A D2C brand in 2026 has three paths to sell online: build a custom storefront they own, sell through marketplaces like Amazon and Walmart, or run both. Each path has a different margin profile, a different customer acquisition cost, and a different level of control over the brand experience. The choice isn't about which platform is "better" — it's about which path matches the brand's margin structure and growth stage.
What does the margin structure look like for each path?
Marketplace-only. Amazon takes 15–45% of the sale price depending on category (referral fee + FBA fees + advertising). The brand gets volume but gives up margin and customer data. At $10M revenue on Amazon, the brand keeps $5.5–$8.5M. The marketplace owns the customer relationship — the brand can't email them, can't retarget them, and can't build a subscriber base from those transactions.
Custom storefront-only. The brand pays for hosting ($200–$2,000/month), payment processing (2.9% + $0.30 per transaction), and customer acquisition (paid ads, content, SEO). No marketplace fees. At $10M revenue, the brand keeps $9.4–$9.7M before marketing spend. The margin is dramatically better, but the traffic isn't free — customer acquisition cost (CAC) on a standalone storefront runs $25–$80 for most D2C categories in 2026.
Hybrid. The brand uses marketplace for discovery and volume while building a custom storefront for margin and retention. Amazon generates awareness, the custom storefront captures repeat purchases through subscriptions, bundles, and loyalty programs. This is where most D2C brands with $5M+ revenue land in 2026 — the margin from owned-channel repeat purchases funds the CAC for marketplace-driven acquisition.
When does a custom storefront make sense?
A custom storefront — meaning a platform built for the brand's specific needs, not a Shopify theme — makes sense when the brand has at least two of these conditions: repeat purchase rate above 30% (the unit economics of owned-channel acquisition work because customers come back), average order value above $80 (the margin per transaction justifies the platform investment), or a product that requires a buying experience marketplaces can't deliver (customization, configuration, subscription, or a complex sizing/fitting flow).
The custom storefront isn't about design. It's about business logic. A brand that sells customisable furniture needs a 3D configurator with real-time pricing that updates as the customer changes materials, dimensions, and finishes. A supplement brand needs a subscription engine that handles swap-and-save, pause, skip, and build-a-box. A fashion brand needs a size-matching algorithm that reduces returns from 35% to 12%. None of this runs on a template.
When should a brand stay marketplace-only?
Marketplace-only works when the product is a commodity, repeat purchase rate is low, and the brand can't differentiate on experience. A brand selling phone chargers doesn't need a custom storefront. The buying decision is price and rating, not brand experience. The marketplace's built-in traffic is worth more than the margin it takes.
The risk of marketplace-only at scale is dependency. Amazon can change its fee structure, algorithm, or category rules at any time. A brand doing $5M on Amazon with no owned channel has zero leverage and zero fallback. The brands that survived Amazon's 2024–2025 fee increases were the ones that had built an owned channel before they needed it.
How does the hybrid model work in practice?
The hybrid model treats marketplace and owned storefront as different parts of the funnel, not competitors. Marketplace handles discovery and first purchase. The owned storefront handles retention and high-margin repeat purchases.
The technical challenge is inventory synchronisation. The custom storefront and marketplace must share a single inventory pool to prevent overselling. Orders from both channels flow into the same fulfilment system. Pricing may differ (marketplace pricing accounts for fees; storefront pricing can offer direct-to-consumer savings or subscription discounts). The system needs real-time sync — not hourly batch updates that let inventory drift between channels.
The customer data architecture matters. Every marketplace order is a missed data point — the brand knows the order happened but doesn't know who the customer is. The owned storefront captures email, purchase history, preferences, and behaviour data. The hybrid model works when the brand actively converts marketplace buyers to owned-channel buyers — through product inserts, post-purchase email sequences, and subscription offers that only work on the owned storefront.
What does a custom D2C platform cost to build?
A headless custom eCommerce platform for D2C runs $80,000–$150,000 for the core build: product catalog, checkout, payment processing, subscription engine, and marketplace integration. Add $20,000–$50,000 for custom features like configurators, loyalty programs, or recommendation engines. Ongoing maintenance and hosting runs $2,000–$5,000 per month.
The ROI calculation is straightforward: if the brand does $5M in revenue, moving 30% of that from marketplace to owned channel saves $225,000–$675,000 in marketplace fees annually. The platform pays for itself in 6–12 months from fee savings alone — before counting the revenue from subscription models, loyalty programs, and improved retention that the owned channel enables.
The platform choice isn't permanent. Start where the brand is. But plan the architecture so the platform can grow with the business — not become the ceiling that limits it.
Building something complex?
Start a project with Madgeek