Jordan Buckner (00:00)
AI search is really changing the way consumers are discovering products. for years, brands have focused on getting discovered through retail shelves, Google search, Amazon, social media, and recently TikTok. But now there's another layer emerging, which are AI shopping assistants. And these are really just built into existing tools like Claude, ChatGPT where people are really just going to asking questions about anything. And consumers are simply asking things like, what's a good low sugar snack? or

Find me a healthy cereal for my kids. And the AI tools will help narrow down the options for them. But by the other design, they're showing the top recommended products. So this creates a new question for CPG founders of does AI actually understand what my product is, who it's for, why someone would buy it, and should it be recommended? So today I have invited on Andy Kurtts, who's the creative director.

And founder of Buttermilk Creative. They do a lot of branding, packaging design, including companies like My Better Batch, Daily Crunch, and Roro. We've had many of those on the podcast before. And Andy and his team have been testing real CPG brands across six different AI platforms. And what they have found should get everyone's founders' attention. So we're gonna jump into it and really discuss how you should think about including this within your brand design. Andy, welcome to the show today.

Andy Kurtts (01:12)
Awesome Jordan. Thank you for having me on. And you hit the nail on the head and I can't wait to jump into it.

Jordan Buckner (01:17)
perfect. you have traditionally worked in branding, packaging design. I'm curious, how has AI changed or is changing how you think about helping founders build a brand, design packaging, and how have you started exploring AI in your your work?

Andy Kurtts (01:33)
Sure. I think it's such a double edged sword AI, because everybody's talking about it. And there's a lot of noise. but it whether you like it or not, whether you're for data centers or you're not, the just all the controversies, it's impacting your brands. And it's impacting brands at

light speed. And so we just notice that, this is happening whether we like it or not. And so we want to learn as much as we can about it. And but we also don't want to be inauthentic about it. We don't want to just jump on the bandwagon of hey, here's a tool that'll help you,

address these data things because that's not what we do it's not our background so what we started to dig into was how are these retailer outlets and then the assistants the AI assistants reading your actual packaging physical packaging front of pack and so that's what we started testing and looking into and it's been really interesting just doing our little minimal tests that we've done so far

and

the things we've learned.

Jordan Buckner (02:28)
Yeah, it's so interesting, right? Because a lot of your work I'm sure is focused before on how can you make your packaging stand out to a consumer in three seconds. And now you have the added question of how do you make it stand out to an AI in a millisecond, right? And so to

Andy Kurtts (02:41)
Yes.

Jordan Buckner (02:41)
really understand what it's doing.

Andy Kurtts (02:42)
Yeah,

So it's similar worlds and it's a tall ask for founders now. We call it the third shelf, and so you've had the physical world with retail shelves, you have the e-commerce world, and now you have this AI agent assistant world, the third shelf. And so now but that has to work on all of those. And luckily the e-commerce one and AI work hand in hand, but you can't just I would shudder to see some of the brands that we work with if they

l literally just took an AI assisted direction for their for redesigning their front of pack because it's probably gonna fall apart in a retail setting. It's not gonna look good. So there's a balance, but you still need to make sure you're following certain aspects of this emerging world that is the AI reading your packaging now.

Jordan Buckner (03:27)
Yeah, so interesting. So I know you started testing out brands across different AI shopping experiences and I wanna know what did you what were you trying to understand and what surprised you about some of those results?

Andy Kurtts (03:39)
Yeah, so one in particular. So we took a a snack brand.

And we asked six different AI assistants to recommend something in its category. And we we had questions, we had 14 or 15 questions, queries. And the queries ran from broad to very specific, brand level stuff. rec tell me more about this specific brand. and so we just tested them all, and one stuff was like healthy afternoon snack.

