cover the value chain? And if not, I'd love to hear more and then also where their hopes lie mostly.
David Feinleib (06:41):
Yeah, what a great question. And we talk about that. I didn't come up with this, one of our customers gave it to me, so I just like to talk about it. But we think about the trifecta of content and the trifecta is quality, speed and cost. And historically it's been a real challenge to get all three of those. If I wanted to move fast, I'd have to make some trade-offs in the other areas. If I wanted to have low cost, I'd have to make trade-offs. But the thing about AI and automation having the right workflows in place is that you can really get to these three elements, speed, price and high quality. High quality being adhering to retail requirements and staying on brand. And you can do that across mobile heroes, across product copy, retail media for banners, sell sheets, all sorts of different product formats.
(07:39):
So we think that's really exciting. The second part of your question, which is what's the performance look like? We do then see brands look at the lift and the sales performance and things like that. But the first step is really can I get to that efficiency like you were talking about, speed, quality, the price point, and put all that together to get the content life cycle where I want it to be. Peter, I know you've been in the space I have for a number of years, and Lauren, you've seen this from the brand side, from the industry side. And I think if we look back a few years, we'd go, how often does a brand really want to update their content? And if you rewind five, six years, you'd say, gosh, if I could just get through item set up and get one photo and some text up on the site, I'd be a hero, right? A weird, and I'd be incredibly happy. It was so hard just to do that. Now you talk to a brand like Newell for example. They're updating content and imagery and so on multiple times per year. That's really a new effort and a new approach in the space. And they're doing that because the space requires that one, they can do it now. Now because got the tech and the workflows, and two, they need to do it because the competition in the market is updating the content much faster to keep the content fresh.
Peter Crosby (09:16):
And that's where the growth is coming from. It's because we've kind of seen on the digital shelf, the first flush of it was a decade, but the first flush of, oh, I can sell things online and when I put 'em up there and I add some really good core content, wow, look how things get better. But now we're talking a game of scale and inches in a lot of ways. And so we talk about the goal for product experience management has always been to optimize every touch point on the digital shelf. Now, that was never possible for every touch point, but we knew that at someday, not exactly knowing how it would be and that every brand to compete to your point, would need to be able to get to that level of coverage and quality and then that extra little zing of slightly better than anybody else around me. That's the performance piece. And that's so exciting about this moment is that you're seeing that start to come into place where that goal is over the next few five whatever it is achievable for every brand and it's going to be a minimally viable capability I think at some point to survive. I dunno, am I overstating?
David Feinleib (10:39):
Maybe I am. I don't think you are. I think this is a requirement. I was on a panel as an example at the Digital Shelf summit with Nikolai Sogard from Newell, and he said, it's a requirement now to be able to refresh this content because he's seeing the competitors, there's upstarts in the category, they're seeing them take very agile approaches to content. So his big question was how can I as a big company, A CPG, leverage my scale using some of these tools? Liken its rapid, for example, using some of these workflows. And so I get the benefit of my scale, but I need the agility of some of these upstarts. What's the enabler to let me put those two things together to win on the digital shelf? And I thought that was such a great perspective and way to think about it. And it's true.
(11:36):
If you have the right workflows and process and some tech, then you can really build this content at scale and you can iterate it. I think that's the really exciting thing here is before it was kind of one and done and you felt so much pressure to get it perfect, and now you can put that content up there, whether that's images or product copy, some video, and then you can change it and see what changes in the scorecards, what changes in the rankings, what are the conversion rates do, are you excited about the efficiency of eight weeks down to one week and what does that enable?
Lauren Livak Gilbert (12:15):
And I think this is perfect for something like seasonality. And even last week when I was at Cannes Lions, they were talking about micro seasonality, which is even a double click into seasonality. So when you think about regular seasonal, its like you Easter or Christmas or Halloween, and then micro seasonality are things like matcha month or some retailers are actually designating like a baby month. So what baby month at what retailer at what time, and how do you create the content to compete with that? And that's going to be that little level up where you have your basic content, but how do you personalize it from a seasonality and even a micro seasonality? And I seeing this being that enabler to do that,
David Feinleib (12:56):
It's a great point. And if you take that a step further, you have categories like chocolate for example. We have a number of clients like in a lint or Russell Stover as an example. We all know what the big selling opportunities are for them it's day, it's mother's day, and then gosh, what is it? And so that's the big question. And so to the extent that they can now personalize video content and images and really take advantage of those micro moments, God, they are really excited about that because that lets them stay current and present and relevant for the shopper all throughout the year. And it aligns also with their budget and their investment because obviously if you know that the historical revenue is around one or two periods in the year, it's very hard to go then ask for budget for the rest of the year.
