Marketecture: Get Smart. Fast.
00:04:05 · Ari Paparo, Sreenath Reddy
[00:04:05] S0: Right. Right. So you might be looking at things like your inventory and your margin and, you know, where you're not selling as much as you'd like and what the top sellers are and all those sort of things. So it's a mixture of availability, advertising stuff. And then your system, Intentwise, is... Can be used for analytics, but it also could be used to actually run new ads or to change settings and things like that. [00:04:25] S1: Correct. And in fact, like, they go hand in hand. Right? So our observation is, like, 70 to 75% of a ad manager's time actually goes into analysis. Yeah. Right. That's a lot. Right. Yeah. They can't really be so especially in a marketplace world. Right? It looks like you have experience from like the Google and Meta world. The marketplace world is a little different, right? [00:04:50] S0: Right. [00:04:51] S1: Where ads are influenced by so many other factors like content, pricing, do you own the buy box or not, inventory, so on and so forth. And that leads to a lot more time being spent on analyzing what's going on because there's a lot more drivers. So to us, edge and analytics are not really separate, but they evolve as two separate stacks pre Gen AI, but Gen AI is converging them together which is super exciting by the way. Yeah. I bet. So so yes, advertising and analytics
00:04:51 · Sreenath Reddy, Ari Paparo
[00:04:51] S1: Where ads are influenced by so many other factors like content, pricing, do you own the buy box or not, inventory, so on and so forth. And that leads to a lot more time being spent on analyzing what's going on because there's a lot more drivers. So to us, edge and analytics are not really separate, but they evolve as two separate stacks pre Gen AI, but Gen AI is converging them together which is super exciting by the way. Yeah. I bet. So so yes, advertising and analytics [00:05:24] S0: combined. So yeah, so let's talk about AI because it's obvious use case for AI. And, also, you'd recently done some case studies and integrations with the Amazon MCP server that we wanted to talk about. So take me through what are your clients? What did Amazon announce? What did you... What was your integration like? And then what are clients doing with this stuff? [00:05:43] S1: Yeah. So I think there's a few different things. Right? So when it comes to ad management, obviously, Amazon has sponsored ads, which is ad search for open search and then display advertising. Right? So Amazon opened up MCP capabilities several months back around just ad management, creating campaigns [00:06:02] S0: Right. [00:06:03] S1: Optimizing campaigns, so on and so forth. We are fairly tightly integrated into that. And so we went from being a read only MCP where people could analyze, build dashboards, and maybe automate some workflows to, oh, now I can take action too. Right? [00:06:17] S0: Right. Right.
00:06:17 · Ari Paparo, Sreenath Reddy
[00:06:17] S0: Right. Right. [00:06:17] S1: So the the use case I just talked about earlier, which is, like, classic because inventory management is often done separately by maybe a separate team and a separate function from the one that runs ads, oftentimes they don't talk to each other. And imagine having and a as the catalogs get large that problem gets more complex. But now you can go ask the analytical question, hey, what are my products where I don't have enough weeks of cover? And by the way, where am I spending ad dollars? And then I can turn around and say, go for the ads on those products. Right there. Yeah. That's an example. That's an example. So that's that's one part. Right? The other part is Amazon launched what's called Amazon Marketing Cloud. Are you familiar with what it is Ari? [00:07:01] S0: Yeah. I think I I think they talked at our last conference but yeah. Okay. [00:07:05] S1: So I mean, you know, you know, for for... If folks are listening to this and don't know what it is, I'll just give you a quick Sure. Fifth... Fifteen second primer, which is if you think about bra... Of a brand and your exposure to shoppers, all the way from an impression to click to conversion and everything in between like add to carts, engagement with the reviews or gift list, every one of those events is now captured within Amazon Marketing Cloud at an individual shopper level. So it's clean room. [00:07:34] S0: Right,
00:07:34 · Ari Paparo, Sreenath Reddy
