Live from Cannes Lions with Jeff Cohen from Amazon Ads

Transcript

Our transcripts are generated by AI. Please excuse any typos and if you have any specific questions please email info@digitalshelfinstitute.org.

Lauren (00:44):
Hi everyone. We are live from Cannes with Unpacking the digital shelf and I'm here with Jeff Cohen.
Jeff (00:49):
Hey, great to be back with you. Last time we were able to kind of do this while we were here too, so it's so awesome to have you back and back in the Amazon Port studio and creating some awesome content for all the viewers.
Lauren (01:02):
Yeah, year two. It's a little hotter.
Jeff (01:04):
It's a little, yeah. Sticky is the word of the year.
Lauren (01:07):
A little sticky. Just don't look at us on camera, but so excited to do this. Let's chat a bit about what's different about Can this year? Any observations, any themes? What are you seeing?
Jeff (01:15):
Yeah, I think one theme is the retail media networks seem to have even expanded across the Croce, if I'm pronouncing that corset.
Lauren (01:29):
We're clearly both not
Jeff (01:30):
Right? I think that there seems to be more people
Lauren (01:35):
Here. I agree. And younger,
Jeff (01:36):
Right?
Lauren (01:37):
More influencers and creators.
Jeff (01:39):
Yes, influencers and creators are definitely, it was starting to build as a theme last year. It's definitely a stronger theme this year. I think that obviously creative is still a strong theme here and AI tends to be kind of dominating the conversation.
Lauren (01:55):
Agreed. But do you feel like AI is kind of out of the hype cycle and it's now more in a place of practical examples and how we're using it and how it's really going to enable us to do more? I feel like there's more of that.
Jeff (02:07):
Yeah. I think last year we were trying to figure out, well, what is AI and what does it mean and what's it going to be and what's it going to deliver? And I think that we're starting to see some of the development of those tools, right? Amazon has developed a series of tools that were released at unboxed last year around the creator studio that allow you to go from an ASIN to an image to a video. And so now we're seeing, I mean, one of the candidates for the awards that was a finalist was a totally AI generated
Lauren (02:38):
I know, crazy
Jeff (02:39):
Campaign. So I think that it's way more in practice today than it was a year ago,
Lauren (02:45):
But let's go a couple years back and think about how it's changed what you're seeing. What are the big differences, how we should be thinking about ai?
Jeff (02:54):
Yeah, I mean, when we look at a company like Amazon, AI has been in our DNA from the beginning. A lot of our optimizations are driven by large language models and machine learning, which is all a function of ai. I think the big introduction with chat GBT and those type of tools was the ability to start having chat-based ai. So I can now ask it questions and it can give me answers. We're now moving into this world of ag agentic ai. Well, I guess between that was the copilot phase, right? True. The copilot phase was where it sat over what you already had and was kind of providing you some insights and some recommendations. We started to see this, I'll say commercially with tools like a deep research where we started, where AI was starting to understand what we were looking for, and then it was taking all of that to do a bunch of research and time, so it wasn't an immediate response like chat and coming back to us with a more polished answer, if you will.
(03:57):
And so when you look at some of the tools our partners are creating, we have two partners who have released a agentic AI type tools, sky and Exner. And while they're different in what they do, they kind of have the same general functionality, which is that you tell it what you're trying to do. I want to build an MBR R-A-Q-B-R-A report from my executive manager. It then comes back to you with an outline of what that report could look like, the type of metrics that it's going to follow, the type of analysis that it's going to do. And then once you align with the AI on what you want, you do a little bit of back and forth, it then goes out and it does that work for you. And it's like having a whole bunch of business analysts who go out and take all these different pieces and pull all this data in, and then it collects it all, and it presents it back to you in a way that with one click could be downloaded to a PDF or a Word doc for you to then share with your client or share with your boss whether it's internal or external.
(05:02):
And I think that's the next phase. And we're seeing some of that in the commercial world. There's obviously, there's deep research that some of the tools have, but even I use Claude. It's one of my personal AI tools. And when I write with Claude, and I fully admit that I write with Claude, and
Lauren (05:19):
I mean it's faster.