(13:54):
But now they can come back and say, with that same tool set, I can now create content for these micro moments and I can experiment with it. I could put something up that really only has a shelf life of a couple weeks, maybe a month or two, and then I'm going to replace it with more content or different content, some other ideas from the team or opportunities for the year that's really exciting for them. Or here even is a trend and I want to do something that's in the moment. How do I be agile enough to first of all put the brief together, then produce it, and then get approval? It's a fairly complex set of steps if you're a larger brand to do this. And so that I think is really exciting. One of our clients was telling me, it's back to the basics. I said, what are the basics? He said, listen, now we can automate syndication. We can automate analytics. We've got those tools so we can actually do what we all wanted to do in e-commerce, which is create great content and help bring that content to life for the shopper. So I thought that was a great way to think about it. That's amazing.
Lauren Livak Gilbert (15:04):
So Dave, let's get into a specific example because I think that'll be really helpful to think about how it's applied. So Pepsi, I know you've done a lot of work with Pepsi. Can you talk us through how they've kind of used ai?
David Feinleib (15:16):
For sure. So first of all, I got to say I'm incredibly thankful and grateful for our partners at PepsiCo. They're a very large organization. They work with vendors in a really, really unique way like us. They truly treat us as a partner and we have the privilege of having kind of an inside seat, if you will, on their content production and banners and things like that. So I'm really, really excited about that. And what I'm going to do is take one example from there, if it's all right with you and one example from Newell and talk about both of those with Pepsi, you think about the scale of their product portfolio, tens of thousands of SKUs worldwide. Everything you do has to be at this scale. And so when you think about anything that's manual today, let's take mobile ready hero images, there's something that if you're only doing five or 10 of them, hey, you know what?
(16:17):
No big deal. Open up Photoshop. Maybe I've got my favorite design tool, whatever it is, create a couple images, heck even I can do it. And so no problem. Now you scale that up back to our trifecta of the key elements of delivery. I want to do it for a hundred products. I want to do it for a thousand or 10,000 products. Now the question is, well, how do I take the data, which is probably in a system like a PIM system, like a Salsify or stibo or other systems of record that a company like Pepsi has, how do I take all that data? How do I take my templates which are specific to each brand and retailer, and I kind of merge all that and produce the results. So when we think about a Pepsi workflow, there may not be as much ai, but there's a heck of a lot of automation in order to scale it up.
(17:15):
And when we think about ai, we might say what it's really doing is taking all these different data sources, all these templates, all these elements, putting 'em together and taking something that would've taken eight weeks, turned it into a week, maybe even a day or an hour. So that I think is one really great use case. Mobile heroes. If you then look at something like a renewal where they're doing product copy on a very high frequency basis, that's something where we might have written that by hand historically. Now it's very hard to do that. Even if you have the best team in the world to try and update your product copy on a monthly or quarterly basis there, we can connect to ai, we can really dial in the prompts. I find this to be the hardest part for our customers is they read the news and they go, I went into such and such tool, I wrote my own prompt and look, I got some cool content as a result.
(18:18):
And that works really well for writing a sales email or a blog intro. If you want to have the brand voice specific to the retailer for the right keywords, and it adheres to your claims requirements and it's going to get approved by all the folks who need to approve it, you really have to have the right prompts and the right workflow to do it. So there's a case where Newell leverages this entire workflow from prompt creation to review, to approval, to delivery back to the PIM system to really scale the content creation. So two different forms of what I'd consider ai. One is for automating the creation of images. The other is for scaling the creation of Copy
Peter Crosby (19:06):
Dave out in the world with sort of the large language models that we hear about. That's sort of where the idea that AI lies and it lies confidently has sort of arisen. What is its rapid doing to, because you are really using data that Newell has, correct? Yeah. You're not going and grabbing stuff from there. You're not sharing Newell's data out into OpenAI or anything. Can you just describe some of that and why it's less likely to lie?
David Feinleib (19:46):
Yeah, yeah. So let's break down the manual process of we had no ai. We would write a brief, we'd have some existing content, and we would say, Hey, can you go get some keyword data and rewrite that? By the way, you need to follow the Walmart or the Amazon style guide while you're writing that. Oh, did I mention that we really want to highlight this certain trend that's going on that's current right now. So you have to take all those factors and put them together into whatever you're going to write, and then someone has to look at that content and say, yeah, that looks right to me, but what is right? Well, it adheres to the style guidelines and the keyword and all those other requirements. Now, if we look at the AI process for that, we're going to drop AI in the middle and we're give it a set of those same instructions.
