By Peddlum Admin
Published: 2026-05-07
Most marketplace literature treats "two-sided" as the natural shape of the system. After building one, I think it's a local maximum — and the global maximum has three sides, not two.
Most marketplace literature treats "two-sided" as the natural shape of the system. Sellers on one side, buyers on the other, the platform sits in the middle and takes a cut. This framing has been so dominant for so long that we've stopped noticing it's a choice.
I've been building a digital product marketplace for the last several months, and the more I sat with the two-sided model, the more it felt like the wrong abstraction for what's actually happening in commerce in 2026. This post is an attempt to articulate why — and what changes when you build for three sides instead of two.
Here's a question worth taking seriously: when someone buys a digital product in 2026, what's the actual sequence of events leading up to the purchase?
It almost never starts with a marketplace search. It usually goes:
They watch a TikTok, a YouTube short, or scroll past an Instagram reel.
A creator is using or recommending a product.
They search for that product (often on the marketplace, sometimes on Google).
They buy.
The creator is in that path. They're doing the discovery work. They're doing the convincing. They're the reason the sale happens.
But on most marketplaces, that creator is invisible to the system. The marketplace doesn't know they exist. The seller doesn't know who drove the sale. The creator doesn't get paid by anyone — unless they cobbled together an external affiliate program on a separate platform with separate attribution and separate payouts.
This is the path mismatch. The actual customer journey runs through three parties, but the marketplace data model only contains two.
Two-sided marketplaces are good at a specific thing: matching supply and demand within a closed system. eBay matched buyers to sellers of physical goods. Airbnb matched travelers to hosts. Uber matched riders to drivers.
These work because the platform owns the discovery layer. You search on Airbnb, you see Airbnb listings, you book on Airbnb. Discovery happens inside the system.
For digital products in 2026, discovery happens outside the system. You don't search Gumroad for Notion templates. You see a creator using one on TikTok, then go find it. The marketplace is the transaction venue, not the discovery venue.
This is a profound shift, and most marketplaces haven't internalized it. They still optimize their entire experience for in-system search and recommendation, when the actual leverage point is the creator who drove you there.
The standard response to this is to bolt on an affiliate program. Sellers can offer a 20% cut to anyone who refers a sale. Most marketplaces have something like this.
I think this is fundamentally a bad solution, for three reasons.
Affiliate programs assume creators will find products. They don't. The flow is wrong. Creators don't have time to browse marketplaces looking for things to promote. They post about what's relevant to their audience, and incidentally affiliate-link it if it happens to be supported.
Affiliate programs treat creators as a long-tail acquisition channel. This is dismissive. The right framing is that creators are a first-class participant in the transaction, not an extension of the seller's marketing budget.
Affiliate programs have terrible attribution. Last-click attribution gives most credit to the search engine. First-touch attribution gives most credit to the creator but is rarely supported. Multi-touch is impossible without identity. The result is creators getting underpaid for sales they actually drove.
A real fix isn't a better affiliate program. It's a marketplace where creators are structurally recognized.
What if instead of "sellers list, buyers buy, creators are a marketing afterthought," the marketplace was built around three first-class roles?
Sellers list products. They post campaigns asking for content. They approve creators who fit.
Creators browse open campaigns. They apply to ones in their niche. They make content. They earn a fixed task fee plus commissions on sales.
Buyers discover products through creators. They get verified-purchase reviews and direct messaging with sellers. They earn loyalty rewards.
The three sides aren't bolted on. They're the structure.
Some practical consequences:
Sellers stop chasing creators. Instead of DMing 50 TikTok accounts hoping for replies, sellers post a brief and creators apply to them. Discovery direction reverses.
Creators stop waiting for brand deals. Instead of refreshing their inbox, they browse a list of campaigns they can apply to today. The "wait to be discovered" model is replaced with a "hunt for relevant work" model.
Buyers get implicit social proof. They're not buying from a faceless marketplace listing — they're buying something a creator they trust is actively promoting.
Attribution becomes trivial. Every creator-driven sale is tracked through their unique link. The seller knows exactly who drove what. Payouts split automatically.
Compounding loops emerge. Every campaign becomes UGC content the seller can reuse. Every review compounds into trust. Every transaction can fuel an affiliate funnel that drives the next sale. Growth is partially a function of the network promoting itself.
I want to be honest about what's hard, because the failure modes of three-sided marketplaces are different from two-sided ones.
The cold start is harder. Two-sided marketplaces have one cold start problem (chicken-and-egg between supply and demand). Three-sided marketplaces have three. You need sellers to attract creators to attract buyers, but you also need buyers to make sellers want to list, but creators won't apply to campaigns with no proof of past payouts. You're solving three coordination problems simultaneously.
The honest answer is that you have to bootstrap one side at a time and accept that the system is sub-economic until all three reach minimum density. We started with sellers and creators; buyers came through the creators.
Quality control is non-negotiable. A two-sided marketplace can survive some bad listings — buyers complain, sellers get banned, system corrects. A three-sided marketplace amplifies bad listings. A creator who promotes a bad product damages their own audience and won't promote your products again. So we admin-review every product before it goes live. This is operationally expensive and slows seller onboarding by 24 hours, but it's load-bearing.
Creator tier design is a research problem. We use Bronze/Silver/Gold/Platinum with multipliers up to 2×. The tier structure has to balance "rewarding growth" against "not punishing new creators." Get it wrong and either small creators leave (multipliers too punishing at the bottom) or it's not worth growing (multipliers too flat). We've iterated on this multiple times.
Payouts are a long tail of integrations. Stripe Connect handles most of the world. But if you want to serve creators in markets where Stripe is weak, you need local rails — bKash, Nagad, etc. Each one is a separate integration with its own KYC, dispute, and reconciliation logic. This is unglamorous infrastructure work that shows up nowhere in the pitch deck but is genuinely 30% of the engineering budget.
SaaS fulfillment is harder than downloads. A digital download is "give them a signed CDN URL." A SaaS subscription is "provision their account, send them credentials, handle renewals, handle cancellations, handle disputes, handle the case where the buyer's email is different from the platform's email." We use a webhook model where the seller's endpoint receives subscription events and provisions access — this works well, but documenting it for non-technical sellers has been an ongoing challenge.
The platform we've been building this on is at peddlum.com. It's a digital product marketplace with three structural roles: sellers (90% revenue share, can post campaigns), creators (browse campaigns, apply, earn fixed fees + commissions with tier multipliers up to 2×), and buyers (admin-reviewed catalog, verified-purchase reviews, direct seller messaging).
The product types are downloadable digital files (templates, ebooks, design assets, code) and SaaS tools (webhook-fulfilled). The tech stack is NestJS / PostgreSQL / Redis on the backend, React / Vite on the frontend, Cloudflare R2 for file delivery, Stripe Connect for payouts. None of that is the interesting part.
The interesting part is whether the three-sided structure is actually a better local optimum than two-sided. I think it is, for digital products specifically, but I'm genuinely interested in arguments against — particularly from people who've worked on adjacent systems and found that the third side adds more complexity than value.
Here's the thesis, stated as plainly as I can:
For digital products, the two-sided marketplace is a local maximum. It's a stable shape, but it's not the global maximum. The global maximum is a three-sided marketplace where creators are a first-class participant in the transaction, not an external acquisition channel.
The reason most marketplaces haven't moved here is path dependence — the existing infrastructure, mental models, and business operations are all built around two sides. Adding a third side isn't a feature addition; it's a re-architecture.
I think the marketplaces that win the next decade of digital commerce will be the ones that did the re-architecture. I'm building one of them. Tell me where I'm wrong.