Building an Etsy-Style Marketplace: A Developer's Checklist for Fees, Escrow, Discovery and Sell-Through
Every few months a founder sends us a spec that boils down to one line: "Etsy, but for X." The screenshots are usually Etsy's listing page, Etsy's shop page and Etsy's checkout. What the spec almost never describes is th
Every few months a founder sends us a spec that boils down to one line: "Etsy, but for X." The screenshots are usually Etsy's listing page, Etsy's shop page and Etsy's checkout. What the spec almost never describes is the machinery underneath: how money moves, how a buyer finds the right item among thousands, how a new seller earns trust and how the platform keeps enough buyers and sellers active at the same time.
That machinery is the actual product. This post treats an Etsy-style marketplace the way an engineer should: as a set of interacting systems with real constraints, not a UI to clone. It is based on Journeyhorizon's breakdown of how to build a marketplace like Etsy, reorganised as a working checklist.
Start with the business model, because it shapes the schema
Etsy holds no inventory. Independent sellers keep their own stock, ship directly to buyers and pay their own fulfilment costs. Etsy makes money from two fees: $0.20 per listing plus 6.5% of the transaction value.
That model has direct consequences for your data model and services:
- No warehouse logic. You don't need stock reservation across a central inventory, but you do need per-seller inventory and per-seller shipping rules.
- Two revenue events. A listing fee fires when a listing is created or renewed; a commission fires on a completed transaction. Those are different code paths with different failure modes.
- Money that isn't yours. Most of every payment belongs to the seller. Your system has to split it, hold it when needed and release it at the right moment.
A minimal fee model might look like this:
listing_fee = 0.20 per listing
commission = 0.065 * order_subtotal
seller_payout = order_subtotal - commission (- payment processing, depending on your provider)
platform_revenue = listing_fees + commissions
The constants are Etsy's; yours will differ. The point is to model fees as configuration from day one, because you will change them.
The systems are coupled, so design them together
The source article makes a point that is easy to miss: these systems are interconnected. Payment architecture affects how disputes play out. Seller discovery affects seller retention. Poor discovery leads to churn, churn leaves stale listings, and stale listings make discovery worse for everyone.
In other words, you have a feedback loop:
weak discovery -> sellers get no sales -> sellers churn
^ |
| v
stale listings <---- abandoned shops stay live
When you plan your architecture, trace that loop explicitly. For each component, ask what happens downstream if it underperforms.
Table stakes vs differentiators
A useful way to scope an MVP is to split features into two buckets.
Table stakes (solved problems, don't reinvent them):
- [ ] Payment processing
- [ ] Basic search
- [ ] Seller profiles
- [ ] Reviews
Possible differentiators (where custom work can pay off):
- [ ] Your discovery or ranking algorithm
- [ ] Your seller verification process
- [ ] Your payment model
Buyers expect table stakes to work and won't choose you because of them. Differentiators are where your niche either gets served better than on a generalist platform, or doesn't.
Decisions that are expensive to reverse
Experienced marketplace teams know which corners can be cut early and which ones turn into migrations later. Three are worth flagging in any architecture review.
1. Payment processor and escrow logic
Choosing a payment processor and designing the escrow logic are among the hardest choices to undo. Payouts, refunds, disputes and seller onboarding all depend on them. Pick a provider that supports split payments and holds for your niche, and wrap it behind your own service interface so business logic doesn't leak provider-specific assumptions everywhere.
2. Seller onboarding
Friction in seller onboarding is a leading indicator of churn. Instrument the funnel from sign-up to first published listing to first sale. If sellers stall at verification or at payout setup, you'll see it there before you see it in revenue.
3. Search ranking
Search ranking needs to be tuned to your own inventory mix rather than borrowed from a general ecommerce platform. A handmade jewellery marketplace and a vintage furniture marketplace have very different listing volumes, price ranges and buyer intent. Keep ranking configurable, and log the queries and results so you can tune it with real data.
Instrument the one metric that tells the truth
If you only build one dashboard, make it sell-through rate: the share of listings that sell within a given period.
The source gives a clear example. A marketplace with 10,000 listings and a 1% sell-through rate looks busy, but it isn't working. Low sell-through points to a matching problem: buyers and the right listings aren't finding each other.
-- Sell-through rate for a period (illustrative)
SELECT
COUNT(DISTINCT CASE WHEN o.listing_id IS NOT NULL THEN l.id END)::float
/ NULLIF(COUNT(DISTINCT l.id), 0) AS sell_through_rate
FROM listings l
LEFT JOIN orders o
ON o.listing_id = l.id
AND o.completed_at BETWEEN :start AND :end
WHERE l.active_during_period = true;
Track it per category and per seller cohort. When it drops, look at discovery and onboarding before you look at marketing.
Pick the stack for your stage, not your ambition
Founders usually face limited runway, a small team and the need to validate before investing heavily. The common options are:
- A no-code marketplace builder, which can go live in weeks with payments, reviews and storefronts, but with limited flexibility.
- Sharetribe, a middle ground with proven infrastructure plus customisable transaction flows, seller tools and integrations.
- WooCommerce or coding from scratch, which give more control and more work.
The source's rough timelines are useful for planning:
| Approach | Typical time to launch |
|---|---|
| No-code marketplace | Weeks |
| Custom MVP | 4â6 months |
| Feature parity with an established platform | 12+ months |
Custom development makes sense when your niche needs features or payment models that no-code builders can't support. It does not make sense for a general marketplace that hasn't found product-market fit yet.
Go narrow before you go wide
Etsy took nearly 20 years to reach its current scale, and it spans categories with very different buyer intent and seller behaviour. Copying that breadth is the wrong starting point.
The better engineering brief is depth: pick one niche, understand it and build what that community needs. The most useful question in your discovery sessions is "what problem does this niche have that Etsy doesn't solve well?" The answer usually tells you which differentiator to invest in.
Why launches fail (it's rarely the code)
Most marketplace projects fail in their first two years, and usually not because of technical flaws. The common causes are misunderstanding the chicken-and-egg problem, underestimating operational overhead, not validating assumptions with real buyers and sellers, and building payments, reviews and search ranking that don't fit how the niche actually behaves.
A few FAQs from the engineering side
Do we need custom code to launch? No. A no-code builder can get you live in weeks. Bring in custom development when you hit its limits or need a technical edge.
What commission should we configure? Etsy's 6.5% is the common benchmark. Lower fees attract sellers but limit revenue; higher fees need to be justified with more value, such as better reach, tools or higher-intent buyers.
Do we need a mobile app at launch? Not usually. For handmade or vintage marketplaces, a responsive website is enough at first, because many buyers browse on mobile and buy on desktop. Build the app after product-market fit and steady mobile traffic.
Takeaway
An Etsy-style marketplace is a payments system, a discovery system and a trust system wired together. Design the fee model as config, protect your escrow and payment choices, instrument sell-through from day one and launch narrow on a stack that fits your stage.
If you want a team that has worked with more than 200 marketplace founders and operators to pressure-test that plan, Journeyhorizon offers ecommerce marketplace development from MVP to scale.
Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes â full credit and traffic to the original publisher.