This week's episode

Jim Boudreau has studied SEO since 2001 — and even he had to ask ChatGPT what was actually in the box. In this episode, Jim explains why catalogue-scale SEO breaks the old playbook: once you've got hundreds or thousands of product pages, hand-crafted content stops being realistic, and copying the manufacturer's description starts costing you visibility in Google and in AI search. He walks through what actually moves the needle — product-level metadata, structured data, FAQs, and category pages built to answer the questions customers (and the AI) are actually asking. Jim is the founder of CataSEO, an AI tool that optimises ecommerce catalogues for search and generative AI engines, built in partnership with Moz.

Most ecommerce founders are handing serious decisions to AI that doesn't know what it doesn't know, and the results can be expensive. Marissa Rodriguez, founder of Through Experience and a fractional CMO and COO for a decade before that, joins Matt Edmundson to explain why industry-agnostic AI is dangerous for first-time founders. They cover the moment she watched ChatGPT invent a 20% advertising budget for a pre-revenue founder with zero basis, what it actually took to train an AI on her own frameworks (one module grew from six videos to fifty), and where she tells founders to point AI first. Her closing argument is simple. There are only three actions that matter online, click, subscribe, and buy, and everything else is noise.

AI is moving from a tool that suggests to a system that executes, and most ecommerce founders are not sure how to think about that shift. Cyril Golub, founder of Jinnify and a builder of ecommerce software since 2003, joins Matt Edmundson to work out what actually works. They cover the move from ecommerce to "AI commerce", why the harness around a model matters more than the model itself, how to match the model to the job rather than paying frontier prices for simple work, who really owns your data on a marketplace, and using AI to monitor competitors at scale. Cyril's closing line lands the whole episode: AI is a supervised assistant you give a clear goal, not a wizard.

Merchandising might be the most overlooked driver of conversion in ecommerce, and most stores are quietly losing sales and ad spend because of it. Leon Bart-Williams, co-founder of Akikumo and a veteran of NBCUniversal and Amazon, joins Matt Edmundson to explain what merchandising really is and where stores go wrong: out-of-stock products in prime collection slots, ads pointed at sold-out variants, and the 80/20 rule hiding at the variant level. They dig into why curation beats volume, how to test collections rather than launch on a hunch, and how to plan inventory around real sales velocity so cash is not left sitting on the shelf.

Drew Morgans, founder of the Amazon and marketplace agency Marknology, returns to the eCommerce Podcast for a grounded look at AI for ecommerce. He explains how he fed 15 years of his own work into an AI trained to think like him: roughly 500,000 emails, 300 podcast episodes, SOPs, playbooks and won-and-lost proposals. Matt and Drew get into why context is the real problem, the rules layer he calls Drew IQ, running cheap local models for ingest and expensive models only for client-facing work, why one super-agent beats a crowd of bots, and how it all lets him scale the deep work an agency normally cannot afford. Concrete, contrarian and genuinely useful.

Sohun Sanka, founder of Clankers, has audited more than 2,000 TikTok Shop affiliate programmes and seen the backend of over 150 agencies. He joins Matt to explain how the best brands actually win on TikTok Shop, and where AI genuinely earns its keep. The big unlock is showing AI what good looks like before you ask it for anything. From testing your lighting rather than just your copy, to treating affiliates like a sales team the way Javy Coffee grew from 300k to 1m dollars a month, this is a practical, honest guide to getting started with AI on TikTok Shop.

When someone asks ChatGPT for the best hiking boots, it names brands, not links, about 97% of the time. Search veterans Dave McAnally and Alan Osetek explain how AI is compressing the ecommerce funnel, why the answer often comes from YouTube, Reddit and third-party review sites rather than your own website, and how a 6,000-subscriber creator ended up answering for Adidas. Inside, a practical checklist covering FAQ blocks, schema, video and niche-source relationships, how to measure AI traffic in GA4, what's coming with OpenAI ads, and why small brands can still beat the giants by owning the questions no one else is answering.

Everyone suspects email is dead. Laryssa Wirstiuk, who has run email and SMS for a decade, says the channel is fine, it's the way we use it that's broken. In this episode she signs up to 360 brands and finds 44% never send the welcome offer they promised, explains why cutting a list from 20,000 to 1,500 saved a business, and shows why the emails people are actually waiting for aren't the ones you're focused on. Practical, honest and useful whether you send weekly or three times a day.

Gabriel Lindh has spent two years and a hundred weekly newsletters building Nordic Dudes, an anti-chafing swim shorts brand, and still hasn't sold a pair. He and Matt Edmundson unpack the pre-launch playbook: validating demand with paid pre-orders before spending a penny, building an audience while the product is still in development, launching at good enough rather than chasing perfect, and the small packaging decisions that turn a purchase into a moment. An honest look at doing the early, unglamorous work of starting an ecommerce business properly.

Brenden Delarua spent eleven years hitting every ROAS target he was set, until a CFO asked why record ad numbers weren't showing up in the bank account. He explains why platform ROAS measures correlation not profit, the free version any small brand can start with (multi-touch attribution plus post-purchase surveys), the YouTube campaign that looked like a 0.2 ROAS disaster but was secretly 3x profitable, and the one post-purchase survey question worth more than the rest.

Cem Atik took a client from $15m to $50m. Then they paid an outside firm 15k to tell them what was going wrong, and his reply was that they already had 100,000 customers who would have told them for free. Cem Atik, Co-Founder of Harucon Ventures in Dusseldorf, joins Matt Edmundson on The eCommerce Podcast to talk about the two mistakes that stall ecommerce growth between $2m and $10m, why a rising customer acquisition cost is not automatically bad news, why retention rather than paid marketing is where the profit actually sits, and the five metrics every operator should be able to recite from memory. He also explains the packaging change that took 25% out of one brand's costs, and the $250,000 he burned on a second business after his first one hit $7m.

After seven years and 300 episodes, Matt's most useful lesson isn't a tactic. It's that "will it move the needle?" says yes to everything worth writing down, which is why notebooks fill up and nothing ships. This episode is the honest report on a 90-day Instagram experiment he committed to on air and didn't deliver, and the framework he should have applied to it. Two questions, a drop-delegate-systematise diagnostic for when an idea fails, and the one-thing rule for what survives. Plus the podcast conversation that took mobile conversion up over 400%, and what 300 episodes have actually been worth.