Niche Marketplace MVP: Five Live Sites, Audited by How They Take Money

Almost every "marketplace MVP case study" post is a list of features with a founder quote attached. That is the least useful thing you can copy, because features are cheap to describe and expensive to verify, and because nobody's feature list is the reason their build took nine months instead of six weeks.

The money model is the reason. It is also the one thing about somebody else's marketplace you can check yourself, today, without an interview or a screenshot from X: it's written on their pricing page, in their help centre, and in their terms. So this is an audit of five live niche marketplaces along a single axis — where the money enters the platform — and what each of those answers forces into your first release.

How this audit was done

Every figure below was read off the operator's own page on 4 September 2026 and linked so you can check it. No third-party roundups, no revenue estimates, no traffic tools. Where a company does not publish a number, this post says it does not publish a number instead of borrowing one from a blog that guessed.

Two candidates were dropped for exactly that reason. Thumbtack's lead-price documentation renders client-side and could not be retrieved from its own domain, and the only dollar figures available were forum posts from individual pros — not a source. Bring a Trailer's site refuses automated retrieval outright. Both are interesting businesses; neither made it in, because the alternative was quoting hearsay about someone's pricing.

What's left is five operators covering three ways of taking money, which turns out to be the whole ladder.

The five, and where the money enters

Three tiers, in ascending order of what they cost you to build: charge for the listing, charge for the introduction, charge for the transaction. Everything else in a marketplace MVP is downstream of which one you pick.

Tier 1 — charge for the listing

We Work Remotely sells a $299 standard job post and, on the same screen, a $1,999 "Mega Placements" tier and bulk posts "from $179/job". Remote OK sells the same product shape at "starting from $299 for 30 days" — and then does something more interesting.

Remote OK's post-a-job page states plainly: "Pricing of job posts and extra features is dynamic and may change based on how many jobs are posted every week, for example to avoid too many sticky jobs at one time." On 4 September 2026 the button at the bottom of that page read Post a job — $447, against a headline price of $299, because the default add-on selection was priced in. The same page also carries a policy most listing sites don't dare publish: "Guaranteed 200+ [Apply] clicks or we auto bump it for free."

That is two competitors in one niche, at the same headline price, with opposite architectures behind it. One has a price list. The other has a pricing engine with supply-side inputs, an add-on matrix, and a service-level promise it has to measure to honour.

Both companies also renew by default. Remote OK's terms say posts "renew automatically after 30 days" unless the poster disables it, with a reminder email seven days ahead and self-service refunds within seven days of renewal. That single sentence is a whole quarter of engineering the feature list never mentions: recurring billing, a pre-notice job, a self-serve refund path, and the dunning states around all three.

Tier 2 — charge for the introduction

Bark sells credits. A pro sees leads for free and pays credits to contact one. The mechanics are published in detail: "No commission, no hidden fees", costs "calculated based on the service, the value of the job, and the supply and demand in the area", credit cost shown before you confirm, follow-up messages free, and "All Credits are valid for 3 months from the date of purchase."

The price is not published. Anywhere on that page. You cannot find out what a plumbing lead in Phoenix costs without registering as a professional.

That is a deliberate choice and worth understanding rather than sneering at. Publishing a per-lead price invites arbitrage against the pro's own margin and makes every price change a public event. Not publishing it means the platform can price by demand without renegotiating anything. The cost is that the pro's first experience of your economics is a signup wall — and the trust deficit that creates is exactly what Bark's credit-back guarantee ("we'll return all your credits if you don't [win business], no questions asked") exists to pay off.

The build implication is bigger than tier 1 by a wide margin. A prepaid balance is a ledger: purchases, holds, debits, expiry, refunds, and a dispute path when a pro says a lead was fake. You cannot fake a ledger with a counter column. And the credit-back guarantee is a support workflow with a policy behind it, not a feature.

If lead capture is the model you actually want, DirectoryLaunch ships the request-a-quote side of it — single-listing and fan-out requests, double opt-in so leads are real, and paid lead subscriptions billed through Stripe. The Quotes & Leads docs describe the gating switch that decides whether free-plan owners see full contact details or a masked lead. What it does not ship is a prepaid credit ledger; subscriptions are a simpler object than a balance, on purpose.

