Selling products online in 2026 rarely means choosing just one place to sell. Your own store — a direct-to-consumer site on a platform like Shopify — gives you full control of the brand experience, your pricing, and, crucially, the customer's email address, which is what makes Module 6's retention flows possible. A marketplace such as Amazon gives you something your store can't: millions of shoppers already searching with their wallets out, plus fast fulfillment, in exchange for a referral fee on every sale (15% in most Amazon categories in the US) and very little access to who your buyers are. Social commerce sits between the two: shops built into social platforms turn discovery content and creator partnerships into sales. On TikTok Shop the buyer can check out without leaving the app, while Instagram and Facebook now send shoppers to your own site to finish the purchase — and what's available varies by country, so check each platform where you sell.

The right answer is usually hybrid, but the order matters. Compare channels on contribution margin per order, not revenue: a channel that doubles your sales at a thinner margin can leave you with less profit and fewer customers you can ever contact again. This lesson's scorecard makes that trade-off explicit before you commit inventory or ad budget to any channel.

A product detail page (PDP) has one job: answer every question a buyer has before they'll pay, in the order they have them. That means a title that says what the product is in the words people actually search for, images that show it in use and at real size, benefit-led bullets rather than a wall of marketing copy, and no surprises on price, shipping, or returns — among the most common reasons shoppers abandon at the last step. Reviews do heavy lifting here, which is exactly why fake reviews, reviews paid for on condition they're positive (or negative), and undisclosed insider reviews are now prohibited in the US under the FTC's 2024 rule on reviews and testimonials: collect real ones and show them honestly.

The less visible half of the job is product data. The same facts on your page also need to live in structured form — a product feed in Google Merchant Center (its required fields include an ID, title, description, link, image link, price, and availability, plus brand and a GTIN for most new products), schema.org Product markup on the page itself, and the listing fields on each marketplace. That data is what places your products in Google's free listings and Shopping ads, and increasingly what AI shopping assistants read when they answer a question like "what's the best coffee for pour-over?" Apply Module 2's CRO thinking to the page and search thinking (Module 3) to the data.

Shopping ads work differently from the search campaigns in Module 7: you don't write keywords or ad text — you supply a product feed, and Google matches your products to searches using that data, which is why Lesson 9.2's feed work is also ad work. Performance Max campaigns use the same feed across Google's networks with automated, AI-driven bidding. Marketplaces run their own ad systems on top of the listings you already have: on Amazon, Sponsored Products ads promote individual listings in search results, Sponsored Brands ads feature your brand and several products, and Sponsored Display ads reach shoppers again after they've browsed, on and off Amazon.

Marketplace sellers usually measure ads with ACoS (advertising cost of sales: ad spend ÷ ad-attributed sales), which is simply ROAS turned upside down — a 4x ROAS is a 25% ACoS. Neither tells you whether you made money. For that you need contribution margin: what's left of the price after product cost, fees, fulfillment, and returns. Your break-even ACoS equals your contribution margin as a percentage of price; spend beyond it and every ad-driven order loses money. Set your target below break-even to leave a profit, and watch TACoS (ad spend ÷ total sales, including organic) to see whether ads are building sustainable sales or just buying them. Automated bidding is only as smart as the target you give it.

E-commerce is one of the places AI pays off quickest, because stores often have hundreds of products and a lot of repetitive content. AI can draft product descriptions, titles, and image alt text in minutes, suggest which products to recommend alongside each other, and power on-site search that understands what shoppers mean rather than just what they type. The risk is just as specific: AI will confidently invent a spec, an ingredient, or a sustainability claim, and on a product page an invented claim isn't merely embarrassing — it's a misdescription that drives returns, complaints, and potential legal trouble. Every AI-drafted product claim gets checked against the real product before it's published.

AI also changes how shoppers find you. AI shopping assistants and AI answers increasingly recommend products by reading structured product data and reviews, so the feed work from Lesson 9.2 is how you show up there too — the product-side version of optimizing web pages for AI answers (Lesson 3.3). After the sale, the biggest profit lever is the second order: post-purchase and abandoned-cart flows, triggered by what each shopper does (how to build them is in Lesson 6.3), bring customers back without paying to acquire them again. Track them with GA4's e-commerce events — view_item, add_to_cart, begin_checkout, and purchase — and report them on a dashboard (Module 8 shows how to build one) so you can see where buyers drop out and whether your flows win them back.