Latest / Talk Commerce / AI Agents Are Reshaping Retail Media for CPG Brands with Sai Koppala
Transcript
- Brent Peterson (00:01.388) Welcome to this episode of Talk Commerce. Today I have Sai Coppola. He is the CMO at Commerce IQ. Sai, go ahead, do an introduction for yourself. Tell us your day-to-day role and one of your passions in life. Sai Koppala (00:14.797) Sure. Hi, this is Sai here. Glad to be on the podcast here. I am the head of marketing or CMO at Commerce IQ. I've been doing one idea of marketing leadership roles in growth stage companies the last 10, 15 years in Silicon Valley. And outside of work, I have two boys and they're into sports. So I'm an active cheerleader on weekends. Brent Peterson (00:41.524) That's perfect. Cy, before we start getting into content, we'll go to, we'll talk a little bit of go to market and about some, your customer advisory board that you had talked about earlier. I do this thing called the free joke project. I'm just going to tell you a joke and you just give me a rating eight through 13. So here we go. I asked my wife, will you still love me when I'm no longer attractive? She said, I do. Sai Koppala (01:10.243) As a married man for 21 years, I'll give that a 12 Brent Peterson (01:15.534) All right, perfect. Thanks. I appreciate that. right, give us a 10,000 foot view of Commerce IQ. Tell us a little bit about what it does and how it helps your customers. Sai Koppala (01:30.797) So we primarily work with CPG brands globally. We help CPG brands optimize their retail media spend, digital shelf, and sales using AI agents and human experts. Brent Peterson (01:50.254) Perfect. Talk a little bit about the AI agents. You know, I think about a year ago, I was at eTail West and AI-gentic was the new buzzword and now it's an old buzzword, but the idea of AI-gentic agents coming and shopping hasn't quite come to life in retail, but on the website and from maybe behind the scenes, AI agents are much more prevalent. Tell us your idea of where AI agents are and where they should be. Sai Koppala (02:19.255) In many ways, this is a reaction to what is happening on the retailer side. If you look at Amazon and Walmart, they've been doing it for a while, whether it is winning the buy box or what price you are to bid for your particular keyword to show up on your spend or what's a share of voice on their site for a particular keyword. That's all automated today on the retailer side. They are using a set of signals. including sales, ratings and reviews and content descriptions to show the right products to the right consumers. All of that is automated today, On top of it, in the last year or so, we are seeing agents like Rufus, Amazon has launched, or Sparky Walmart, where even the consumer now is engaging agents on the retail websites to figure out what product they want to buy. So it's no longer your website or your product description page is the front door. It's actually these agents which has been spawned by either retailers or even literally Chad's GPT, right? Or is becoming critical in terms of the consumer flow. given that context, the... On the brand side, can't rely purely on your human experts and analysts and human power agencies to counter that. You need to have automation on your side. Therefore, lot of this has been happening in the back end, not visible. But increasingly, we're realizing, given retailers investing more and more agents, brands have to invest in agents to really automate some of these workflows. to be able to react at the speed at which the retailers are reacting. Brent Peterson (04:18.71) Yeah, that's really interesting. think the traditional idea of AI has, or not traditional, what consumers see AI as is genitive in making videos or making images, but really the magic is in behind the scenes and giving, empowering, I should say, the merchant to be able to serve up precise content to the users. And even then that's emergent to find places where they can improve. Tell us a little bit about some of those things you've been learning from your customers. Sai Koppala (04:54.435) Yeah, let's take a simple example. If you look at retail media, over the last year in the retail report we did on 2025 e-commerce in the US, retail media spend actually went up 20 % across the board. And by the same time, return on ad spend has gone down about 8-something percent. Your cost per clicks have gone up by 8%. So it's not the most efficient and today in most cases what a brand would do is they would work with an agency to manage their retail media spend. Hey, use my retail media spend dollars and optimize it for return on ad spend. Even with the best rules-based automation which been around for a few years, what people call goal-based optimizer, you probably can do a few hundred automations a day. But if your bid amounts or bid prices are changing second by second, if you use AI and use that to optimize whatever goals you're optimizing for, it could be a CPC, could be a ROAS, you can do 4,000 plus automations, optimizations a day. You can figure out based on your share of voice, for example, what we do in our case is we use 50 plus signals, including real-time signals on share of voice and inventory availability and things like that, and also keywords. What are the highest performing keywords? What are the long tail keywords where the cost per click is lower? So we are able to optimize to the nth degree all