And it didn't show up in any of those six area assistants. but when we asked for it by name, it won every time. So the product was great, the brand was strong, but it was a it was invisible to the shopper exactly when they needed it. And that's the thing that people have to think about now is it's like when people are like you mentioned, when people are shopping via an agent, they're not thinking about

Okay, I'm walking through the grocery store and now I'm in the produce department. So I'm looking for produce and now I'm on the energy drink aisle and I'm looking for energy drinks. It's like, no, I need something that is gonna solve my problem of crashing at 3 p.m. What's a product that can help me do that job? one more thing with that brand. So we then went to that had a very strong seed oil claim. they are seed oil certified. And so on Amazon, where that claim

was written in.

into the listing the AI recommended it and literally said this one is explicitly seed oil free it was part of the top listings but on another platform where the it's the same claim but it only appeared on the pouch as the

Jordan Buckner (05:05)
No.

Andy Kurtts (05:05)
certification the AI couldn't see it and so they didn't get credit for it because they hadn't put it in the data field so that's the thing and that's these things go hand in hand it's optimizing that front app the

physical packaging so those vision models are reading it and then it's also making sure going through all these different platforms and inputting the data fields.

Jordan Buckner (05:26)
Yeah, I think that's so important because

Different tools are using different data sets to inform what they showcase. And even remember hearing that Amazon's actually transitioning to shortening headlines for their products, like the title, they're shortening those to be more user-focused, especially on mobile. And so reduced are the days of the keyword stacking and the end of the headlines on Amazon. they're relying more on AI to read the packaging design.

The images in which are showing the packaging to highlight the benefits of the product versus stacking as many of these hidden keywords into the back end. And so it makes such

Andy Kurtts (06:04)
Exactly.

Jordan Buckner (06:05)
more of a difference of highlighting the actual product and attributes and the key components on the packaging in the images so that that's what the AI is reading from.

Andy Kurtts (06:14)
Yeah, exactly. Yeah. and it just is

also interesting in there's one Cyndigo and we looked at we were looking at my better batch and her listing it was weird. It was like it included, the everything from her packaging. It had her founder story, and then after that it included the her cooking instructions, and then after that it included some other brand story stuff. And so we were like, it's literally taking

Everything from her packaging, stripping it off, OCR, I guess, and then pulling it into the listing data. And so it's like, that was interesting. It's like, well, if we want some of those occasions and we want some of those jobs to be done to be included, it needs to be in that particular platform, it needs to be included on the packaging. So it pulls it. Yeah, exactly. Yeah. Because it's literally

Jordan Buckner (06:58)
Explicitly. Yeah. So when you were doing

Andy Kurtts (07:02)
taking it, it's literally taking it from that printed packaging and populating.

Jordan Buckner (07:07)
Yeah, especially for a lot these platforms where on Amazon as a brand, you can control a lot of the data that you're inputting, but there's a lot of other platforms

Andy Kurtts (07:13)
Yes.

Jordan Buckner (07:14)
where you are not controlling that data, and so you're at the mercy of whatever they decide is important. Yeah.

Andy Kurtts (07:19)
Yep, exactly. Yep.

Jordan Buckner (07:20)
Did you find that through your research that there's a bias towards having more established brands show up more frequently versus emerging brands, or was it mixed?

Andy Kurtts (07:28)
No, I have not I honestly feel like from the research we've done and the testing we've done

the a lot of the brands, yes, the the probably the massive CPG brands are investing a ton of money in getting this right. But a lot of the, let's say, established artisan brands that show up in these searches, I think are doing it through dumb luck. know, like they

Jordan Buckner (07:48)
Mm.

Andy Kurtts (07:48)
just have, just I don't think anybody has thoughtfully went into their data fields and optimized them for this. and so no

I don't think there's a bias and I do think that there is that's where the opportunity is because it's so new right now that founders, especially emerging founders that are emerging brands that are just, out there s they're setting these things up probably in real time.

