How to build a website like Whatnot

What is Whatnot and how does it work?

Whatnot is a live video shopping marketplace where sellers run auctions and fixed-price sales while broadcasting to an audience in real time. Buyers watch a stream, chat with the seller, and bid or buy without leaving the video. Founded in 2019 by Grant Lafontaine and Logan Head, the company started with trading cards and collectibles and has since expanded into fashion, electronics, comics, coins, and sneakers. Whatnot was valued at $20 billion in an August 2026 funding round and has reported gross merchandise value in the billions of dollars annually (CNBC).

The platform works because it solves a trust problem that plain e-commerce photos can't. A buyer looking at a photo of a graded trading card can't tell if the corners are soft or the surface has a scratch; a seller holding the card up to a camera and turning it under a light can show them directly. That live demonstration, combined with a chat window where buyers can ask questions before bidding, recreates the feel of a trading show or hobby shop rather than a cold online listing.

How Whatnot makes money

Whatnot takes a commission on every completed sale, with a standard rate of 8% of the item price in the US, Canada, and Australia, though some categories like coins and electronics carry lower rates, plus a separate 2.9% + $0.30 payment processing fee (Whatnot Help Center). There are no listing fees and no charge for hosting a show that doesn't result in sales, which lowers the barrier for new sellers to try the format before committing to it as a business.

Beyond commission, Whatnot generates revenue from seller-side tools: promoted placement that increases a show's visibility in the app's discovery feed, and analytics dashboards that help higher-volume sellers understand what's converting. The company has also piloted brand partnerships, where consumer brands sponsor shows or work with top sellers to move new inventory, and it has tested advertising placements within the browse and discovery experience.

The commission-only structure aligns Whatnot's incentives with its sellers: the platform only earns when a sale happens, which pushes it to invest in features that increase watch time, bidding activity, and repeat purchases rather than features that just extract fees from inactive accounts.

What makes Whatnot work: key features

Live streaming built for commerce, not just video. Whatnot's broadcast tools overlay bidding, current price, and chat directly on the video feed so buyers never lose context. Low latency matters enormously here: a half-second delay between a bid and its confirmation breaks the auction dynamic that drives urgency.

Real-time bidding mechanics. The platform supports English auctions (ascending bids), Dutch-style price drops, and "breaks" where buyers pay for a slot in a group pack opening. Auto-extending timers prevent a buyer from sniping a win in the last second, which keeps bidding fair and keeps sellers' trust in the system intact.

Seller-to-buyer relationship building. Following a seller, getting notified when they go live, and building a rapport over dozens of shows turns one-time buyers into repeat customers. This is closer to how independent card shops and comic stores built loyalty in person than how a typical online marketplace works.

Consolidated payment and shipping. A buyer who wins five items during one show gets one payment charge and one combined shipment, rather than five separate checkouts. That reduces cart abandonment risk during a fast-moving show and lowers shipping cost per item for the buyer.

Discovery through an algorithmic feed. Whatnot's app surfaces live shows based on category interest and past purchase behavior, similar to how short-video apps surface content. This is a meaningful departure from search-driven marketplaces like eBay, where buyers type in what they're already looking for.

Buyer protection and dispute handling. Because purchases happen fast and buyers can't inspect items physically before buying, Whatnot backs transactions with a buyer protection guarantee and a structured process for reporting items that don't match what was shown live.

The competitive landscape

eBay Live is eBay's answer to Whatnot's format, letting sellers stream and run live auctions inside the broader eBay marketplace. It benefits from eBay's massive existing seller base, established buyer trust, and integrated search traffic, so sellers don't have to build an audience from zero the way they often do on a newer platform. Where it falls short is focus: eBay Live competes for attention against eBair's traditional fixed-price and timed-auction listings, and the live format feels like an add-on rather than the core product. That gap creates room for a platform where live selling is the entire experience, not a feature bolted onto a 25-year-old marketplace.