the long tail of your skew items across the long tail of keywords all using AI. That's just not humanly possible. And top of that, we still need the human in loop, frankly. Yes, the mechanics of it, the execution of it, which is in many ways, you really don't want a human to do that because it's hands on keys again again, right? That is automated. But the strategic aspects of it still is human driven. So you need a combination of both AI from an execution perspective, but at the same time, a human in the loop and human to work with our brands to identify what are the Sai Koppala (07:19.299) priorities for you. What do you want to optimize for? And therefore coming up with a strategy. So you need a combination of the two to succeed. Brent Peterson (07:26.658) Yeah, you've mentioned long tail a few times there. Is the long tail now really in play with, especially with the use of AI that you can really find these long tail words and these words that people are naturally using to search for in terms of finding your products? Is that kind of the new frontier for PPC? Sai Koppala (07:51.243) Yes, it's not just, it's accelerating even more, right? Because if you look at how consumers are changing their behaviors, they're asking for specific questions. I'm doing, for example, it'd be like, I'm creating a specific recipe and I want to search for items for that. So what item should I purchase? And it gives a list of items. So very much because the consumer discoveries change to more Q &A style, how humans speak, the agents like, Rufus and things like they're also prioritizing those kind of things. So therefore downstream the brands also have to optimize their content so that it gets surfaced by the Rufus and Sparky's of the world. So you are right we Because the consumer behavior is changing and how they are searching for stuff on Gemini or you know, Chad GPT and also retail websites like Rufus and Sparky That is downstream impact in terms of how brands need to optimize their content and media spend to surface their power. Brent Peterson (08:58.03) Do you think consumers right now are patient with some of these brands? And I've used the Walmart version and I find it, I found it in the very beginning anyways, not that great, it didn't find what I wanted. And even I would say that Rufus isn't the best LLM out there for finding what you need. And you would think since they have such a big catalog and because they have such a huge tech budget that they could make something better than even OpenAI, right? But are consumers patient right now to deal with what they're seeing and knowing that something is going to come out that's better. Sai Koppala (09:34.435) Andy Jassy in the last earnings call said that 10 billion of their transaction on Amazon has come through roofers. Is there room for improvement? Of course, this is still, I would say, early, early days in the world of AI agents, right? And the technology is just improving at such an exponential pace. In a matter of a year or two, it will be significantly better in terms of, you know, adapting to consumer needs. Yes, is it ideal in terms of interaction? No. But even with that, you're seeing so much amount of revenue going through it, it'll only get better. So it's no escaping that. Brent Peterson (10:19.648) Is there some risk in retail media with LLMs and them suggesting other products? And especially you as a consumer, you're asking for a product or something like a product, and theoretically somebody's paying for that ad, but it's also going to give them something else, right? I think is there a risk for you to get shown up first, but then somebody, something else is going to show up as well? Sai Koppala (10:46.563) Frankly, it's early days, I can come up with an answer, but it's early days still figuring out how these agents are surfacing the right products. We know that they rely quite a bit on ratings and reviews and Q &A because one of the things they're realizing is the best, because on the one end you have consumers talking. asking like a human being the question say what are the ideal products for what I'm looking for. They're reliant quite a bit on reviews because which also return by human beings to surface the right things along with PDP description and additional description that fields that are provided. Having said that, it is early days. This is one of those things where the entire industry is figuring out as we speak. Brent Peterson (11:42.86) Yeah, you had mentioned to go to market. What is your kind of go to market strategy that has changed dramatically in the last couple of years? Sai Koppala (11:54.069) in terms of working with brands. Brent Peterson (11:56.237) Yeah, with the brand, yeah. I'm assuming that the go-to-market strategy has to change just because of the dynamics of the market and the way it's changing. Sai Koppala (12:03.469) Yeah. Look, we were primarily an analytics company a few years ago, where we would enable brands to, hey, we can tell you what happened. We can look at all your data from the past and tell you what happened. And then the industry shifted to not just telling you what happened, but more of our insights, why it happened. Now we're in a world where we're moving from not just what happened and why it happened, what should you do? How should you fix it? And automation, that's where it comes in, In terms of our go-to-market strategy, one of the challenges, even