A marketplace looks like a shop with more sellers. Commercially and technically it is a different business. A shop has one party taking money and one party shipping goods. A marketplace has three parties, buyers, vendors and the operator, and the operator sits in the middle of every order, every payment, every dispute and every refund. The decision to build, rent or install a plugin is therefore not a catalogue question. It is a question about who holds the money, who owns the data, and how much of the commercial model you can change after launch.
This guide is written for founders and operations leads who need to choose a route before spending budget. It covers what a marketplace must do, the three routes and where each one breaks, how money moves between buyer, operator and vendor, a worked ledger example with every assumption exposed, and a short validation plan you can run before committing to a build.
What a Marketplace Has to Do That a Shop Does Not
Strip away the design and a marketplace is a set of workflows layered on top of a normal ecommerce flow. Every route you evaluate must handle these, either out of the box or through work you will pay for:
- Vendor onboarding. Application, document collection, approval, suspension, and the ability to reject without deleting history.
- Catalogue ownership. Each product belongs to a vendor. Vendors edit only their own products, and the operator may need to approve new listings or changes before they go live.
- Split orders. A basket with items from three vendors is one payment for the buyer and three fulfilment jobs for the operator. Each vendor needs to see only their part.
- Commission rules. A percentage, a flat fee, a rate per category, a rate per vendor, or a combination, applied consistently and recorded at the time of sale.
- Payouts. A defined moment and method for turning vendor earnings into money in the vendor’s account, with a record that survives refunds and chargebacks.
- Disputes and refunds. Partial refunds on a split order, returns routed to the right vendor, and a clear rule on who absorbs the cost.
- Trust signals. Vendor ratings, verified badges, product reviews, and moderation of all three.
- Communication. Notifications to vendors on new orders, to buyers on status, and a record of messages between them if you allow contact.
A shop with a few extra roles can imitate the first two. The rest is where marketplaces are won or lost, and it is where the three routes differ most.
The Three Routes
Route 1: Rent a hosted marketplace platform
Hosted marketplace products give you vendor management, commission settings and payouts as a subscription service. You configure rather than build. The advantages are speed and a smaller operational surface: hosting, security patches and upgrades are the provider’s job. The constraints are the mirror image. You work inside the provider’s data model, you depend on the payment methods the provider supports, and your vendor, order and customer data lives in a system whose export options you must test before signing. For an Egyptian operator, the first question is not features. It is whether the platform can collect payment through the gateways your buyers actually use and pay vendors in Egyptian pounds to Egyptian bank accounts or wallets. Ask for that in writing, with the gateway named.
Route 2: Build on WooCommerce with a marketplace plugin
The common open-source route is WooCommerce plus a multi-vendor plugin such as Dokan, WCFM Marketplace or MultiVendorX. You own the site, the database and the hosting. The plugin adds vendor dashboards, vendor-scoped products and orders, commission settings and a withdrawal workflow. Public documentation from Dokan, for example, describes vendors requesting withdrawals from their dashboard, administrators approving them, a minimum withdrawal threshold set by the site owner, and withdrawal methods that can be manual, scheduled by the administrator, or custom. Approval inside the plugin does not itself move money: the documentation notes that the actual transfer depends on the payment gateway and the processing time. That detail matters, and we return to it in the money section.
This route suits operators who want to validate a marketplace idea quickly with a standard feature set, and who are comfortable that the plugin’s model of commission, order splitting and payouts is close to their own. It becomes expensive when the commercial model diverges from the plugin’s assumptions and every difference becomes custom code on top of code you do not control. For the wider trade-offs of the WooCommerce ecosystem for Egyptian merchants, see our comparison of WooCommerce, Shopify and a custom store.
Route 3: Custom build
A custom marketplace is designed around your commercial model: your commission logic, your vendor lifecycle, your fulfilment rules, your reporting. In practice that usually means a Laravel back end with a purpose-built data model for vendors, sub-orders and a ledger, and a front end built for your category’s search and browsing behaviour. The advantage is that nothing in the system exists unless you need it, and nothing you need is missing. The cost is that you pay for everything the plugin gives you for free, including vendor dashboards, notification plumbing and admin screens, and you take on the maintenance of all of it. Our marketplace development work sits in this space, and for the framework behind choosing the technology, read WordPress or Laravel.
Money Is the Hard Part
Most failed marketplace builds fail on payments, not on pages. There are two basic ways money can move, and the route you choose determines which of them is realistic.
Model A: Collect, hold, then pay out
The buyer pays the operator through a normal checkout. The operator holds the funds, calculates each vendor’s share after commission, and pays vendors later by bank transfer or wallet, either on a schedule or on request. This is the model that fits plugin-based marketplaces in most markets, and it is the one the Dokan withdrawal documentation describes: earnings accumulate, the vendor requests a withdrawal, the administrator approves, and the operator transfers funds. It works with any ordinary payment gateway because the gateway only ever sees one merchant, you.