StockX takes a completely different approach to the same underlying problem of authenticity in collectibles and sneakers. Instead of live demonstration, it uses a bid-ask pricing model and physical authentication: every item is inspected by StockX before it ships to the buyer. This works exceptionally well for standardized products like sneakers and graded cards where condition can be verified against a checklist, but it removes the social and entertainment layer entirely. There's no seller relationship, no live chat, no story behind the item. That leaves an opening for categories where the story and the seller's expertise matter as much as the physical verification, which is exactly where Whatnot has built its edge.

GOAT operates similarly to StockX, combining marketplace listings with authentication services, primarily in sneakers and now expanding into apparel and accessories. Its strength is buyer confidence: a verified authentication stamp reduces the fear of counterfeits that plagues secondhand sneaker resale. It shares StockX's limitation, though: transactions are async and impersonal, with no live element to build seller-buyer trust or urgency.

Heritage Auctions represents the high-end, traditional auction house model applied to collectibles, sports memorabilia, and fine art. Its live auctions are professionally run, often for items worth thousands or millions of dollars, with expert authentication behind every lot. This works for serious collectors and investment-grade pieces, but the formality and price floor exclude the casual collector who wants to spend $20 to $50 on a fun evening of bidding. That casual, high-frequency, low-stakes segment is where Whatnot built its early user base and where new entrants can still find room in adjacent categories Heritage doesn't touch.

Mercari offers broad, general marketplace functionality across many categories, including collectibles, with a simple fixed-price buying experience and integrated shipping. It's easy to use and has wide reach, but it has no live auction functionality and none of the entertainment value that keeps Whatnot's most active buyers watching for hours. Mercari competes on convenience and breadth; it doesn't compete on the community and discovery experience Whatnot has built.

Across all five, the pattern is clear: platforms either offer scale and authentication without a social layer, or formality and expertise without accessibility. The opening for a new entrant is a specific category where live demonstration and seller expertise matter, but where neither Whatnot nor its imitators have built deep community yet: think vintage tools, houseplants, model trains, vinyl records, or regional produce and specialty food.

How to build a marketplace like Whatnot

1. Define your niche within Whatnot's category

Whatnot itself started narrow, focusing on trading cards and Pokemon before expanding. You should do the same. Look for a category where three conditions hold: buyers currently struggle to verify condition or authenticity from static photos, sellers already have some following or credibility (a hobby, a shop, a social media presence), and the goods are distinctive enough that seeing them in motion or hearing the seller describe them changes the buying decision. Vintage watches, houseplants, model kits, vinyl, and regional artisanal food all fit this pattern better than commodity goods that don't benefit from live demonstration.

2. Validate demand before writing a line of code

Test the live-selling format cheaply using tools you don't have to build. Run a few shows on Instagram Live or a Discord voice channel with screen share, take payment through Venmo or PayPal manually, and see whether people actually show up, chat, and buy. This tells you whether your chosen niche has the right dynamics (an engaged, chatty audience willing to bid in real time) before you invest in custom infrastructure.

3. Design your commission and seller economics

Decide your take rate with reference to Whatnot's 8% to 12% range and to what your category can bear. A hobbyist reselling $15 items needs a lower friction fee than someone moving graded cards worth hundreds of dollars per lot. Decide early whether you'll charge listing fees (Whatnot doesn't) and how you'll structure payouts, since sellers running frequent shows expect fast, consolidated payments rather than per-item settlements.

4. Choose your development approach

Vibe coding from scratch. AI tools like Cursor, Lovable, and Bolt can produce a working Whatnot-style prototype quickly, complete with a video embed, a chat window, and a mock bidding interface. For pressure-testing a concept with friends or demonstrating an idea to investors, this is genuinely useful. For launching a platform that holds real buyer money during a live auction and settles it to a seller minutes later, it's probably not sufficient. What vibe-coded outputs reliably don't produce is the infrastructure underneath: payment escrow, dispute resolution, fraud detection, and compliance with money transmission rules. A documented Sharetribe experiment ran 60+ hours to reach demo quality but revealed a critical checkout exploit that would have allowed any user to manipulate transaction prices via a direct API call. Bringing that output to production standard takes significant additional time and a clear understanding of how the pieces fit together. It's a useful prototyping tool, not a cost-effective path to launch.