Jeff (05:21):
I did a whole blog post about it or a LinkedIn post about it, and I'm happy to share it with you because yeah, please. I had learning disabilities, I have learning disabilities, I guess you never really fully grow out of them. And it helps me with the areas where I've struggled. And so I can use the voice as a way to express my meaning, but I don't always put 'em in the right words. And so all of the thoughts are mine and they're original. But what Claude does is when you are writing, it actually opens up a second panel and it becomes kind of like your draft pad and you can say to it, change this out and make it do this and change this and make it do that. And so I can actually make it and perfect it within the system before I then copy it to take it to how I want it to be finished. And that is, I think, a great example of how we're using these types of technologies that are different. And obviously we've also had the expansion of shopping bots,
Lauren (06:21):
Rufuss,
Jeff (06:22):
Right? Rufuss with Amazon and others that are out there that are just allowing us to do things that, and to start experimenting in ways that we weren't doing before.
Lauren (06:32):
And I think you addressed two things. It'll change the way we work internally in our organizations, but it'll also change the way consumers shop, the way they experience shopping, the way they think about shopping. So you mentioned agentic AI and AI agents. Let's talk about conversational search, which I think is really exciting, right? Because it's not a point and time. It's like, Hey, I'm planning a dinner party. What do I need? How is that though going to change how brands think about their strategy and think through
Jeff (06:58):
Things? My simple answer is it shouldn't change your strategy. You should never develop a strategy to work for an algorithm. You should develop a strategy to have sound basis in what you do. But let's talk about some of the social trends that maybe are new influencers in that, right?
Lauren (07:19):
Yeah.
Jeff (07:19):
So there's social commerce, there's social media, there's ratings and reviews, there's product attributions. All of these things become data signals or data that's feeding into the large language learning models that are then sharing the answers that are coming out. And so if you have a social media campaign with influencers who are creating a bunch of content, it could actually, and that content does well, it could be a signal to a large language model that this is a product that people are interested in and people have tried. Now, I'll give you an example of,
Lauren (07:57):
Ooh, I'm ready
Jeff (07:57):
Of how shopping commerce can potentially work. I was planning a trip to Miami
Lauren (08:04):
Just as hot as can.
Jeff (08:05):
Yes. It wasn't when I went, but I decided to let the chat develop a trip for me. But I told it what I like to do, where I'm staying, what I like to eat, what I don't like to do, what I don't like to eat. And it gave me a whole itinerary. And then of course in the chat, I could be like, well, I don't want to do that on my second day. I want to do that in my third day, and I don't want to go there in the morning. I want to go there in the afternoon. And so then when I asked it something else, it was like, well, do you want to go there in the morning or afternoon? So it learned that I had a preference of mornings and afternoons. It learned that I had a preference for certain types of restaurants and I didn't like others. I would say that the restaurants were, I'm going to go 50 50. We ate at two. One was a hit, and my wife liked it, although it was a very social influencer type of restaurant, lots of
Lauren (08:55):
Pictures.
Jeff (08:56):
We were at the Versace Gardens
Lauren (08:59):
Cool.
Jeff (09:00):
Which the food was actually very tasty, but everybody was taking pictures of themselves there. It was kind of comical.
Lauren (09:06):
Influencers in the wild,
Jeff (09:08):
Great people watching.
Lauren (09:09):
Yes,
Jeff (09:10):
People literally had cameras set up at their dinner table, the whole food and everything.
Lauren (09:15):
But can we just talk about that for a second? Because I think that's moving towards more experiential shopping. That's what consumers want. They want to feel connected. And I think AI agents are giving you the ability to create those scenarios where it would've taken really long to identify what that trip looks like so you can spend more time on building out a better experience.
Jeff (09:37):
Well, the second restaurant that we went to was a brunch beach bar a lot with a lot live dj. And
Lauren (09:47):
In the morning,
Jeff (09:48):
In the morning
Lauren (09:49):
It is Miami.
Jeff (09:51):
And that's when I kind of realized that the vibe that it was giving back to me in both of these places was it was really looking for trendy places. But I don't know that I was in the age group of who it was actually trying to connect to. So that means, but
Lauren (10:06):
It had a lot of content about it, so that's why I was pulling all that information.
