Latest / Talk Commerce / Transforming Enterprise Web Performance Through Edge Computing and AI with Drew Chambers
Transcript
- Brent Peterson (00:01.86) Welcome to this episode of Talk Commerce. Today I have Drew Chambers. He is with Harper and they are a marketing company. But Drew, why don't you go ahead, do an introduction for yourself. Tell us your day to day role and one of the things you're excited about for this coming year in business. Drew (00:19.832) Sure, thanks for having me, Brent. Yeah, so I'm the EVP of marketing, so essentially CMO at Harper. We are a platform technology company. we like to say that we're a platform as a service for large enterprise customers that are looking to drive better application and web performance. So think technical SEO, think conversion rates on your website, think better user experience, think scaled applications and large catalogs and distribution globally. All of those things are sort of the sweet spot for Harper. And I guess for this year, I'm excited to see how we continue to see e-commerce grow. Obviously, there's some geopolitical challenges that might make things interesting for the rest of 2025. But we do continue to believe that the trend will be towards online. and continuing to see growth and interest in that space and particularly for the large enterprise customers that we work with, I would anticipate seeing a lot of emphasis on performance and how that can continue to improve. Brent Peterson (01:33.606) How about passions outside of work? What do you like to do? Drew (01:37.87) Yeah, I tend to be very active. So I live in Denver, Colorado. So it's all the standard stuff of skiing and cycling. I also like to row and box. I have a young daughter and she enjoys dancing and soccer. So it's been a lot of watching that stuff as well. So that's been a lot of my life as a player. Brent Peterson (02:02.36) That's awesome. right, Drew, before we get started, we talk about some content and maybe talk a little bit about platforms and performance, which I'm always interested in. I'm going to tell you a joke and all you have to do is tell me or give me a rating eight through 13. I call it the free joke project. So here we go. a job interview, I filled my glass with water until it was overflowing. Nervous, asked the interviewer. No, I always give 110 percent. Drew (02:31.374) I'll say 12. That's pretty good. Brent Peterson (02:35.014) All right, thanks. Appreciate that, Ju. I would have given it a lower score, but I think that we've got to get some performance marketing into there, right? Drew (02:42.99) going to say, well, you know, there's categories of jokes, right? And like, as far as, you know, dad corporate jokes go, that's pretty good. Brent Peterson (02:49.414) All right, perfect. So I'm always interested in this topic. And I've been writing a lot about SAS first pass first on-prem. And there's some things you can control and some things you can't control for performance, right, and even an on-page SEO. So talk about the importance of how fast your page loads or. you know, how important that is in terms of your ROI, things like that. Drew (03:20.802) Yeah, I think it's, I think most owners of stores, most marketers, most brands understand that speed is key, right? You have a very finite window of attention. You're trying to mimic, you know, an in real life experience as much as possible in the digital space. And so having something that is extremely performant and responsive to a customer is, is, is key. But beyond that, You know, from a technical SEO standpoint, we know that Google and all the major search engines and now, know, chat, GBT and the AI search elements are going to prioritize things like core web vitals, right? So they could, you know, largest content full paint and, you know, layout shift and interaction, the next pain, you know, all of those pieces are really sort of table stakes right now. And so the. the more complex that you're trying to make your technology stack, inevitably you're adding latency to the equation. You're adding elements that need to be rendered and need to be loaded depending on where they're loading from, whether that's the edge or whether that's from on-prem and origin. That can create an incredibly laggy experience for your customers. And not only is that going to hurt customers, know Amazon has said that every 100 milliseconds of latency is about 1 % of total sales for them. So that's a really back of the envelope math that you can do for your own company. But beyond that, you're also just going to get fanged on your organic stuff because you're going to lose the box on shoppable tiles in Google. You're going to lose know, positioning to your rivals that have prioritized this stuff. So it really is like paramount across all elements of what you would think is the digital experience. Brent Peterson (05:24.982) We talked a little bit AI in the beginning, but how is AI now playing a role in measuring and helping you with that performance on the page loads? Drew (05:38.51) Well, I don't know that, and this is kind of an interesting thing because I don't know that AI is necessarily helping a lot of people on their page loads. I do think that prioritizing all the same things that you would have traditionally prioritized for the Googlebot are important. I guess the term now is a GEO standpoint, right? There was SEO, now there's GEO for the large language models. But really, again, prioritizing these elements that are going to create, one, a better mobile first experience, I should say. That's another unspoken thing here is that everything is being prioritized towards a mobile experience because that's where most people are going to be interacting with your ads. That's going to be where most people are coming to your website from on their phones. And So I think you need to think about that. And then you need to think about like, OK, is the information that I'm rendering, is that going to be valuable from an AI's perspective? Is that going to answer the question that I think that people are asking? Again, very similar to how you were probably thinking about things five to 10 years ago with regards to your blog content and all those pieces that you were putting together for Google. It's just that on steroids for. for AI and it's about making every landing page, the PDPs to the blog, to the content that responsive and that interesting. Brent Peterson (07:13.594) Yeah, sorry, I should have been more clear on that question. I think people see AI as generative still, but they miss the fact that AI is really good at finding patterns and problems within your code base and within your tech stack and even within your catalog. And lot of people don't see the advantage of