Tier 3 — charge for the transaction

Reverb charges nothing to list — "Listing your gear is always free on Reverb" — and takes 5% of the sale plus 3.19% + $0.49 in payment processing, which the page adds up as "= 8.19% + $0.49 Total Fees". It headlines this as: "When your instrument sells, you earn 91.8% of the selling price."*

That 91.8% is honest about the percentage and silent about the flat fee, which matters at small basket sizes. Run their own formula: on a $500 guitar the seller keeps $458.56, or 91.7%. On a $25 patch cable they keep $22.46 — 89.8%. The headline figure is the asymptote you approach as the order gets bigger, not the number a seller of cheap accessories experiences. Two points of margin is nothing on a guitar and a meaningful chunk of the reason nobody lists $25 items if your niche's average order value is $25.

Rover is the most instructive of the five, because it runs two different published fee structures in the same product.

In most of the US: the provider pays 20% ("you take home 80% of your earnings"), the owner pays an 11% booking fee, and — the detail that decides your data model — "the list price shown on a pet care provider's profile includes the service fee paid by the pet care provider."

In California, "for legal reasons specific to California", the whole thing is inverted. The provider enters the number they take home. The platform adds a 25% marketplace fee "of the rate pet care providers set", rounded to the nearest dollar, and displays that as the public rate; the 11% booking fee still applies at checkout.

Run the arithmetic and the two structures land on the same money:

Twenty per cent off the top is twenty-five per cent added to the bottom. The California variant is a disclosure requirement expressed as a pricing rule, not a price increase — and it means Rover's pricing engine, search results, profile pages and checkout all branch on jurisdiction, in both directions, with rounding rules attached.

The side that never pays

Read all five pricing pages back to back and one thing is identical across every model. Bark's page opens with "it's free to sign up and see leads on Bark". Rover's help centre states "It's free to sign up for a Rover account as a pet parent". Reverb gives sellers "Unlimited Free Listings". Both job boards publish a posting price for employers and no price for seeing the jobs.

In every case the party that pays is the one with a commercial motive, and the party that generates the demand signal pays nothing, ever. That is not generosity; it's the only sequencing that works. Whichever side you charge, the other side has to be there first, and it will not arrive if there's a paywall between it and the thing it came for.

Which side is which differs, though, and it decides your whole go-to-market. On the job boards the employer pays and the candidate is the inventory. On Rover and Reverb the person supplying the service or the guitar pays. On Bark the pro pays for access to demand the platform generated. Before you design a checkout, be clear about which of your two sides is scarce — because that is the side you cannot charge.

What each tier costs you to build

This is the table the feature-list posts should have printed instead.

The gap between row one and row five is not incremental. It is the difference between a project you can estimate in hours and one you should estimate in quarters — which is the same conclusion the engineering-hours estimate arrives at from the other direction, and the reason the cost-to-build breakdown refuses to give one number for "a marketplace".

Where we fit, and where we don't

DirectoryLaunch covers rows one through four. Premium listings are a one-time charge, promotions and sponsor slots are recurring Stripe subscriptions, paid lead subscriptions bill owners for a stream of matching leads, and the store format turns the same codebase into a single-owner shop with native Stripe Checkout and optional print-on-demand fulfilment. All of it is documented in Payments.

Row five we do not do, and it isn't a roadmap tease. The store format is single-owner: the buyer pays you. There is no Stripe Connect, no seller onboarding, no split payment, no payout ledger. If your model is "buyers pay sellers and I keep a slice", Sharetribe is built around exactly that flow and has spent years on it — which is what our own three-way comparison says too. Buying a directory boilerplate and then bolting a payments platform onto it is the most expensive possible route to a take-rate marketplace.

The other honest note: nothing above removes the operational cost. A lead marketplace still needs someone deciding whether a disputed lead gets refunded. A listing site still needs someone rejecting the scam post. None of the five companies in this audit built their way out of that, and neither will you.

Audit your own shortlist

Do this for the three sites nearest your niche before you write a line of code. The answers change the estimate more than any feature decision will.

What to do tomorrow morning

Open your own pricing page draft and answer one question in writing: at what event does money enter the system? Not "how will I monetise" — the event. A listing being created. A contact being unlocked. A booking being paid.

Then find the two live sites nearest your niche, read their fee pages the way this post read these five, and write down every number with the date next to it. If a competitor doesn't publish theirs, note that too — it tells you what the market tolerates.

If the honest answer to the first question turns out to be "money enters when a buyer pays a seller", stop shopping for directory boilerplates, including ours, and go price a marketplace platform. If the answer is any of the other four, you're building a directory with a payment surface, and the work in front of you is measured in weeks.