Jeff (10:09):
Correct. And so you then have to train it on what you like and what you don't like, and it will hopefully improve over time. And you just have to understand that. And I think that's the same as even with the agentic models, when you're asking it to run reports, the first report that comes out may not be exactly what you want. And if you look at that report and you go, well, this isn't what I want, this tool doesn't work for me, you're not giving it the time that you need. Got to give it the energy that it needs to get to the point of being helpful for you. But
Lauren (10:42):
I think that's a perfect example. If you don't create content about your product and it's not in all of the places that consumers shop, then the AI agent is not going to pull in that information and make any of those recommendations.
Jeff (10:51):
And so the strategy is more about how you're doing content creation, not how are you developing for the algorithm.
Lauren (10:59):
And let's talk about creative and creation of creative agents, how you're thinking about the suite of tools that you have to create content. How do you think that's going to change the way brands work, the way retailers think about their experiences with images on the PDP?
Jeff (11:14):
Yeah, I think that we think of the tools that we're creating around really democratization and efficiency. So democratization, meaning that we want the tools to be available to all. So whether you're a small seller who has no team and you want to start with an ace and then go, or whether you're a large seller and you have prime video ads or other social media ads that you're running, national TV ads that you're running, you can upload that into our creative studio and it will make all the individual pieces of content that you need. And so that's how we're thinking of it. And then the efficiency comes in and the ability to change your content, making your content seasonal, focusing on micro seasons, focusing on tent pole events. And those are the things that I think are challenging for brands because Prime Day is coming up, 4th of July is coming up, what if you want different images for 4th of July, but then you want to change them all out for prime day? Well,
Lauren (12:13):
That's a quick turnaround.
Jeff (12:15):
It's a quick turnaround. But with a tool like that, you, because you can make different images for a lot of your PDP pages. Now, I always tell people, focus on the 80 20 rule, focus on the 20% of your PDP pages that are driving 80% of your business. But ultimately, as you're making those, you can look at Amazon Marketing Cloud and understand how your sales are doing during those periods of time. You can't necessarily directly correlate what's happening with the image, but you can start to see what type of influence that image creation has and whether you want to keep spending time on it.
Lauren (12:51):
And you announced Prime Day dates, right? July 8th to the 11th. Did I get that right?
Jeff (12:55):
You know what, I think it actually happened while we were here. It did. I saw that. I don't know them off the top of my head. I actually had somebody share with me that I have my Prime day dates, and they showed me the tape on their box, which I found funny. So yeah, I think it's around the second week of
Lauren (13:11):
July and it's four days this year.
Jeff (13:13):
It's four days, yes.
Lauren (13:14):
What are your thoughts on that?
Jeff (13:15):
Well, listen, bigger, better, more opportunities. I think that Prime Day has gotten better and better for both our shoppers and our brands every year. I think that brands need to really go back and look at what happened for you for Prime Day last year? What did you do? What did you do wrong? And use that to analyze what you want to be testing into this year. I don't necessarily think that means you necessarily need to run a four day deal, although you could look at what you do for Cyber five or Cyber 11. It was this year. It seemed like Thanksgiving was so late that all
(13:56):
The, it was really late. All the sales started early. But we've done, we've had five day periods between Thanksgiving Day and Black Friday and Cyber Monday where you've used promotional strategies and email strategies and other things that you've used to really energize your brands. Some of the interesting data that was shared on LinkedIn demonstrated that there were bumps each day at different time periods. So go back and look at your marketing, Amazon marketing stream data to understand where those inflection points occurred and use that as knowledge and direction for where you should be putting your effort this year.
Lauren (14:40):
And speaking about announcements, you had some other announcements happen at Cannes, right? Other than Prime Day?
Jeff (14:44):
We did, yeah. What do we
Lauren (14:45):
Got?
Jeff (14:46):
There were I think three or four, if I get these. All right. We had a really exciting one with Roku where we're connecting our signals with Roku signals to be able to cross identify people who see Amazon ads on Roku devices and be able to measure the success of those. I think it reaches just around 80% of connected TV viewers, which is That's huge. Yeah, it's really exciting. And it's another example of where Amazon is working, not just in our owned and operated, but in our third party supply, which leads to the second announcement, which was with Disney, which is creating identifiers between Disney's system and ours, which again is giving you the ability to buy Disney, Hulu, ESPN and their other properties through the Amazon third party supply. We also announced Amazon Live and their data being in A MC. We announced some updates to our video generation tool, which was super exciting. And really the one that I was really, it actually struck me because a lot of pride of the company. I work for Andy Jassy, our CEO, I guess if you don't know who he is,
Lauren (16:06):
Anyone listening,
Jeff (16:07):
Just in case you didn't know was named the 2025 Media Person of the Year saw that. So he's actually accepting his award right now while we're recording this
Lauren (16:18):
Right across the street.