(20:46):
But what's neat about it is, one, it's going to follow all those instructions almost to the T, but we do need to check it and vet it. Two, we can run it multiple times. So this is where this iteration really comes into place, which is I can run it. I can see the result. Now I want to tune it a little bit to adjust some wording, adjust the style. I can rerun it just like that and see the result again. So it's really iterative really quick, and I can kind of test and learn myself and then do that multiple times. So I think that's what's really exciting about the AI enablement is certainly the speed, but also this ability to do it really quickly. Now, the question then is how I do that for a hundred products or 500 products on a monthly basis or a quarterly basis there, I really need a workflow to help me manage which items did I even optimize?
(21:45):
What were the results? Which ones have I approved? Which ones did I send back to my him? Can I compare my original with my results so I can really see what the differences are? So all of those elements, again, maybe not so hard. If I've got a couple of items, I could do it by hand, a few hundred items and the frequency of doing that, all of a sudden it's this sort of exponential problem that I've got to scale. How do I do that? Quality, price, speed, I need technology again in its rapid or other tools to be able to do that.
Lauren Livak Gilbert (22:24):
And Dave, I think one of the, I guess cautions that most brands think about here is where's the human in the loop, right? Because we don't want to just put AI content out there without out there making sure. So can you talk a little bit about being able to scale with ai, but still having the human in the loop and how that kind of combination works? Because I think that's in the minds of a lot of brands as they're starting to think about leveraging this.
David Feinleib (22:52):
Yeah. Yeah, I agree. And also when I create content myself, I want to be able to sign off on it and make sure it's authentic to me, and I'm putting my name on it just like everyone else does. And so I want to feel confident that what's going out into the world represents me well and adheres to all these requirements. And that's why we have this review process in our platform, which is I've created the content, now I can go through and it's going to highlight for me what are the differences between my original content and the resulting content, and I can really see that visually and go approved, or here's a comment. And so now I've got, let's say maybe I've approved 90% and I can publish that back to my pim. Great, I'm good with those. Now let me go look at the 10% that I want to take a second look at.
(23:47):
I can either make comments, I could edit in line and then run that, or I can even run it through again and republish it at that point. So I think having this workflow for get my content in, run the op and then review and approve or comment on the results, that's really important because that lets you look at the history of what you've done, lets you compare the content side by side, and then you can really mark easily, hey, which ones are good to go and which ones need some further review? By the way, that would be the same process if I were doing it by hand. It's just that now I'm doing it much higher frequency, much larger volume of items. And like you said, I do have this kind of question mark. I really do want to be able to vet the content and make sure I can sign off on it. And having a workflow is a perfect way to do that.
Lauren Livak Gilbert (24:44):
So Dave, let's put our, I feel like we need tapes for crystal ball watching around what's going to happen in the next three years in terms of scale, what do you think that looks like for brands leveraging this in the next two, three years? If you could look into the crystal ball.
David Feinleib (25:01):
Yeah, I would say two things. First of all, it's doing this at scale, which is what you just said. So I think we're seeing brands move from test and learn into production, and we're going to see more and more of that. The questionnaires, how do I do this at scale? What's my workflow? What's my approval process? Those are all maturity questions. We're moving from kind of the early stage into adoption. Everyone's kind of using it in their emails, they're seeing it, so things like that. And so we're just going to see more and more scale and more and more adoption. So I'm excited about that. The other part I'm excited about is how do I combine something like the outputs from an open AI with the sort of image automation we have in its rapid, put those two things together. So I could actually write a high level brief.
(25:57):
I want a banner ad campaign for barbecues for the summer, and I'm now going to get 10 variations of my banner ads, and I can just pick and choose the ones that I want. Now that's a little further out. It's onto your one to three year time horizon there. But I think it's possible. It's really exciting, and it's something where now I can say I'm still in control of the brand tone, the voice, the overall look and feel, but now the engine can really put some dynamic content together for me and I can choose what I want to run, or I can even run a whole bunch of variations. And we are seeing this a bit with clients like Pepsi where they're testing multiple variations of retail media and just figuring out what performs the best. So that I think is the second area that I'm super excited about.
Peter Crosby (26:56):
Dave, so far, and this is often the case with new areas of software, the early adopters often are the bigger brands that they might have some loose change around. They can flex because run
David Feinleib (27:12):
A pilot.
Peter Crosby (27:13):
Yeah, exactly. Exactly. Is there anything preventing the medium size brands or others from jumping into this arena with the same kind of well, or with their own brand of how do I test and learn this and can I afford it and all of that? What are you finding with maybe the smaller set of your customers?
David Feinleib (27:35):
Yeah, what a great question. This is one of those technologies where small and mid-size customers can take advantage of it right out of the box. And so I imagine we've all tried out a chat, GPT, prompt ourselves or what have you. The question I think is for a small and mid-size brand, I'm an owner and I only have so many hours in the day. I'm trying to do a hundred things. So I've got that same question. My big brand has, which products did I try it out for? Do you have anything that'll just make it more efficient for me to run it and rerun it so I can get the result I want and kind of streamline the process for me? So in a way, they have many of those same challenges, nothing holding them back from doing that. And we often have, for example, through our partnership with Instacart, we have small and mid-size brands come to us who want to do a project or a banner campaign or something like that.
