Everything in this course has been building to one operational question. When your platform pays someone, where does the money come from at the moment it leaves, and when are you made whole? That gap, measured in seconds on one rail and days on another, is your funding model. Get it wrong and you are not running a payments business. You are running an unsecured lender that did not mean to.
We have covered the mechanics. Authorization is a promise, settlement is the cash. Gross rails move value transaction by transaction, net rails defer it to a cycle. Card authorization and card settlement are separated by days. RTP and FedNow are prefunded and final on arrival. Now we put the operator's hat on and ask how you fund all of it without lighting your balance sheet on fire.
The two ways to fund a payout
There are only two. You either prefund or you settle on credit. Every arrangement you will ever build is a blend of these two, applied per rail.
Prefunding means you hold money at the settlement point before you need it. On the RTP network, participants fund a joint account at the Federal Reserve, and your sponsor bank's position in that pool has to cover what you send. FedNow settles through a Fed master account or a correspondent's, again funded ahead of the send. Prefunding is the price of finality. Recall from the finality and Herstatt modules why instant rails insist on it. There is no clawback, so the money has to be real before it moves.
Settling on credit means someone advances the funds and you repay on the normal settlement cycle. This is the card world. Your acquirer or processor pays out merchants before interchange and network settlement fully clear, and a line of credit, your own or your sponsor's, absorbs the gap. Credit is cheaper on working capital because you are not parking idle cash, but it has a counterparty and a limit, and limits get tested at exactly the wrong moment.
The mismatch that defines the problem
Here is the trap the whole industry walks into. The reporting is real time. The money is not.
A card authorization confirms in under two seconds. Your user sees "paid." But card settlement lands T+1 or T+2, and a chargeback can pull it back for months. If you pay your merchant or seller out instantly off that authorization, you have fronted your own cash against money that has not settled and might never. That is the fast-payout working-capital trap, and it is the single most common way platforms blow up their treasury.
Float is not a bug you can engineer away. It is the structural gap between when money is reported and when it is actually settled and final. Your funding model is a decision about who carries that gap and who gets paid for it.
Instant payouts are a real product and customers love them. But every instant payout is you, the platform, lending your own balance sheet to cover a settlement that has not happened. Marketplaces offering instant seller payouts are fronting capital against unsettled card volume, and that capital scales linearly with growth. The faster you grow, the more cash the float consumes, and the gap is invisible on a P&L that only shows fees.
A worked example
Take a platform processing $10 million a day in card volume, offering instant payouts to sellers.
Card settlement runs T+2. So at any moment you are carrying roughly two days of volume, about $20 million, that you have already paid out but not yet been funded for. That $20 million has to come from somewhere. Either you hold $20 million of your own cash idle as a prefunded buffer, or you draw on a line of credit.
At a line cost of, say, 9 percent annually, financing a steady $20 million float costs roughly $1.8 million a year. If your platform margin on that volume is 60 basis points, you are earning about $21.9 million a year in fees on $3.65 billion of annual volume, and the float financing quietly eats about 8 percent of it. Now add a bad week. Chargebacks spike, a seller cohort turns out fraudulent, and the unsettled exposure you already paid out never settles. The line does not cover losses, only timing. That hole is yours.
The lesson is not "never offer instant payouts." It is: price the float into the payout fee, cap instant-payout eligibility by risk, and size your facility for the bad week, not the average one.
Intraday liquidity is the daily reality
Prefunded rails do not let you fund once and forget. RTP and FedNow run 24/7, so your settlement position drains and refills around the clock, including nights and weekends when the wholesale funding rails you would normally use to top up are closed. FedNow shipped a liquidity management tool precisely so participants and their funding agents can move balances into and out of the account outside standard hours. Treat that as core treasury work, not an afterthought.
Intraday liquidity management means knowing, hour by hour, how much funded position each rail needs, and never letting a prefunded account run dry mid-cycle. On Fedwire and CHIPS, the same discipline applies at far larger ticket sizes, where a queued payment waiting on liquidity can cascade. The smaller you are, the more this lands on your sponsor bank, which brings us to the arrangement most platforms actually live inside.
Sponsor banks and where the credit really sits
Most fintechs do not hold a Fed master account. You sit behind a sponsor bank, and your customer funds usually live in a For-Benefit-Of account on the bank's books, FBO all your end users. The sponsor is the regulated party at the settlement point. Its prefunded position and its appetite for extending you intraday credit are the real constraints on what you can move and how fast.
This is why the central-bank-versus-commercial-bank-money distinction from the previous module is not academic. When you prefund at the Fed through a sponsor, you are as good as final. When you rely on the sponsor advancing credit, you are exposed to that one bank, and your "instant" product is only as resilient as their willingness to keep the tap open. Negotiate the intraday credit line, the haircuts, and the funding deadlines as hard as you negotiate the per-transaction price. Those terms are your funding model.
A framework for choosing, rail by rail
You do not pick one model. You pick one per rail, and the same three questions decide each.
- Is settlement final and irreversible on this rail? If yes (Fedwire, RTP, FedNow), you prefund. There is no recourse, so the money must be there first. If no (card, ACH), credit is available because the cycle gives you time and recourse.
- What is the timing gap between pay-out and funded-in? Seconds means you carry almost no float but need standing prefunded liquidity. Days (card, ACH) means a financing decision: hold idle cash or draw a line.
- Who is best placed to carry the gap, and are they paid for it? You, your sponsor, or the customer through a fee. If nobody is explicitly paid to carry it, you are carrying it for free, and that is the silent leak.
Run those three questions across each rail you support and the model falls out. Instant rails: prefund, manage intraday, size the buffer for weekend peaks. Card and ACH: settle on credit, but reserve against chargeback and return risk and price the float into instant-payout fees. Wires: prefund, watch queue and liquidity. Net rails: hold settlement reserves sized to your worst projected net debit position.
Closing: where the course lands
We started with a promise and a payment and spent nine modules pulling them apart. Authorization is not settlement. Gross is not net. Reported is not final. Instant is not free. Every one of those distinctions was leading here, to the desk where someone has to decide whose cash covers the gap tonight.
A funding model is not a spreadsheet you build once. It is a standing position you manage every hour the rails are open, and they are open all the time now. The operators who survive are the ones who treat float as a real liability with a real cost, prefund what cannot be reversed, reserve against what can, and never confuse a real-time confirmation with money in the account. Money moving is the easy part. Funding the movement, on time, every time, without lending your balance sheet by accident, is the whole job.