Jeff (16:20):
And so I look forward to hearing what he had to say during his sit down fireside chat. I'm hoping that he walks around Port and I have a chance to maybe meet him. I guess I'm a fanboy of the company that I work for, but I was also a fan boy before I ever came to this company.
Lauren (16:38):
You just have to figure out who knows him and you can just stand in the right place and wave.
Jeff (16:43):
The people that know me and know him might tell me not to go near. He may not want a picture. I don't know. Although I've seen him in offices taking pictures with people. So
Lauren (16:51):
Well, this is the place to run into celebrities.
Jeff (16:53):
It is. Yeah. I ran into Dirk Naski and Blake Griffin from the Amazon Prime NBA team last night at the
Lauren (17:03):
Concert. That's awesome.
Jeff (17:04):
Yeah, that was kind of cool. My son actually thought I was cool for that.
Lauren (17:08):
Wow. Very cool.
Jeff (17:09):
Got some cool dad points.
Lauren (17:11):
Yes, you got points. You're cool. I love it. I have a funny celebrity story. I walked up to someone on the Quas set who I thought was someone else, and it was Dylan Ephron. I was like, Hey, Sam. And he's like, Uhuh, I'm Dylan. And I was like, oh my gosh, you are nice to meet you. So I had my celebrity running.
Jeff (17:28):
We were talking about influencers as being a theme. And what's really interesting is my daughter's 19, my niece is 21. Both of them have texted me.
Lauren (17:36):
They're like, you're at Cannes?
Jeff (17:37):
Yeah. They're like, did
Lauren (17:38):
You meet anyone? Cool?
Jeff (17:39):
No. They're like Every influencers at can. And it's like they've known, I've been coming here for years, and it's just kind of like, oh, my dad's going to France. Oh, my dad's got a trip to France. It's like this year when we talk about the rise of influencers in this space and what's happening here, that's what I mean is it took a different level.
Lauren (17:58):
I
Jeff (17:58):
Agree. And if you're here or you want to come next year, they are all hanging out at the Carlton Hotel. They see
Lauren (18:05):
TikTok,
Jeff (18:06):
Right? All the sports people and all the influencers are there, so just lobby lurk there, and you're guaranteed to run into a few.
Lauren (18:14):
I love it. We're going to have so many people just coming to can next year being like, can I see someone famous?
Jeff (18:17):
Right.
Lauren (18:18):
Okay. To close that, I want to do something fun. I'm going to hit you with a topic, a thought, and then you tell me your perspective. Yes. You ready?
Jeff (18:24):
Sure.
Lauren (18:24):
Okay.
Jeff (18:24):
I do really bad at these.
Lauren (18:26):
You're going to crush it. Will websites matter in 10 years?
Jeff (18:30):
Oh my Lord. Will websites matter? Yes. Websites will matter because brands will always need a home for telling their story. And I do believe that the D two C market is still thriving and will continue to thrive. And Amazon believes that too, because if you're not familiar with this, you can buy ads on Amazon and drive traffic to your D two C site. So we clearly believe that the consumer wants to choose where they're going to shop. And so I do believe that websites will matter.
Lauren (19:03):
And I agree with your place of telling the story, right? Yes. Because you also need a place where your data exists that the agent can scrape it to be able to pull it in until we get to an agent to agent world, which is a little bit of a different story, but I do agree that they will be around, but in a different capacity.
Jeff (19:17):
Yes.
Lauren (19:18):
Okay, second one, ready? Is AI going to shop for us in 10 years without us intervening because they know who we are?