[00:07:34] S0: Right, [00:07:34] S1: right. So what that makes possible which is not possible otherwise is a deep understanding of shopper journeys including things like lifetime value, repeat purchase behavior, and this ability to create really fine grained fine tuned audiences for your brand. So deep shopper insights and the ability to create audience, that's what AMC is. Yeah, so the other announcement was and historically because it's a clean room, there are some pre built insights and audiences, but to get very specific around use cases that are unique to your brand, you have to know SQL. Have to write SQL statements. [00:08:10] S0: Yeah. Yeah. Sure. And I never... Since AI came around, I I said I will never write SQL again. [00:08:16] S1: But it takes some time for natural language conversion to SQL to be accurate. Right? Like Okay. Yeah. So I think that's the other announcement is around Amazon Marketing Cloud they launched in beta MCP around that too where I can get new insights as well as create audiences just using natural language. I don't need to be a SQL guru to do that. [00:08:35] S0: So do your clients know when they're using, say, the Marketing Cloud versus some of the other APIs, or do they just kind of put in a query like, tell me who's most interested in my product and and you figure it out? [00:08:48] S1: It's a spectrum. Right? Yeah. There are a small minority of clients that know each of the systems very well. Yeah. Okay. Right? [00:08:57] S0: Small minority, yeah.
00:08:57 · Ari Paparo, Sreenath Reddy
[00:08:57] S0: Small minority, yeah. [00:08:58] S1: Yeah but a large group doesn't maybe fully understand what is possible with AMC. Yeah. So they're a bit agnostic about where the signal comes from. Right? So we think about it in six buckets by the way. There's advertising. There's retail and operations. There is shop... Shopper intelligence from Amazon Marketing Cloud. There is cross channel data. I mean, one of the common questions these days is how is TikTok influencing my Amazon performance? That's cross channel [00:09:25] S0: use case. Right. [00:09:26] S1: Yeah. Can you get an answer to that in your in your analytics? Well, I think what we... Because we connect to both platforms, we can certainly, a, we can do the basics, which is correlated use of what's going on with some time. That's one. You're also taking an algorithmic path to understanding causality. But the key there is to have the underlying signals stitched together the right way so you can actually do these things, which is what [00:09:51] S0: we solve for. And so Amazon's offering connections to all their services. It's complicated. It's hard for many marketers to deal with it. Intentwise is aggregating it all, offering the AI gateway, and also connecting across retailers so... Or across media plans like TikTok, whatever. What are what are some success stories you've seen of your clients so far using using this stuff? They're going to Claude. What sort of things are they asking?
00:09:51 · Ari Paparo, Sreenath Reddy
[00:09:51] S0: we solve for. And so Amazon's offering connections to all their services. It's complicated. It's hard for many marketers to deal with it. Intentwise is aggregating it all, offering the AI gateway, and also connecting across retailers so... Or across media plans like TikTok, whatever. What are what are some success stories you've seen of your clients so far using using this stuff? They're going to Claude. What sort of things are they asking? [00:10:16] S1: I think the frame that we have settled on after watching a lot of their use cases is, again, back to four buckets, strategy and planning questions. Yeah. Right? Execution, simplifying the execution. Performance monitoring, is a big deal, and then troubleshooting. So one of the most common... Look. When you have a large catalog, there's always some... Something going on. Right? [00:10:38] S0: Yeah. Sure. [00:10:39] S1: Right? So, hey. Automatically knowing there's a sales anomaly here, either good or bad Yeah. And then understanding why is that happening. [00:10:48] S0: Right. [00:10:48] S1: Right? Right. Is it a Amazon drop price because it matched some price drop somewhere else and now it's not making enough money so I'm gonna... They can no longer can advertise. Yeah. Or or is it that prices drop, conversion has gone up and it's doing well? So identifying anomalies and understanding root causes is one of the most common use cases. [00:11:10] S0: Yeah. And how much of that is automated versus the versus the operator having to ask the questions?