Marketplaces and platforms

Marketplace development in Australia

We built a national marketplace and then maintained it for nearly a decade, which is long enough to have learned what actually breaks. It is almost never the transaction.

Tell us what you are matching

The cold start is the whole problem

A marketplace has to be useful to one side on a day when the other side has not arrived. That is the constraint everything else follows from, and it is a product problem rather than an engineering one. Most marketplaces that failed were technically fine.

Which is why the first thing we do is strategy, and why one of the more common outcomes of that work is a recommendation to launch as something narrower. If you can be genuinely useful to sellers alone, or buyers alone, you have a product that works before the network exists and a reason for the second side to show up. We would rather tell you that now than a year in.

What we have built

Yume Food: a national surplus food marketplace

Food manufacturers end up with stock they cannot sell through normal channels, and matching a pallet of it to the right buyer at the right price inside its remaining shelf life is a discovery and logistics problem. Yume solved it as a marketplace, with donation running alongside sale through the same catalogue. We were the development team from 2016 until the company wound up. Over the platform's life Yume reported returning $30 million to Australian manufacturers and redistributing 11.5 million kilograms of food. Read the Yume Food case study.

Community Choice: a redemption platform

Different domain, similar mechanics. Relief is issued through a network of more than 25 agencies, redeemed at participating local venues, and reconciled and reported for the Foundation that funds it. Multiple parties, value moving between them, and an obligation to account for every cent. Read the Community Choice case study.

AirVendor: two sides meeting at a door

Not a marketplace in the commercial sense, but structurally the same shape: facilities on one side, industry representatives on the other, and software that has to serve both without letting the rules drift apart. Read the AirVendor case study.

What we have learned the hard way

The listing model decides your next five years

A marketplace whose listing model does not natively understand expiry, partial quantities and more than one exit route for the same item will fight itself for the rest of its life. This is the least glamorous decision in the build and the most expensive one to get wrong, which is why we settle the data layer before anything is drawn on a screen.

Admin is not an afterthought

Somebody has to resolve disputes, verify sellers, adjust a transaction that went wrong and answer the question nobody anticipated. Marketplaces that treat the operator's tooling as a phase two end up with a team doing that work in the database, which is exactly as bad as it sounds.

Money makes everything stricter

The moment value moves between parties you inherit reconciliation, refunds, partial fulfilment and an audit trail. Designing so history is preserved rather than overwritten is what lets you answer questions about the past, and in anything financial you will be asked.

Build it API-first or regret it

Marketplaces grow surfaces. A web app, then an operator console, then a mobile app, then a partner integration. We build API-first with Laravel underneath and Vue on top, so a rule is written once and every surface obeys it.

What it costs and how we start

A marketplace costs more than a single-sided product of the same apparent size, because you are building two experiences plus the mechanics between them. A focused first version generally starts around $60,000 to $80,000, and once there are payments, messaging, disputes and an operator backend it commonly lands between $80,000 and $200,000. The fuller picture is in what it costs to build an app in Australia.

We start with strategy, then scoping, then a staged build. If the honest first step is a narrower product, MVP builds is where that conversation goes.

Marketplace development: common questions

How much does it cost to build a marketplace?

More than a single-sided product of the same apparent size, because you are building two experiences and the mechanics that connect them. A focused first version generally starts around $60,000 to $80,000, and anything with payments, messaging, disputes and an admin backend commonly lands between $80,000 and $200,000. Strategy is where that number gets decided rather than discovered.

What is the hardest part of a marketplace build?

Not the transaction. It is that both sides have to be present before either side gets value, so the software has to be useful to a seller on a day when no buyer turns up. Every marketplace that failed technically was fine and failed at that. It is a product problem before it is an engineering one, which is why we start with strategy.

Have you built a marketplace before?

Yes. We were the development team for Yume Food, a national marketplace for surplus food, from 2016 until the company wound up nearly a decade later. Over its life Yume reported returning $30 million to Australian manufacturers and redistributing 11.5 million kilograms of food. We also built the redemption platform behind Community Choice, which has similar mechanics.

Should I start with a marketplace or a single-sided product?

Often the single side. If you can be useful to sellers alone, or buyers alone, you have a product that works before the network exists and a reason for the other side to arrive. We will tell you if we think your marketplace should start life as something narrower, because that conversation is cheaper now than a year in.

"One of the most impressive aspects of working with OK200 was our direct interaction with Paul, the CTO and Founder, throughout the entire process. It is rare to have such direct and sustained access to a company's leadership."
Arkan Youssef, Co-Founder, Leaflet

Drop us a line