Jeff (19:26):
I'm going to maybe answer it more broadly and say AI is going to take actions for us. And I do believe shopping will be one of them. We're already starting to see it. I don't think we need to wait 10 years. So Alexa Plus was announced maybe two months, three months ago, I had a sit down conversation with the chief evangelist for Alexa. The technology within Alexa plus is game changing, and it's super exciting. I haven't had the chance to play with it myself just yet, but you're able to see things within the Alexa devices and tell the Alexa app to take actions, whether that action is schedule a haircut or whether that action is order food, or whether that action is buy something. And so I don't think we're that far away from the world where it can take action for you. And again, I think it'll go back to the story that we told, which is like, you're going to want to be comfortable. So you might not say, go buy me dog food, but you might say, Hey Alexa, I need to buy dog food. And it might say, okay, the last time you bought dog food was this, and this is what you bought. Is this what you want to buy? And you say Yes, right? Or dog treats or whatever it is. You want to insert that
Lauren (20:36):
Into it. Buy a lot of dog food. I have 105 pound dog.
Jeff (20:39):
Oh, that's a lot of dog
Lauren (20:39):
Food. Yeah, a lot of dog.
Jeff (20:40):
So I do think that we're going to have a world where that exists.
Lauren (20:44):
Okay. I agree. I think that it will help us do things faster. It will learn our trends and what we've purchased and help us kind of predict some of those future experiences.
Jeff (20:54):
Right.
Lauren (20:55):
Okay. Last one. Ready? It's 2030 and you need to, I don't know, either go to the grocery store, you need to plan a trip, you need to buy a new backpack. How do you do it?
Jeff (21:07):
20, 30. So that's five years away.
Lauren (21:09):
Yes. I have to think about that. Yes. In five years
Jeff (21:12):
I wish you used fives. It could have made it really hard.
Lauren (21:14):
Even numbers.
Jeff (21:17):
I think I'll be using, I don't know if I'll be using AI to do everything that you said, but I'm already planning trips. I mean, I shared one, but I've planned a couple of trips with ai. I haven't gotten to the point where the AI is just kind of doing everything for me in terms of booking the hotels and booking the air and things like that.
Lauren (21:35):
But it's not there yet.
Jeff (21:36):
It's not there yet. But I do think there are some things that it can do. And so I do think that AI will be a big influencer in that maybe I'm older and more skeptical, so I'll probably add inspection to it. And so the term that I've been using a lot this week is this term of inspection, which is that AI, by its nature wants to provide you a result. And it's not necessarily graded on how right that result is it hallucinates. Right. And if you ask it in the general ai, if you ask it the same question multiple times, you get different answers. So as we get more refined and in less hallucination and it understands our personalization more, it'll become more trusting in the things that we do. I can tell you this. As I said, I have a 19-year-old daughter, an 18-year-old son, and a 14-year-old son,
Lauren (22:34):
And they think you're cool. We have to reiterate that.
Jeff (22:37):
Yeah. I talk to them about AI every single day ad nauseum. How
Lauren (22:43):
Do they feel about it? Actually? Are we
Jeff (22:45):
Talking
Lauren (22:46):
About it? No, no, no, no, no. Not necessarily that, but about ai. Is it just ingrained into their vocabulary?
Jeff (22:51):
Not yet. I would say that they were very hesitant to it because they've been told by their teachers that they can't be using it.
(22:58):
But both my older children are, one is at Ohio State, one will be attending Ohio State, and Ohio State was one of the first universities to announce that they're going to have an AI literacy program. So when my son starts at school his first semester, they'll be teaching AI from the get-go. And I commend them for doing that because that is what's important for these kids. You can't fear technology. You have to embrace it and understand how it's going to impact you, and you need to learn how to use it the right way. And I think that's the biggest lesson that I want them to learn. I don't tell them to Google things anymore. I tell them to use AI and get the answer.
Lauren (23:38):
Yeah. I love it. Awesome. Thanks for the chat, Jeff. This was great.
Jeff (23:42):
It was great to see you. And we can't wait. Maybe we'll do it again next year maybe, or I'm sure we'll see each other
Lauren (23:46):
Before, or maybe we'll be in a spaceship someday on can, I don't know. Predictions.
Jeff (23:51):
Right.
Lauren (23:51):
Thanks
Jeff (23:52):
Jeff. Welcome to talk to the other CEO, Jeff Bezos, and you could send us up to space to record the next one
Lauren (23:56):
On the moon. Amazing. Next podcast. See you there.
Jeff (23:59):
Take care.
Lauren (23:59):
Bye. Time. Awesome. I liked that. I liked where we went. And then I liked the end of.