Custom development from scratch. Hiring developers gives you full control over the live streaming and real-time bidding mechanics that make Whatnot distinctive. A live auction marketplace with reviews, seller verification, escrow, and dispute handling sits in the production-grade tier: $80,000 to $200,000 over 16 to 28 weeks (Codica, RaftLabs). If you also want native iOS and Android apps with real-time video and push notifications at launch, expect the high complexity tier: $150,000 to $350,000-plus over 26 to 52-plus weeks. The low end of each range assumes offshore teams billing $15 to $40 an hour; US or Western European teams push you toward the top. This route makes sense once you have specific live commerce requirements that no existing platform, including a customized Sharetribe build, can meet.

Building on a marketplace operating system like Sharetribe. You start at roughly 90% done on the standard marketplace foundation: payments, user accounts, listing management, messaging, transaction flows, fraud detection, and compliance. Your time and budget go toward the features that make a live auction experience actually work. Three paths within this approach:

• No-code builder. Configure listings, categories, seller profiles, and a commission-based transaction flow from the Console without writing code. This gets you a functioning marketplace for scheduled sales and timed auctions (list an item, set a closing time, let buyers bid) quickly, even before you add live video.
• AI-assisted development. Connect Claude Code, Cursor, or Codex to Sharetribe's open APIs and open-source template to build the live-specific layer: a video embed integrated with a third-party streaming service (Agora, Mux, or Amazon IVS), a real-time bid overlay, and chat that syncs with the stream. Because Sharetribe already handles payment holds, splits, and refunds, AI-assisted development on top of it carries much lower risk than building the payment layer from scratch.
• Custom code. Build directly on the developer platform for deeper integrations like proxy bidding logic or auto-extending auction timers, or hire from Sharetribe's Expert Marketplace if you want specialized help with the live streaming layer.

Most founders start with the no-code builder to launch scheduled and timed auctions, validate that sellers and buyers want the format, then layer in live video and real-time bidding through AI-assisted development once demand is proven.

5. Solve the cold start problem

Live auction marketplaces face a sharper version of the standard chicken-and-egg problem: an empty show is worse than an empty listings page, because there's no content at all if no one is broadcasting. Whatnot solved this early by recruiting sellers directly out of existing trading card Facebook groups and Discord servers, often reaching out to individuals with modest followings and helping them run their first shows personally.

Two or three tactics work here. First, go supply-first and recruit five to ten sellers who already have some following in your niche, whether that's a regional plant shop with an Instagram audience or a vintage watch collector active on forums. Ask them to commit to a regular schedule (twice a week beats once a month) so buyers learn when to show up. Second, seed a tight geographic or community focus rather than going broad. A marketplace for vintage tools sold live to woodworkers in one region builds momentum faster than the same idea marketed to "everyone who likes tools" nationally. Third, consider hosting the first shows yourself, sourcing inventory and running the camera personally, to prove the format works before asking outside sellers to take the risk. A tight launch with five committed, active sellers outperforms a broad launch with twenty inactive ones, because buyers who show up to an empty or inconsistent schedule rarely come back.

6. Launch and build the show schedule

Once you have committed sellers, set a public, predictable show calendar and promote it inside the communities where your target buyers already spend time. Cross-post short highlight clips (a great auction moment, a rare item reveal) to TikTok, Instagram Reels, and relevant subreddits or forums, since these clips work as organic acquisition for people unfamiliar with the live format. Track the funnel from viewer to chat participant to bidder to buyer closely in the first weeks; if viewers aren't converting to bids, the issue is usually pacing or trust, not traffic.

7. Iterate on the format and expand categories carefully

Use seller and buyer feedback to refine auction pacing, bid increments, and chat moderation before adding new product categories. Whatnot expanded from trading cards into fashion, electronics, and general collectibles only after its core format was proven, and each new category launch required rebuilding seller supply and buyer trust from a lower base. Resist expanding into a second category until your first one has a healthy rhythm of repeat sellers and repeat buyers.

Do you need to build everything Whatnot has?