using it to find those SEO gaps. that either are bloating your system. one of the marketing things that we always saw when I used to run a digital agency was the site would suddenly load in less than one second. Then a couple of weeks later, the site loads in three minutes because I'm being facetious, but three, let's say 10 seconds, because the marketing team, no offense against the marketing teams, but they've loaded a video or some file that's like a megabyte or 10 megs or whatever that number is. where they've just completely loaded down the system because they can, right? So are there tools that are coming out that help marketing teams optimize that through analytics? Drew (08:14.95) for sure. Yeah, I think there's a number of partners that we work with that can go to as granular detail as, it's this JavaScript that's essentially bloating your LCP, and it's this video, or it's this image, or it's this element of the page. I think more and more we're seeing options like Webflow and some of the CMS tools that people are using integrating these types of AI to... you know, leverage Gemini and ChatGPT and whatnot to help you identify that because yeah, we've gotten to the point where you can drag and drop and build whatever the heck you want, but you're not necessarily doing so with, through the lens of like, is this the most performant solution? And I think where, you know, where we sit typically at the very high enterprise level of things where they're using, you know, 15 different pieces of technology and what's typically. considered like a composable, a headless or a composable architecture for their e-commerce sites. There's all sorts of things that just don't work together all that well. And they weren't intended to work together. And they might be the best at the thing that they do, but they're, again, not communicating as well as they could be. so AI can be very helpful in identifying what we would consider latent pain points within your architecture. that can be potentially done more efficiently. Brent Peterson (09:46.96) Do you think that this trend towards AI is going to help? I mean, I shouldn't even say that. It will help marketers, but maybe marketers need a little more education outside of the generative AI zone. Drew (10:01.966) Yeah, I think marketers have been using AI to help with the copy and the graphics for a bit now, but I think there's going to be a whole other element similar to engineers leveraging AI to write better code and more efficient code and leveraging the technologies that are really building themselves as AI first. Marketing is going to have the same thing. And again, I know You know, Meta and Google and, and, Tik Tok have started to roll out some, more AI tools within their platforms, which I think is getting marketers a little bit more comfortable with how AI slots in with, with their typical daily workflow. But it's, it's going to be a, it's going to be a steep learning curve to be perfectly honest, because the implications aren't just on the front end. It's not just on what you're necessarily inputting into the system. I think as many of us know, the way that Meta is using AI on the backend to identify learnings and to push those back into your campaigns has kind of shifted the entire way that you need to think about things, right? They've prioritized creative as being the new targeting. And so they want you to go abroad and they want you to not do all of this audience targeting and instead let the creative work and then using AI to analyze that creative. And in so doing, I think a lot of marketers and lot of brands realize that all of their creative is the same. in the eyes of an AI, it's not all that different. There isn't, you know, They haven't explored these ranges and as such they're behind the eight ball. They're losing out on reach, they're losing out on potential new customers. And so I think that's a pretty steep learning curve that people are gonna have to educate themselves on and better understand as they look to be at the forefront of paid media, earned media and organic. Brent Peterson (12:10.022) You mentioned Composable earlier and I wanted to just kind of talk about the enterprise level. You hear a lot about Composable, Headless, but I read an really interesting post this morning from a friend of mine. He says, Composable commerce, it's not a religion. Headless is not a business model, it's an architectural choice. Composable is not a competitive advantage, it's a technical option. And then he goes on to state that The projects blow up and cost the MVP takes a year to ship. The ownership is split between 10 microservices, three agencies and a part-time DevOps. Do you think that sometimes the, like they hear the buzzword from the salesperson because they're an enterprise company and they said, we have to do this. Drew (13:00.334) It's a really good question. And I think everything that your friend said makes sense. think in the typical execution and what engineers might talk about as the traditional rip and replace, or we're going to take everything that we have and we're going to rip it out, we're going to replace it with something different, that is a very challenging undertaking. It's a lot of. getting buy-in from multiple teams, right? At the enterprise level, you probably have, you know, it's not just one marketing team. It's, have a marketing team, you have a product marketing team, you have a brand team, you have your paid team, have your organic team, and then you have an entire development organization, right? Which is maybe split between web performance, technical SEO, catalog, you know, what have you. It could be 15 different ways. And so all of those people have like an invested stake in in this project working. Because as we talked about at the beginning, having an architecture that is inherently performing is going to rise. It's a rising tide that can lift all those ships, right? So that right there presents a challenge, because that right there is going to blow the length of time that any project takes. Beyond that, Yes, cost is always going to be a factor here, right? And what you're trying to do is you're trying to get the best solution at every single one of these points, right? You're trying to get the best database. You're trying to get the best hosting architecture. You're trying to get the best API layer and caching layer and messaging and all those pieces. But again, if those things aren't designed to work well together, you're going to see bloated costs. You're going to see all these sorts of interactions and connections that you weren't anticipating originally. I won't go into my shameless plug here, but looking at that project holistically and really trying to