though we have completely rebuilt our capabilities from ground up to be more native agents and agentic infrastructure, there is a lot of AI washing, let's say, going on in the industry. Every company out there will say, yeah, yeah, we are AI too. So one of the biggest challenges for brands has been, I don't know who to pick. There are 50 companies, they all say they do AI. And our recommendation, of them, a go-to-market strategy coming back to you is like, we believe in our technology, so we are offering free pilot, right? Hey, try it out at no risk to you. We'll work with you and have you actually solve business problems. With AI, it's almost AI is afterthought. Frankly, a brand doesn't care how you're solving. They want a problem to be solved, whether it is optimizing their retail media spend for the lowest CPC or whether it is fixing their content score. AI is a technology enabler. And we believe in our technology so much that we are using essentially a free pilot for people to try out the product, to see the value. And that's a big shift. In the past, would like, you know, is long sales process and but now we're saying hey don't even pay us just try it out you'll know for yourself is worth it what your money or not Brent Peterson (14:04.62) Yeah, think in the past year or now AI has become the requirement instead of the, hasn't, I think a year ago, even people were still bolting on AI to their product saying this is going to be new and exciting. And now if you don't have that as part of your product, people think you're behind the times. Sai Koppala (14:24.931) Yeah. So that's another reason people, whether they have anything or not, they'll slap an AI label on top of it. One of the interesting things for us, the learning experience for the last two years has been, it's building, know, collecting data and putting some LLM on top of it is relatively easy. The challenge always is LLM, think of it as a really powerful intern, lack of a better word, but what you really need is context. Brent Peterson (14:31.074) Right. Sai Koppala (14:55.883) And that's what, because we work with the world's top 20 global brands, we are able to glean in, and retailers too, right? So we are able to provide the LLM context around macro context like, hey, in the U.S. we're seeing due to inflationary pressures, people are willing to spend less on the same product. or they want a package with fewer items, right? Or it could be like, hey, Amazon dings you if you don't have three plus compliant images. Or if it's a category level in the pet food, it could be like, lately we're noticing that any protein claims, those are getting better visibility. Or there's a brand context, hey, this set of products are lower margin. So all this context is needed to make the right business decisions. the past, so what we would do in the past is somebody in the company or set of people in the company would know these things intuitively and make decisions. So what we have to do is train the LLMs across all these different contexts. That's how you get the right decisions. Otherwise, it'll just hallucinate, right? Brent Peterson (16:16.47) Yeah, absolutely. Sai, what is your prediction now for this year? Where do you think our economy is going? Where do you think retail is going in this next first half of the year now? Sai Koppala (16:28.675) think we're seeing a continuation of, I think people call it a K-shaped economy, where on one side, it's essentially a tale of two cities, right? On one side, significant part of the population is still strapped economically, right? So they are still hunting for low-priced, everyday value kind of goods. So therefore, there's significant pressure on brands to maintain price and not necessarily extremely price-sensitive buyers. And their focus there is maintain price and capture market share. On the other hand, you have a set of buyers, talk with the top 10 percentile, who are willing to pay a higher price for better value. Brent Peterson (17:08.142) Perfect. Oops, sorry, go ahead. Sorry. Sai Koppala (17:22.977) So they are less price sensitive and we're brands make appropriate adjustments given all the additional costs they've been having lately. So that's kind what we're seeing in terms of how price is impacting consumer behavior or vice versa. Brent Peterson (17:41.582) Perfect. Si, we have a few minutes left as I close out the podcast. I give everybody a chance to do a shameless plug about anything they like. What would you like to plug today? Sai Koppala (17:52.437) My advice to brands out there is AI is fundamentally changing how you can win in the retail business. And yes, there are a lot of solutions out there. My recommendation is pay close attention to the vendors out there. with a pilot, you will see the results. And from a commerce IQ perspective, If you're looking to optimize your content or retail media spend or identify what the sales gaps are and how they can be improved upon, Commerce IQ has a full-fledged both the agent suite plus human expertise to help you. Brent Peterson (18:40.238) Perfect, and tell us how did they get in touch with you? Sai Koppala (18:43.693) They can send me an email, si.k at commerceiq.ai or at info at commerceiq.ai. Brent Peterson (18:53.838) Perfect. Thank you, Sai. Sai Kopal is the CMO of Commerce IQ. Thank you so much for being here today. Sai Koppala (18:59.885) Thank you. Thanks, Brent. Appreciate it.