No. Whatnot's algorithmic discovery feed, its native mobile apps with sophisticated push notification logic, its in-house content moderation team, and its brand partnership program were all built over years with a large team and significant funding. None of that is required to launch. What's essential at launch is a working transaction flow (list, bid or buy, pay, ship), a way to broadcast video with a chat overlay, and enough trust signals (seller profile, ratings, buyer protection policy) that a stranger will hand over their payment details.

The instinct to "just vibe-code a clone" of Whatnot is understandable given how fast AI tools can produce a plausible-looking demo, but it underestimates what's underneath the demo. A convincing video call interface with a chat box is a fraction of the actual system: you still need to handle payment authorization at the moment a bid wins, hold funds until a dispute window closes, split commission automatically, and prevent the kind of price manipulation that a direct API call can expose in an unhardened build. Starting on Sharetribe's foundation is faster even with AI tools in your workflow, because the payment and compliance infrastructure already exists and passed real-world use; your AI tokens go toward the live video and bidding layer that actually differentiates your platform, not toward rebuilding a payment system that already works.

Trust and safety for a Whatnot-type marketplace

The biggest trust risk in live auctions is the mismatch between what's shown on camera and what arrives in the mail. Lighting, camera angle, and a seller's framing can make a flawed item look mint, and buyers have no chance to inspect the physical object before bidding closes. Counterfeits are a related and serious risk in categories like trading cards, sneakers, and designer goods, where fakes are sophisticated enough to fool an untrained eye even on video.

Standard verification steps in this space include seller identity verification before they're allowed to host shows, a buyer protection policy that guarantees refunds when an item materially differs from what was shown live, and, for high-value categories, a third-party authentication service that physically inspects items before shipment (similar to what StockX and GOAT do, but applied selectively to your highest-risk categories rather than every item). Sharetribe provides the transaction backbone this depends on out of the box: payment holds so funds aren't released to a seller until a dispute window passes, structured transaction flows that can require specific steps before a sale completes, and user accounts that support optional identity verification. What founders need to add is category-specific: authentication partnerships if you're in sneakers or cards, a moderation policy for live chat behavior, and a documented dispute process buyers can trust before they place their first bid.

Running a marketplace like Whatnot

Day-to-day operations for a live auction marketplace center on seller support, dispute handling, and schedule management rather than the heavier logistics operations a delivery marketplace requires. You'll spend time helping sellers plan and promote shows, reviewing disputed transactions, and moderating chat behavior during and after live events. None of this requires the scale of Whatnot's in-house trust and safety team on day one; a founder or a single ops hire can handle disputes and moderation for the first several hundred transactions manually.

Sharetribe handles the parts that would otherwise consume your early engineering time: payment capture and payout timing, commission calculation and split, hosting and uptime for the marketplace itself, and the underlying data model for listings, users, and transactions. That leaves your attention free for the things that actually differentiate a live auction platform, like show quality, seller coaching, and community tone, which are what buyers remember and what keeps them coming back to the next show.

Development costs and timeline

Three realistic scenarios:

Vibe coding from scratch: Free or very cheap to start, and a working Whatnot-style prototype (video embed, mock chat, fake bid buttons) is buildable in a day or two with AI tools. What you can't get from here is a production-ready platform: real payment authorization at the moment of a winning bid, fund holds through a dispute window, commission splitting, fraud detection against shill bidding, and compliance with money transmission requirements all require work that AI coding tools don't shortcut. A documented Sharetribe experiment logged 60+ hours to reach demo quality and still turned up a critical checkout exploit that would have let any user manipulate prices via a direct API call. It's a useful proof-of-concept tool, not a cost-effective path to a live business handling real transactions.

Custom development from scratch: Cost depends on complexity tier. A live auction marketplace with reviews, seller verification, escrow-style payment holds, and dispute flow sits in the production-grade tier: $80,000 to $200,000 over 16 to 28 weeks (Codica, RaftLabs). If you want native iOS and Android apps with polished real-time video from day one, budget move into the $150,000 to $350,000-plus range. The low end assumes offshore development teams; US or Western European agencies push costs toward the top of each range. Ongoing maintenance typically runs 15% to 25% of the original build cost per year on top of hosting. This route makes sense once you have specific requirements, like proprietary bidding algorithms or exclusive streaming tech, that no existing platform can support.