dissect what is the KPI that we really would like to improve in the next. Drew (15:14.732) three months, what is a thing that we would consider a success? I think is the strategy that I would recommend as opposed to wholesale saying like, hey, we want a composable architecture tomorrow, we want a new composable architecture tomorrow, we're gonna go headless tomorrow because you're just gonna encounter a lot of different obstacles and a lot of different hurdles. Whereas if you can prioritize like, this is the thing that I wanna see improvement on, again, landing page speed, LCP. my technical SEO ranking and search engine rankings, you can maybe get a little bit more surgical with what exactly you're going to see the most ROI from within a composable architecture. And you might be able to peel off certain use cases and test them out at a lower scale before you're flipping your entire domain over to a new structure the week before Black Friday's ever. Brent Peterson (16:11.974) And of course you would never recommend anybody flipping over to new structure a week before Black Friday. Drew (16:17.23) No, I've seen it happen, unfortunately, but no, I would not recommend that. Brent Peterson (16:23.366) Last topic, do want to kind of talk about the idea of this Edge. You had mentioned Edge and how important that is and maybe today's architecture. And a lot of companies can achieve it with a SaaS or a PaaS or an on-prem, right? Just maybe give us a, just explain to the listeners what the Edge architecture means for delivering your website. Drew (16:45.71) Yeah, I mean, I think the simplest way, and I I preface every one of these conversations with, am the least technical person at my company. So that is, which is good for a marketer, think. But I think the easiest way to think about, you know, the advantage of the edge, right, is that you are essentially creating the most personalized experience and the quickest experience close to where that person is, right? So rather than, you know, me being in Denver, having to make a call to, let's say, Origin in Europe for a company, I maybe am able to hit a pop or a node, whatever you want to say, much closer to me. I'm having a very, it's a much quicker user experience with the website. Theoretically, I can start adding some personalized elements there. So depending on what your product is and depending on what you sell, Maybe I'm seeing something slightly different than somebody that is in Asia or Europe or Australia or whatever it is. So you can really start to leverage that to your advantage to not only have a more performant experience and a faster experience for the user, but also potentially that personalization that everybody is hunting for. As marketers, we have all heard that we just need to have as personalized as custom and experience as possible for our potential customers, right? They need to see an ad that speaks directly to them, that solves a very specific problem or paints this wonderful story, this picture for them of an outcome. And then they land on the landing page that speaks directly to that ad that they just saw and like everything is this like contained. ecosystem of essentially walking into my favorite store and having the salesperson know my name and hand me something that I know I'm going to love. And the Edge makes that a lot more feasible for you as the e-comm brand. Brent Peterson (18:52.442) Yeah, that was perfect. That was a perfect explanation from a non-technical person. it's great. Drew, you know, we have a few minutes left as we close out the podcast to give everybody a chance to do a shameless plug about anything they like. What would you like to plug today? Drew (18:55.414) Alright. Drew (19:06.83) Well, I'd love to plug Harper, obviously. It's very, very excited to be a part of this team. And a lot of what we've discussed today is the challenges that Harper helps enterprise customers face. So as I mentioned at the beginning, we are an enterprise platform as a service that is focused on performance, specifically web and application performance. So you can think about building some really cool componentry on top of Harper. It is a fully integrated back end stack. So this is really the big difference between us and sort of the architecture that I've described today, where you would have a traditional database, a cache layer, a messaging layer, an application layer. All of that is fused into one in memory process. And so it's just naturally faster. Everything is happening concurrently. You're removing serialization, deserialization problems, cross-data center replication. And as a result, you can leverage a lot of what we've talked about today in terms of better customization at the edge, know, LCPs that are like 180 milliseconds. So just the fastest possible experience you could have for a user. And I think what's really exciting to me and is relevant to this discussion and If I have a second here, I'll just say part of the challenge that we see with AI is that a great use of AI would be to customize every landing page to the user that lands on your page. I know what Drew likes. I'm going to make this landing page speak directly to him. Wear this shirt to this cool event that you're going to. But we can't do that because as we've all used ChatGPT, it takes a little while for ChatGPT to spit out that perfect answer. And nobody is going to wait for five to 10 seconds for your page to load. So how do we get AI onto the landing page, onto the PDPs? And I think what we've been talking about is by leveraging AI with Harper, you can create a landing page experience that's essentially pre-rendered and is as fast, if not faster than a traditional landing page experience, but is using AI to generate multiple pages for various personas. And that's really exciting to me. think that's going to be table stakes in the next five years for Drew (21:29.656) for most brands, if not the top players in this space. So that's the shameless plug. I appreciate you giving me the time to do it, but it's also very exciting. think it's very relevant to what we're talking about. Brent Peterson (21:41.73) Andrew, tell us how we get in touch with you if somebody wants to contact you. Drew (21:45.422) Yeah, mean, feel free to email me. My email is drew at harperdb.io. Or you can go to our website, which is harpersystems.dev. And yeah, happy to walk through anything with anybody there. also, we have open source componentry that you can go take a look at on GitHub. So play around, give us a ring, let us know what you think. Brent Peterson (22:13.51) That's perfect. Drew Chambers, the EVP of Marketing for Harper. Thank you so much for being here today. Drew (22:18.722) Yeah, thanks so much for having me, Brian.