The cost of this model is on your side of the table. You are holding other people’s money. You need a reliable record of what is owed to whom, a payout process that is accurate and auditable, and a clear position on what happens to a vendor’s balance when an order is refunded after payout. Whether holding and redistributing funds needs any particular licence, registration or tax treatment in Egypt is a question for a qualified lawyer and accountant before launch, not something to discover afterwards.
Model B: Split at checkout
Some gateways offer a marketplace or split-payment capability, where the buyer’s payment is divided at the moment of sale and the vendor’s share is settled to the vendor directly. This reduces the funds you hold and can simplify trust, but it depends entirely on what your gateway supports and on onboarding each vendor with the gateway. We have not verified split or marketplace features for each Egyptian gateway and we do not assume them. If this model matters to you, ask the gateway’s technical contact for the exact capability, the onboarding requirements per vendor, and the settlement timing, and get the answer in writing before design starts. Our guide to connecting Paymob, Kashier or Fawry covers the single-merchant integration that Model A builds on.
What a payout process needs in either model
- A ledger that records every financial event as an entry: sale, commission, refund, adjustment, payout. Balances are calculated from entries and never edited by hand.
- Commission captured at the time of sale. If you change your rate next month, last month’s orders must not change.
- A hold period before earnings become withdrawable, long enough to cover your return window.
- Handling for refunds after payout: negative balance carried forward, deduction from the next payout, or recovery from the vendor.
- Payout batches with a status for each line (pending, approved, sent, failed) and a reconciliation view that matches payouts to bank statements.
- A record of the vendor’s payout details, with changes logged and, for high-value accounts, a confirmation step to defend against account takeover.
Commission Models and Their Consequences
Commission design affects behaviour as much as revenue. A flat percentage is simple to explain and to implement. A per-category rate lets you take less on low-margin categories and more where your platform adds the most value. A per-vendor rate rewards large or strategic vendors but creates an operational burden, because every exception has to be stored, explained and audited. A subscription for vendors, with a lower or zero commission, suits marketplaces where vendors are professionals with steady volume. A listing fee suits classifieds-style platforms where transactions complete off-site.
Whatever you choose, test it against the case that breaks simple models: a multi-vendor basket with a coupon and one partial refund. Who funds the coupon, the operator or the vendor? Is commission calculated before or after the discount? Is shipping part of the commissionable amount? A plugin will have an opinion on each of these. A custom build lets you set your own. Either way, write the answers down before you pick a route, because they are the specification.
A Worked Ledger Example, With Assumptions Exposed
The numbers below are illustrative assumptions, not Egyptian market data and not Kemetova pricing. They show why the ledger matters.
Assume a basket of two items from two vendors. Vendor A sells an item at 1,000 EGP. Vendor B sells an item at 600 EGP. The operator charges a 10 percent commission on item price, excluding shipping. Assume a gateway fee of 2.5 percent of the total paid, and assume the operator absorbs the gateway fee rather than passing it to vendors. Assume shipping of 80 EGP is charged on Vendor A’s item and belongs to Vendor A.
- Buyer pays: 1,000 + 600 + 80 = 1,680 EGP.
- Commission: 10 percent of 1,000 = 100 EGP from Vendor A; 10 percent of 600 = 60 EGP from Vendor B. Total commission 160 EGP.
- Vendor A earns: 1,000 – 100 + 80 = 980 EGP. Vendor B earns: 600 – 60 = 540 EGP. Total owed to vendors 1,520 EGP.
- Gateway fee at the assumed 2.5 percent of 1,680 = 42 EGP, paid by the operator.
- Operator net on this basket: 160 – 42 = 118 EGP, before hosting, support, fraud losses and staff cost.
Now a partial refund: the buyer returns Vendor B’s item. Refund of 600 EGP goes back to the buyer. The ledger must reverse Vendor B’s 540 EGP earning and the operator’s 60 EGP commission. Whether the 2.5 percent gateway fee on that 600 EGP is recovered depends on the gateway’s terms, which you must confirm. If Vendor B had already been paid, the reversal creates a negative balance that must be recovered from later sales or collected directly. A system that stores only a running balance per vendor cannot reconstruct what happened. A system that stores entries can, and can explain it to a vendor who disputes the figure.
The same arithmetic shows how thin the operator’s margin can be at a modest commission rate. Before choosing a route, run your own numbers with your real commission, your real gateway terms and a realistic refund rate. A route that makes the model unprofitable on paper is the wrong route, however attractive its feature list.
The Technical Questions That Separate the Routes
Order and sub-order structure
Plugins typically create a parent order for the buyer and child orders for each vendor. Check how status changes propagate. If one vendor ships and another cancels, what does the buyer see, and what does the parent order show? In a custom build you define this explicitly, usually with an order group and independent vendor fulfilments, each with its own state machine and its own refund path.
Search and catalogue scale
Ten vendors with a few hundred products work on default search. Hundreds of vendors and tens of thousands of listings do not. Faceted search, vendor and category filters, and relevance ranking usually need a dedicated search engine such as Elasticsearch, OpenSearch or Meilisearch, fed by an indexing pipeline that updates quickly when a vendor changes price or stock. Decide early whether stale search results are acceptable for a few minutes or must be near real time.
