12 Marketplace Decisions You Have to Make Before Launch

If you get one meeting with the board about the marketplace, this is the list worth talking through. Not because the rest does not matter. The rest follows from these twelve.

These decisions share one trait, and that trait is why they come first: changing any of them after launch costs you a rewrite rather than a fix.

This article breaks down:

• Why are these decisions hard to reverse?
• Which 12 decisions have to be made before launch?
• How do you spot a decision that was made by accident?
• What does each one cost if you get it wrong?

Key insights

• Irreversibility has three sources: the data model, documents already issued, and contracts with third parties. All twelve decisions touch at least one.
• The first three settle who you are in the transaction, and they follow from the platform model you pick.
• A decision made by a default setting is still a decision. If nobody remembers making it, the installer made it.
• Three of the twelve need a lawyer or a tax advisor, so plan that time into the schedule.
• A vendor who prices a project without asking these questions is pricing a license rather than an implementation.

Why are these decisions hard to reverse?

It pays to understand the mechanism, because then you can judge any new decision the project brings up by yourself. Irreversibility has three sources, and only the first one is technical.

1. The data model

If you are not recording a piece of information from day one, you cannot reconstruct it afterwards. You cannot add a field that never existed to two hundred thousand offers, because nobody knows that value.

You do not, and neither does the seller who posted the offer eighteen months ago.

2. Documents already issued

The invoice, the correction, and the settlement statement have gone to the seller and to accounting. Even if you change the configuration technically, the history stays in the old arrangement, so from that point on your books hold two parallel worlds.

3. Contracts with third parties

The seller terms, the contract with the payment provider, the commission grid published as a PDF. Changing how the roles are split means amending your entire seller base rather than changing a setting.

Any decision that touches even one of these three belongs on this list. All twelve touch at least one.

Which decisions settle who you are in the transaction?

1. The platform model

A marketplace on commission, your own sale with an invisible supplier, one invoice from you at a price the seller sets, or several models running at once. It settles everything below it: who invoices, who accounts for the tax, how you earn, and whose money passes through your hands.

In many systems, this is a setting made when the platform is created, and it cannot be changed later. Even if your vendor lets you change it, the consequences in the data and the documents are the same.

More in "Which marketplace platform model should you choose".

2. Who is the merchant of record, flow by flow

Who is a party to the contract, who accounts for the tax, whose account at the payment provider the money lands in, and who issues the document. Four roles that can fall to different parties.

Changing this late means an amendment to every seller contract and a rebuild of invoicing. More in "What Is the Merchant of Record in a Marketplace?".

3. Whether you touch seller money at all

Either the payment provider splits the payment on their side and pays the seller directly, or you collect the whole amount into your own account and pay out in cycles. The second option almost always means holding third-party funds, which means a conversation about a license or about a licensed partner.

This question can move a schedule by quarters, so it gets asked before the quote rather than before launch.

Which decisions settle what you sell and how?

4. The catalog model and the matching key

Whether two sellers with the same goods land on one shared product page, or each of them gets their own listing. And what does the system use to recognize that it is the same product: the barcode, the manufacturer number, or something else.

This is the most expensive debt you can build, because it is created in the first month and you pay for it for years. At operators running for several years, backlogs of badly created products run into tens of thousands of items, and nobody is going to fix those by hand.

Changing the matching key after a year means reprocessing the entire catalog.

5. Who is the source of truth for each entity

Product, price, stock level, order, customer, money, document. For each of them, exactly one system is right, and the rest synchronize to it.

It sounds like technical housekeeping, and yet it is the most common cause of chaos, because usually nobody wrote it down and every team assumed something different. Settle it on a diagram, signed by both sides, before anyone starts coding.

6. The relationship between your own sales and the marketplace

Whether your own offer wins ties when the offer on the product page is picked. Whether you police sellers so that they are not cheaper than you. Who "owns" the customer.

This is a business decision with two kinds of consequences. Legal: price clauses come in two forms, wide (the seller cannot be cheaper anywhere, including on other platforms) and narrow (the seller cannot be cheaper on their own site).

In the EU, the wide ones have been carved out of the general exemption for distribution agreements and require an individual assessment, and some countries have banned both forms in selected industries. If your seller terms are going to contain any price clause at all, a lawyer has to see the exact wording rather than the intention behind it.

Political: the sales team and the marketplace team start competing for the same sales plan, and that is a conflict that does not resolve itself.

7. The order split rules

A cart with three sellers breaks into separate orders. But along which lines exactly?

By seller, certainly. Usually by shipping method as well.

Sometimes by the declared dispatch time or by warehouse. Every extra axis means a different number of parcels, different delivery costs in the cart, different documents, and a different course for the payout.

The buyer sees the result at checkout, so this is not a technical decision.

Which decisions settle how the marketplace runs day to day?

8. The moment you take the money

At order placement, at seller acceptance, at dispatch, at delivery. The choice looks like a detail, and it is a financial decision: it settles when you have the money, who carries the risk of a canceled order, and what happens when the card authorization expires before the seller ships the goods.

On a €1,000 cart shipped after ten days, that is the difference between "the money is there" and "the authorization lapsed and we have to ask the customer to pay again".

