Build or buy your D2C stack

The ten parts of a D2C stack

Every web shop, bought or built, has to do the same ten jobs. The question is which ones you want to own.

Two rows stay with you whatever you choose. Item delivery runs on your game server in every model, and you remain responsible for what your game sells and how it presents prices. The rest can move.

What building in-house involves

Building the storefront is the smallest part of the job. The work sits behind it.

Accounts come first. A web shop needs a sign-in that works outside the app and a way to open the shop from the game with the player already signed in. In the US, the court orders in Epic v. Apple bar Apple from restricting “dynamic links that bring consumers to a specific product page in a logged-in state,” so the link from game to web shop is yours to design. If your game has no account system that works on the web, that is the first project, and it has to exist whether you build the shop or buy it.

When a payment succeeds, the provider notifies your server, and your server credits the items. A retried notification must not credit twice, and every credit needs a log that ties the payment ID to the item, the time and the player’s later use of it. That log is your evidence when a player disputes the charge. Apple adds one rule for iOS players: guideline 3.1.3(b) lets them spend web-bought consumables in the game only if the same items are also sold through in-app purchase.

Payment methods come one contract at a time. Appcharge, a web shop provider, reports from its own transactions that in the US the top three payment methods take about 75% of web store revenue, while in Europe no single method passes 30%. Each local method has its own price: on Stripe’s US pricing, Klarna costs 5.99% + 30¢, iDEAL $0.80 and Pix 2%. Covering Europe, Brazil and Japan properly means integrating and paying for methods one by one, and deciding whether to route between more than one processor.

Tax, fraud, chargebacks and consumer law are the expensive part. If you build the whole stack, you are the seller, with every registration, filing, dispute and consumer-law duty that comes with it. Merchant of record vs payment service provider lists each one and prices it at $1M, $10M and $50M a year.

Provider payouts rarely match a single order report. FastSpring, for example, warns sellers that its monthly statement total “will not match a single bank deposit.” When you build, you also build the pipeline that matches orders, refunds, chargebacks, fees and currency conversion to what arrives in the bank.

People are the ongoing cost. We found no public benchmark for how many engineers or finance staff an in-house web shop needs, so we won’t invent one. The calculator’s labelled defaults for running payments yourself are $45,000 a year for half of one person’s time on operations and disputes, and $30,000 for tax advice and filings. Replace both with your own figures. In the Appcharge and GDC survey, 54% of mobile studios said they had no staff dedicated to D2C.

What buying gets you

A web shop provider supplies the storefront, catalog tools, payment methods and, usually, a merchant of record checkout. You integrate your accounts and your delivery endpoint, load the catalog, and the provider is the seller. The fee replaces the tax, fraud and dispute work above, and most of the payments engineering.

Published prices vary in how much they include. Coda publishes 5% per transaction for Coda Webstore, says its pricing “includes handling of local taxes such as VAT or GST,” and says there is no extra cost for its payment methods. It does not say whether Codapay’s processing fees stack on top. Paddle publishes 5% + 50¢ for its checkout. Xsolla and Appcharge publish no merchant of record rate, and Stash’s site says pricing “depends on your transaction volume.” The provider fees table keeps the published figures side by side.

What you give up when you buy:

• Control of the checkout. Stripe’s merchant of record product, for example, accepts only Stripe-hosted Checkout or Payment Links, does not support custom domains, and sends receipts under the Onelink name.
• Some control of refunds and pricing. The contract may let the provider refund players or change prices without asking. The merchant of record guide lists the clauses to check.
• A margin on every sale for as long as you use the provider, including at volumes where doing the work yourself would cost less.
• Ease of leaving. Switching providers means moving players’ saved payment methods, purchase history and open subscriptions. Ask how each one moves before you sign.

Who builds and who buys

The public examples are all large publishers, and they go both ways.

Take-Two’s quote comes from a transcript published by The Motley Fool, not from a company filing; Take-Two has not published a D2C revenue figure. Playtika’s 3% to 4% is its own statement in an SEC filing, and it is the clearest public figure for what payments cost a large publisher that runs its own platform.

EA’s choice is the one to study if you are a large publisher that buys. As an authorized reseller, Coda is the seller the player pays, which is how a merchant of record works. EA keeps the game, the items and the player; Coda carries the sale.

The hybrid options

Most of the decision is where to draw the line between your code and the provider’s. Four set-ups are common:

The second row is the usual middle ground. You keep the parts players see and the data on what they buy, and hand the seller’s duties to the provider. It depends on the MoR offering an API or embedded checkout, and not all do.

Questions that decide it

1. Do your game accounts work outside the app, and can your server credit a web purchase safely? If not, start there. You need both whether you build or buy.
2. What will your web sales be in year one and year three? The worked comparison at list prices shows running payments through Stripe beats Paddle’s list price only while your own tax and payments costs stay under about 2.2% of web sales: $22,000 a year at $1M, $220,000 at $10M. Our judgment is that below about $10M a year few studios can run the seller’s duties within that budget. The calculator shows where your own numbers cross.
3. Which countries matter? Each new market adds payment methods to integrate and, if you are the seller, a tax registration.
4. Who in your company would own tax filings, disputes and reconciliation, and do they have the time today?
5. Which parts of the player experience do you need to control: the storefront, the offers, the checkout page, the receipts? Rule out providers that cannot give you those.
6. How would you leave? Ask each provider how saved payment methods, purchase history and subscriptions move to another provider, and get it in the contract.
7. How fast do you need to launch? If the answer is this quarter, buy the seller’s parts now and revisit at the volume where building pays. The launch checklist sets out the first 90 days.