And they're getting on these platforms. It's like now's the time to learn about this, figure out what should be populated in those fields and go ahead and get it right off the bat versus having to go back and retool. I know how a a lot of emerging brands they set up their Shopify site and then maybe two or three months later they do their Amazon. Two or three months later, they're in Instacart and they're setting that up and they're getting into their retailer accounts. And it's like once you look back, it's like that isn't consistent because they

They've set them up as the brand has evolved and AI awards consistency. And so

Jordan Buckner (08:41)
Yeah.

Andy Kurtts (08:41)
you need to make sure, you have this thoughtful strategy going into, either filling them out if you're new to this, or going back and auditing it, making sure all the fields are consistent and the correct data is there. You're getting credit for

Jordan Buckner (08:54)
Yeah, I know

Andy Kurtts (08:54)
it.

Jordan Buckner (08:55)
Yeah, it's interesting too because one thing I've always been fascinated about is how

Different platforms for discovery, whether it's in person, online, Amazon, you get different learnings around how consumers shop, right? So like walking

Andy Kurtts (09:09)
Yeah.

Jordan Buckner (09:09)
out grocery aisle, certain things stand up, visuals, colors, things like that, in a different way than on Amazon, where it's about high conversion keywords and what people are mentally searching for versus like

Andy Kurtts (09:22)
Mm.

Jordan Buckner (09:22)
emotional search with their eyes or so. And so, I'm curious of how you're thinking.

thinking about how some of this might show up then back on the packaging. So then is it like adding certain jobs to be done keywords on packaging now as you're thinking about it or how does that start to inform that version?

Andy Kurtts (09:38)
Exactly. That's exactly what we're thinking. so there's is certain things let's make sure that seed oil free claim it that certification is slightly larger. that there's just hierarchy that needs to come into play. But then there's also to a degree, maybe not necessarily adding like

Literally a job to be done, but just adding a little bit more detail to your product name or to your product type. And but maybe if you're low on claims, you could put, great for an afternoon snack as a call-out, But it is we are trying to think about that and look at opportunities, still have aesthetically pleasing packaging, but where can we tuck in a certain line or something like that?

Yeah.

Jordan Buckner (10:17)
Yeah, and that's actually a good point too because I love the idea of starting to use some of the data around it. it's interesting, I have not seen a ton of published data around how people are what people are looking for around some of these AI tools, right? I know you can see Google search result rankings, you can see Amazon search rankings,

There's probably some out there already on how people are using AI because I remember with our T Squares packaging, we had three icons on there and those things are afternoon snack, good for energy, but it was very much like a brand driven language, less so

Andy Kurtts (10:50)
Mm-hmm.

Jordan Buckner (10:50)
like how consumers are actually searching. And so I wonder if that starts to play up of looking at what are some of these high conversion search terms that people are looking for, even using some of the Amazon data, what are people inputting and how can you use that?

language on your packaging versus a made-up RAN language so you're like, this represents our personality as a brand, but it's not actually what consumers are searching for in their chats.

Andy Kurtts (11:12)
And it

And yeah, and I think that is for the brands to reckon with. it's like how far do we take this so that we don't lose our soul, but also show up in these searches. And so I think, once they can look and see here's our, very unique and custom authentic brand language, and then look through the list for their category of like wherever that

data lives of what people are searching and then see where this the overlap and then how can you tweak that language because they yeah

it's a probably a good just exercise for a brand to take a step back and say are we doing a disservice to the customer by using this very specific unique brand language? whether it's search or whether it's, literally trying to just speak to a customer on shelf or whatever. 'cause I think brands can get a little they can trip over themselves,

and a little too into in the weeds of that and lose sight of you're talking to normal people everyday people so yeah that's a great point though and I do think that obviously as more and more agents shop for people more and more data's gonna be collected and then we'll start to understand and see how to figure out

how brands can yeah, bring that those search terms into that brand language.