(28:35):
Nothing holds them back from saying, I'm looking for some inspiration for that summer grilling campaign or back to school or holiday or whatever it is. Let me use some tools from, its rapid to give me the ideas, give me some variations, and now I can just go and execute on that and decide how I'm going to allocate the spend and the time and all of that stuff that only I as a business owner can really make the final determination on. But what are the right headlines for my retail media campaign? Can I have some text and imagery for retail media that aligns with my product detail page copy? That's all stuff that can really be automated and save the business owner some time.
Peter Crosby (29:22):
And in my opening when I first welcomed you a board, I mentioned that AI can scale PDP retail media, and then I said, or even in store, what does that look like?
David Feinleib (29:37):
Yeah, in store is a great area. Peter. I just had a guest on my own podcast beyond the shelf, Marlo Nichol from grocery tv. He was a great speaker and he was telling me his big learning from in-store, which is keep it simple. And I thought, what a great insight because there's so much you could do with in-store, you have audio, you have print, you have screens with video, you have the shopper walking around with their own personal device. So in-store is a very broad set of content surfaces for someone to think about. I think what we see brands doing is saying, how can I take the content I've already got that I've made a huge investment in and customize it for a local market? And that might be as simple as, here's something that's in print that I want to have for the New Hampshire market, the Texas market, the California market, whatever it is.
(30:43):
And I've really customized it for each of those areas, maybe even down to a specific city, a specific store. I might have an offer that's custom to that, or I might be doing a tie-in if I'm a larger brand, I have a partnership with a football league or baseball or something like that, and I want to tie that into the game, and I'm going to run that on the in-store screens. Now, my challenge producing the content is how do I pull in the logo and the footage and the messaging for that specific market and do that in a quick way because those budget dollars might not come available, or that opportunity with the retailer might not present itself until it's within a couple of weeks or even a few days of the content needing to go live. So again, that's where you can leverage some tech to say, I've got existing content. I want to do something local personalized, either on a video surface or a print surface, for example, and I'm going to produce that and now deliver that for this in-store experience. So I think those are some of the ways we're seeing brands really take the content they have, use some tech and personalize it for the in-store and the market.
Lauren Livak Gilbert (32:02):
I was talking to a brand, I think this is a perfect example, and they shared something that was super relevant here, connecting the dots between in-store and online, they spent a lot of money on an end cap for a product at a major grocery store, and consumers were seeing the end cap, but didn't actually want to buy it in store. And so they were going on their phone and they were picking up the product detail page, but when they did that, it was out of stock and they didn't have the same packaging, and so they spent all this money on an end cap, but we're actually losing dollars because they weren't connecting of the disconnect. Exactly. So I think what you're saying is so important and being able to use technology to make cohesive story across all of your channels in an omnichannel way is so important because those ad dollars that you're spending, it's not cheap to do an end cap in a major grocery store, but how can you make sure that that's seamless for the consumer and you're actually driving additional sales?
David Feinleib (32:58):
It's such a great point. The way we often talk about it is you're flying the plane with retail media up here, and you've got, whether that's your online banners, onsite, offsite, you've got out of home, you've got retail media in the store now, and then how do I land the plane, which is I've got the product detail page, and frankly, I've got the physical product on the shelf in the store. And if those don't all match, which is really hard to do because you have different parts of the organization working on them, you have different production schedules, things like that. But imagine if you could do all that at once. How powerful is that, Lauren, to your point to say, I'm flying the plane, I'm landing the plane. It's all consistent, and I know that all the way from my banner ad that I ran to my mobile ready hero image to my end cap, the whole thing is consistent, and maybe I've even dialed it in for seasonality or those micro moments or what have you. That is I think, really powerful and really the promise of this creative automation, technology and space,
Peter Crosby (34:11):
Creative automation. I love that. I love that term. Thanks Dave. I'm going to carry that with me for a little bit, and I can't wait for our robot overlords to do all of this for us so that we can have a beverage out on the porch in the summer.
David Feinleib (34:27):
Well, if we could all spend more time at the beach, we would love to do that. But I have to say, I've seen that. I don't know about all of you, but there's a lot of work to be done right now. We're seeing a lot going on at our brand clients. They're doing more than ever before. They're the speed, they're moving at the volume of content. And this ability to just create really fun and exciting and engaging content, I think is really fascinating. I'm excited to be working with these brands and some of our retail partners, Amazon, Instacart, Walmart. So I'm excited to see what more there is to come in the space.
Peter Crosby (35:13):
Yeah, that's what I felt at the digital shelf.