Building on Sharetribe: Subscription pricing starts at $99/month on the Lite plan, $199/month on Pro, and $299/month on Extend, all billed yearly. This covers the transaction engine, payments through Stripe Connect, user accounts, listing management, messaging, and search that a live auction marketplace needs as its foundation, leaving live video and real-time bidding as the scoped, custom pieces you build on top. Most founders reach a live MVP, launching with timed or scheduled auctions before adding live video, within a few weeks rather than months. Custom features added via AI tools or a developer stay narrow in scope because the transaction and payment engine underneath is already built and tested.

Why Sharetribe for building a marketplace like Whatnot

Sharetribe's transaction flow engine maps directly onto what an auction marketplace needs: a defined sequence of steps (bid placed, auction closes, payment captured, item shipped, funds released) that you can configure without building payment logic from scratch. Its built-in Stripe Connect integration handles the marketplace-specific complexity of collecting from a buyer, holding funds, taking a commission, and paying out a seller, which is exactly the flow Whatnot runs at massive scale. The customizable data schema lets you add auction-specific fields like current bid, reserve price, or auction close time to your listings without touching core infrastructure, and the open APIs mean your live video and real-time bidding layer, built either through the no-code builder's extension points or through AI-assisted custom code, sits on top of a payment foundation that's already handled real transactions for other marketplaces.

Frequently asked questions

How much does it cost to build a marketplace like Whatnot?

A production-grade custom build with escrow, verification, and dispute handling runs $80,000 to $200,000 and takes 16 to 28 weeks with an agency (Codica). Building on Sharetribe starts at $99 to $299 per month plus the cost of adding live video and bidding features, and most founders reach a live MVP in a few weeks. Native mobile apps with polished real-time video push custom costs toward $150,000 or more.

Can I build a live auction marketplace without coding?

Yes, for the core marketplace. Sharetribe's no-code builder handles listings, seller accounts, payments, and a timed-auction transaction flow without any code, which is enough to validate demand before you add live video streaming. Live video itself typically requires connecting a third-party streaming service, which is where AI-assisted development or a developer becomes useful.

How does Whatnot make money?

Whatnot charges sellers a standard 8% commission on most items (with some categories like coins as low as 4%), plus payment processing fees, and charges no listing fees (Whatnot Help Center). It also earns from promoted placement for sellers who want more visibility and from brand partnerships and sponsored shows. It doesn't charge buyers directly, which keeps friction low for people trying the format for the first time.

What are Whatnot's biggest competitors?

eBay Live competes directly with a similar live auction format inside eBay's larger marketplace. StockX and GOAT compete for the same collectibles buyers with an authentication-first, non-live model. Heritage Auctions serves the high-end collector segment with formal, expert-run live auctions, and general platforms like Facebook and Instagram offer live selling features without the specialized auction mechanics.

What's the hardest part of building a Whatnot-style marketplace?

Getting the cold start right is harder than the technology. You need sellers willing to commit to a regular broadcast schedule before you have an audience, and buyers won't show up to an inconsistent or empty schedule. Recruiting five to ten committed sellers in a tight niche and helping them run their first shows personally tends to work better than trying to attract many sellers at once.

Is vibe coding good enough to launch a live auction marketplace?

It's good enough to demo a concept but not to launch one that handles real money. A documented Sharetribe test needed over 60 hours to reach demo quality and still exposed a checkout exploit that would have let users manipulate transaction prices, which shows how easy it is to miss serious security gaps in AI-generated payment code. For a live launch, start on infrastructure that's already handled real transactions, like Sharetribe, and use AI tools for the features specific to your marketplace.

Related guides

• How to build a marketplace for antiques
• How to build an adventure gear marketplace
• How to build a marketplace for wedding items
• How to build a marketplace for party supplies
• How to build a website like Depop