Jordan Buckner (12:20)
Yeah, one thing that's interesting too, which I'm curious about, is that I feel like a lot of AI results

When you don't specifically prompt it, we'll prioritize more popular items, whether that's more popular businesses or food brands or larger companies, just because there's more data out there about them. And that leads

Andy Kurtts (12:38)
Yeah.

Jordan Buckner (12:39)
to a signal, right? if there's a million mentions of a brand versus twenty, unless you're asking for the one with lower mentions, there's gonna be a there may be a bias towards that higher mention brand as well. And so I'm wondering if there's also a piece around

more data points and getting those onto the internet, right? Is it having more images and maybe the images have more text about the products and some of those keywords in them? maybe some of it is actual physical packaging

But what the AI is actually scanning are images of the physical packaging. So maybe it's actually the image itself. And so can you put in different cer like keywords or things into the image, but necessarily it's not necessarily on the physical packaging because right, like

Andy Kurtts (13:17)
Hmm. Yeah.

Jordan Buckner (13:20)
if you have a an Amazon picture with the prop in the middle and a white background, can you put white text on top of the bite white background, but in a way that AI can pick up and read but it's not the actual packaging itself.

Andy Kurtts (13:31)
Yeah. No, I love that. And I do think we're gonna have to be creative about that. F

All while following the rules, you don't wanna get banned or your hand slap, but I think there probably is opportunity there, from what I can tell, specifically because we have been narrowly looking at with the vision models specifically reading front of pack. They're not reading necessarily the supporting content on Amazon,

your thumbnails and your product information. And so that's where our focus is is, if you're already going to adjust your packaging, refresh your packaging, for this, let's make sure obviously it works on each shelf. but no, I think it's an interesting thought in how

you optimize those supporting images wherever they live. Because the AI is looking everywhere. It's looking at your website. It's looking at, the yeah, mentions, reviews. yeah, just any touch point of your brand is worth thinking about how does this visually come out there and is it gonna give us the most bang for our buck for this visibility? yeah, that's great,

Jordan Buckner (14:31)
Yeah.

Andy Kurtts (14:31)
great thought.

Jordan Buckner (14:32)
I know that you and your work in brand design, there's a lot of these elements in terms of

having the understanding of how the industry works, some of the fundamentals with actual hierarchy, what information to put on the packaging. There's also the visual and design aesthetic component. I'm curious as how different AI tools may or may not be impacting like your business and how you're utilizing them or not utilizing them as you're building and continuing to grow your agency.

Andy Kurtts (14:59)
Yeah, we use it in terms of a lot of administrative stuff. And so, that's it, just like a lot of people, i it's our executive assistant that helps us do tasks like keep up with projects and help us write proposals and things like that. we

have noticed and it's been interesting and we've tried to to approach it in a friendly way I guess an accepting way when clients

Jordan Buckner (15:22)
Mm-hmm.

Andy Kurtts (15:22)
send us ai images that they want like I think people who founders who have a hard time

Because they don't have a design background, have a hard time articulating what it is in their heads. These are massively creative people, very smart, but they don't have the tools or vocabulary or whatever to get what is in their head on paper. And I think any way that any tool that can help them do that is great.

That doesn't mean that we what a client gives us that they just prompted chat to do, we're gonna take verbatim. But it definitely helps us get into their head and see, okay, this is what they mean by that. okay, I can I'm tracking with them. But we do we accept them, it's okay, but we're not gonna take it and just run with it, know, and so yeah.

Jordan Buckner (16:08)
Right. take this and

then do the final version of it.

Andy Kurtts (16:11)
Yeah, yeah, yeah, exactly.

Jordan Buckner (16:13)
Yeah. I think that's a good point because

Andy Kurtts (16:15)
Yeah.