Vendor uploads and moderation
Every vendor is a content publisher. You need image handling that resizes and compresses uploads, file-type restrictions, malware scanning for anything beyond images, and a moderation queue for new products and edited descriptions. Without these, page speed and brand quality degrade as vendor count grows. Product images are the largest part of most marketplace pages, so image delivery is a performance feature, not a design detail.
Permissions and data isolation
A vendor must never see another vendor’s orders, customers or earnings. In a plugin this relies on the plugin’s capability checks being correct on every endpoint, including AJAX actions and REST routes added by other plugins. In a custom build, you enforce it at the query level with vendor-scoped queries and automated tests that try to break isolation. This is one of the first places we look in a security review of any marketplace.
Fraud, spam and abuse
Marketplaces attract fake vendors, stolen-card testing, review manipulation and off-platform deals designed to bypass commission. Plan for vendor verification, rate limits on registration and checkout, review controls that tie reviews to completed orders, and a policy on contact details in listings and messages. These are product decisions as much as technical ones, and they are cheaper to design in than to retrofit.
Performance under load
Marketplace pages are dynamic, personalised and image-heavy. Full-page caching helps anonymous browsing but must be bypassed or fragmented for carts and logged-in vendors. Plan for object caching, a CDN for images, queue workers for notifications and payout jobs, and database indexes on the vendor, status and date columns that your dashboards filter on. Vendor dashboards are often slower than the storefront because they run heavy aggregate queries on the same database; separate reporting tables or read replicas solve it when it matters.
How to Choose: Six Questions
Score each question honestly. Mostly answers in the first column point to a plugin or hosted platform; answers in the second column point to a custom build.
- How standard is your commission model? A single percentage with a hold period fits a plugin. Tiered, category, subscription or negotiated rates, or commission on shipping and fees, point to custom.
- How do you plan to pay vendors? Manual or scheduled bank or wallet payouts fit a plugin. Split at checkout through a gateway needs verified gateway support and often custom integration.
- How many vendors and listings in year one and year three? Dozens of vendors and a few thousand products fit a plugin. Hundreds of vendors and tens of thousands of listings need search and caching architecture a plugin does not provide.
- Is the marketplace the product or a channel? If the marketplace is the business, own the model and the data. If it is an experiment to test demand, start with the fastest route that can be exited.
- What is unique about fulfilment? Standard shipping fits a plugin. Services, bookings, rentals, scheduled delivery or vendor-specific logistics usually need a custom workflow. If your marketplace sells time or capacity rather than stock, a booking engine is the core of the system, not an add-on.
- What is your exit cost? Ask any supplier, including us, how you would export vendors, products, orders and the financial ledger, and what it would take to move them to another system.
Cost Layers Without Numbers
We are not quoting prices in a general article, because the largest cost drivers are specific to each project. The layers are the same whichever route you choose, and you can ask for each one in a quotation: the platform or plugin licences, the vendor-facing dashboards, the buyer-facing storefront and search, payment and payout integration, the ledger and reporting, administration and moderation tools, security work and testing, content and vendor onboarding, and the recurring cost of hosting, monitoring, updates and support. A plugin route moves cost from build to ongoing licences and customisation. A custom route moves cost to the build and to maintenance you control. Our customer portal and custom system cost model uses the same layered approach and shows how to compare quotations on a like-for-like basis.
A Pragmatic Path: Start Narrow, Keep the Ledger Clean
For many operators the lowest-risk plan is neither pure plugin nor full custom. Validate demand with a small curated vendor group on the fastest route that fits your commission model, and treat the financial data as the asset to protect. Keep your own record of every sale, commission and payout outside any plugin-specific table, in a form you can export and import elsewhere. If the model proves itself and the plugin’s limits start costing more than a rebuild, you will migrate a clean ledger into a purpose-built system rather than reconstructing history from order notes.
If you are building a membership layer for vendors, such as paid vendor plans, the billing design overlaps with our guide to membership websites and how to charge, and the two should share one subscription state machine rather than be built twice.
A Thirty-Day Validation Plan
- Days 1 to 5: Write the commercial rules on one page: commission, hold period, refund handling, shipping, coupons, who funds what. Run three worked examples like the one above.
- Days 6 to 10: Confirm payment capability in writing with the gateway or platform: collection methods, payout options, onboarding requirements, settlement timing. Take legal and tax advice on holding and redistributing funds.
- Days 11 to 20: Build a clickable or lightly functional version with five to ten real vendors and real products. Do not build a full system before real vendors have tried to use the dashboard.
- Days 21 to 30: Process test orders through every path: multi-vendor basket, coupon, partial refund, post-payout refund, failed payout. Check that the ledger explains every figure.
The outcome is a decision backed by evidence: which route fits, what it will really need, and what it will cost to run. A marketplace that can reproduce every pound owed to every vendor from its ledger is a marketplace you can scale. One that cannot is a liability that grows with every order.