9. Who talks to the customer

You first, or the seller first. This settles the cost of the team, the response times you have to hit, the tooling, and the brand risk.

Settle the follow-on question right away, because it is the one that gets forgotten: can the seller contact your customer directly, and what do you do when they start using those messages to steer the customer to their own store? This happens everywhere, and it needs a rule rather than surprise.

10. Geographic and currency scope, and the order of countries

One country or several, one currency or several, in what sequence. Running in more than one country reaches into every area: taxes, shipping, law, translations, documents.

It is cheap to plan for and very expensive to bolt on later. Even if you launch in one country, it is worth knowing whether a second one is in the plan, because that changes the data model.

Which decisions settle who runs the marketplace?

11. The business owner and the operating model

Who answers for the marketplace result, where the team sits in the structure, and who they report to. A marketplace reaches into customer service, the stores, legal, complaints, sales, marketing, product data, and accounting.

Practitioners consistently point to the organizational change as the harder part of the project, harder than the technology. On top of that comes something almost nobody thinks about at the start: how you calculate a salesperson's bonus on a transaction where the company earns less than it does on its own sale.

Without an answer to that question, your own organization will effectively block the marketplace, and nobody will know why.

12. The definition of success and the phase gates

What it means for the pilot to have worked, and what you do if it does not. Without that, the project has no condition for finishing and no condition for scaling.

A marketplace with no investment decision behind it drifts until somebody shuts it down. More in "Why Build a Marketplace, and Which Metrics Prove It Worked?".

How do you spot a decision that was made by accident?

The most common problem is not that one of the twelve decisions never got made. It is that it was made by a default setting.

Somebody clicked whatever was in the configurator, and nobody noticed.

The test is simple and takes a minute per item. For each of the twelve, ask three questions:

1. Who made it? If nobody remembers, the installer made it.
2. Where is it written down? If it lives only in the system configuration, it is a setting and not a decision.
3. What would we do differently if the answer were the opposite? If nobody can answer, the decision was never considered.

The same test works as an audit of a marketplace that is already running, and it usually finds three or four items the team was convinced were settled.

What does each decision cost if you get it wrong?

What do these 12 decisions decide?

A quote made without these twelve is guesswork. Every one of them changes the scope of work, and several change the architecture.

A vendor who prices a project without asking these questions is pricing a license rather than an implementation.

The order matters. Decisions 1 to 3 are the precondition for a sensible conversation about 4 to 7.

You cannot design an order split without knowing who the seller is.

Three of them need a specialist from outside the company: the second, the third, and, in part, the tenth. Plan that time into the schedule, because a lawyer and a tax advisor do not answer in two days.

The remaining chapters of this series expand this list. Money, taxes, catalog, orders, organization: each of them goes deeper into one of the twelve items.

How do you work through the 12 decisions?

The simplest thing that works is one workshop, one day, twelve items, and a strict rule for writing things down.

1. Every item ends with one of three outcomes: settled, settled conditionally, or open.
2. "We do not know" is an acceptable answer, but only with an owner and a date. Without those two things, it turns into a default decision.
3. Write down the reasoning as well as the choice. In a year, someone will ask "why this way", and that question comes back with every larger change.
4. Bring legal, tax, and finance in from the start, rather than at the end for a sign-off. Three of these decisions are really theirs and not yours.
5. Close the ones that change the architecture before the quote: the platform model, merchant of record, the flow of money, the catalog model, the source of truth, geographic scope, and the business owner. The rest can mature during design.

Which mistakes do teams make with these decisions?

1. Treating this as a list of technical tasks

Eight of the twelve are business and legal decisions. The technical team can describe them, but it cannot make them.

And if nobody else makes them, the technical team will make them anyway, silently.

2. Settling them after the vendor contract is signed

By then, the scope is fixed, and each of these decisions can change it. This is the source of most of the disputes over contract amendments in this industry.

3. Recording the choice without the reasoning

In a year, nobody will remember whether "we collect the money ourselves" came out of a legal analysis or out of the fact that it was easier in the configurator.

4. Skipping the twelfth item because it looks obvious

The definition of success is the item on the whole list that gets skipped most often and comes back most often.

5. Assuming a decision, once made, is known across the organization

Practice shows otherwise: decisions live in emails and spreadsheets, and two departments can work off two different versions of the same matter for many months.

Which decisions need a specialist from outside?

It does not tell you how to answer these twelve questions. It tells you that you have to answer them deliberately and in this order. The answers depend on your model, your market, your category, and what you already have in the company.

Items two, three, and ten touch on law and taxes. Confirm them with a lawyer and a tax advisor, for the specific country and year.

This list is also not complete in the absolute sense: every vertical adds its own questions, and services, digital content, and rental have theirs. It is complete in the useful sense.

If you have these twelve closed, you are ready for the conversation about the quote.

Summary: How do you get through the twelve?

One workshop, one day, twelve items, and a strict rule for writing down both the choice and the reasoning behind it. Anything left open gets an owner and a date.

Bring legal, tax, and finance in from the start, because three of the twelve will not close without them. Talk to a marketplace expert if you want to work through the list against your own case.

This is where the Foundations block ends. The next article opens the largest part of the series: How money moves through a marketplace. It runs from the moment the buyer clicks "pay" to the transfer landing in the seller's account.