Jordan Buckner (16:15)
a lot times founders would traditionally send over maybe a mood board of grabbing random images of brands who they like, color, style, and then there's always that translation piece of okay, what is what is the essence of what they actually like? Is it the shape? Is it the color? Is it the juxtaposition between two things? are they the images, right? The photo realistic versus maybe a drawing style, and that those pieces are hard to call out. And so

I do like the idea of it enables founders or people who are not doing the final design to express their ideas in different ways. I feel like have a a part of that process. But I agree, I've done some just test packaging designs for different things and you can get something, right? Isn't it know so

Andy Kurtts (16:58)
Right.

Jordan Buckner (16:59)
when we launched our first packaging for TeaSquares one of my co-founders is a designer but comes from the web space, and we got a packaging out there in the world, but actually failed on shelf.

And it failed on shelf because we didn't understand communication hierarchy and what shoppers were looking for, which is a different

Andy Kurtts (17:14)
Yeah.

Jordan Buckner (17:14)
part of the design science within there that's also necessary to know per category, per product, when to do things that are standard in the category versus when to knowingly push the limit on things. And that comes from that hands-on experience.

Andy Kurtts (17:27)
Yeah,

exactly. And I think it's we've noticed that mood boards can be very conceptual, and it's and some founders have a hard time bridging that gap of either they literally want a font that is included on a mood board and they don't understand that it's just inspiration and it when it so I think

they can be helpful, but I think getting something very close to what they have in mind for one concept just, speeds things up immensely. By no means does it mean, that might even not even make it into our, two or three concepts that we present to them for that first round. But it was just a data point that we were like, okay, we get what they're trying to go here go for and then translate it.

So yeah, we try to be accepting of it. I know that it is very controversial, especially for designers who feel like it might be taking their jobs, but yeah, we try to embrace it to a degree, especially when a paying client is giving it to you.

Jordan Buckner (18:18)
Yeah. Exactly. I

know we started this conversation talking about that coalescence between the d digital information space and physical through the packaging. If you're tracking where things might be going over the next couple of years, what's

What advice would you give to founders to help them set themselves up for success in navigating that balance between physical design and this digital space, whether it's like through AI search or having something understand their packaging? what would you leave founders with?

Andy Kurtts (18:48)
Yeah, I think it is a great time, now, is a great time to reassess your front of pack, your packaging design, and just take a pass and see where the opportunities are around it doesn't have to be a huge facelift. It can be just increasing sizing of certain elements, or adding a one line to speak to jobs to be done or just a different descriptor. there's opportunity on the back panel to include

include those occasions and jobs to be done as well because some of these vision models read the back panels and that marketing copy. So I think right now you're gonna want to do that. And it's a light touch, you don't have to go crazy with it. and then from the data structured text perspective this is also a great time to look into those

go back into those platforms that you set up, make sure all those data fields you might assume that seed oil free is in your Amazon listing, but

Go and confirm it. Make sure it is. It might not be. Or you might have had a certification come through that you just haven't input in there yet. So just to go through, refresh yourself on all that stuff. because that's all gonna be those points of truth for these agent shoppers. And you want to make sure, and it's only getting bigger and bigger, And so, I think, 25% of people said they would be okay with an agent shopping up for a meal for them. So that adopts.

is only gonna get higher and higher. So right now is a perfect time to touch both the physical and the data text areas.

Jordan Buckner (20:11)
love that. Andy, thanks so much for being on and sharing all this. Definitely insightful and something that founders need to be thinking about. it's actually I get excited because I have been guilty of setting up various platforms and not going in and

Andy Kurtts (20:22)
Ha ha ha.

Jordan Buckner (20:23)
and editing and updating them. And so that's something that I'm gonna be working on as well and recommend our listeners do as well too. So thanks so much for being on. If you need help with branding, creative design, Andy has worked with tons of amazing CPG brands and put out really great work into the world.

And so if you want someone who's thinking about this cutting-edge stuff so that you can be ahead of the curve, definitely reach out to him and his team. I'll put info in the show notes. Any, thanks again.

Andy Kurtts (20:47)
Awesome. Thank you so much, Jordan.