What It Costs to Run a Gold Product Inside Your App

The commission rate is the question everyone asks first and it is the least interesting one. What decides whether this works inside your business is the cash mechanics: a partner buy debits your balance with us, not a payment gateway and not your user’s wallet, which means you are carrying float and settling it on a clock.

This is the cost side in full. There is a deliberate omission at the end, and it is explained rather than hidden.

The mechanic that actually matters

When a user buys through your integration, three things happen in an order that surprises people.

Your user pays you, on your side, by whatever means you already use. That part does not run through us at all.

The buy then debits your wallet with us. Not a payment gateway, not the user’s rupee balance. From our side you are the counterparty, and the metal is then held for the end user with our custody partners.

So you are intermediating, and that means float. How much float depends on which billing type you are on.

Prepaid. Your balance must stay at or above zero. You fund it in advance, and your spendable headroom is simply your balance. Simple, and it ties up cash.

Postpaid. Your balance may go negative, down to an agreed credit limit. Your headroom is your balance plus that limit, and dues settle T+2. Less cash tied up, and a settlement obligation on a two-business-day clock.

Neither is better in the abstract. A team with a large float and low volume will not notice prepaid. A team doing volume with thin working capital will care about very little else.

Reconciliation, before you need it

Every movement on that wallet writes an append-only ledger row carrying the balance after the movement. Not a mutable balance field that you have to trust, an immutable line per movement.

That matters at month end. Reconciliation against a mutable balance is an argument; reconciliation against an append-only ledger is arithmetic.

The debit is also atomic with order creation, in the same database transaction. There is no window in which an order exists but the wallet has not moved, or the reverse. If you have ever integrated something where those two drifted apart, you will know why that sentence is in here.

The fees, plainly

Commission on attributed buys is 1% on the Free tier and 2% on Pro. It is charged on our rate, not on whatever you have marked it up to, which matters if you use the markup below.

The Pro tier is ₹50,000, one time. What it buys you is the higher commission share and the ability to set a merchant markup.

Merchant markup is up to 5%, Pro only. That is your margin on top of our rate, and it is yours.

Sell-side commission is 0.25%, and only once monthly volume reaches ₹3,00,00,000. Below that, sells earn you nothing. Read that as a design signal: this is built around buying and holding, not trading.

Platform fee is 0.25% and GST is 3%, the statutory rate on precious metals in India.

Attribution is set once, is exclusive, and runs for 12 months. A user attributed to you stays attributed for a year. That is favourable to whoever gets there first and worth understanding before you assume a user can be re-attributed later.

Per-transaction ceiling defaults to ₹50,000, and daily send to ₹1,00,000. Those are defaults rather than hard walls.

What it costs in engineering

Two numbers that are routinely conflated, and they measure different things.

Approval for production keys is about one business day. That is our review of your submitted go-live wizard.

Integration effort is about three days on the hosted webview and about two weeks on the raw API. That is how long a typical team takes to build, not how long we take to approve.

Do not add them together and do not describe the integration figure as “from wizard to live key”. Sandbox access, meanwhile, is immediate on both paths, so none of that blocks starting.

The omission, and why

The brief for this article asked for what it costs and what it pays back. I am not going to give you the second half, and the reason is worth stating plainly rather than burying.

We do not have partners in production yet. Any payback figure, any worked ROI, any “typical partner sees X” would therefore be modelled rather than observed, and a modelled number presented next to real fee figures reads as though it were also real. Our own compliance ledger forbids inventing figures, and this is exactly the case it exists for.

What we can tell you is the shape of the arithmetic, which you can run against your own numbers better than we can: your revenue per attributed buy is your markup plus your commission share, against a cost of the Pro fee if you take it, the float you carry, and the engineering above. Whether that clears depends entirely on your volume and your existing user base, neither of which we know.

When there is a real number we will publish it and say where it came from. Treat any vendor in this category who gives you a payback figure without saying how many live partners it is drawn from as having answered a different question.

Where the real cost usually is

For most teams evaluating this, the largest line item is none of the above. It is the compliance review, because digital gold is not a regulated instrument and that review is not a formality. We wrote up what your compliance team will ask and the SEBI position behind it.

Start that in parallel with the build rather than after it. The engineering is measured in days and the review is not.

The short version

Commission is 1% or 2% on attributed buys, the Pro tier is ₹50,000 once, and the markup is yours up to 5%. Those numbers are easy to compare and they are not the decision.

The decision is float. A partner buy comes out of your balance with us, prepaid means funding it in advance and postpaid means settling T+2 against a credit limit. Work out which of those your treasury prefers before you argue about a percentage point.