Why affiliate commissions need on-chain settlement, not invoicing

There is a number that nobody in affiliate marketing talks about, because everyone is implicitly used to it. The number is the average days-in-flight on an affiliate commission — the time between the click that earned the commission and the moment that commission shows up in the affiliate’s bank account. Across the major networks, the answer is somewhere between thirty and ninety days.

Thirty to ninety days. For work that the affiliate finished in week one.

Njord’s settlement target is roughly three seconds. That is not a marketing line; it is what falls out of the protocol design once attribution lives on Solana instead of in a billing system. This post is about why that change matters, what it lets you actually do differently, and where the on-chain rail starts to break down so you can plan around it.

What the 90-day cycle is actually paying for

When a network tells you a commission will be paid 30, 60, or 90 days after the conversion, three things are happening underneath.

The first is reversal risk. The customer might charge back. The customer might cancel within the refund window. The merchant might decline the conversion after manual review. The network is keeping the money so it does not have to claw back from the affiliate.

The second is float. The network is sitting on the commission for the entire holding period, earning treasury yield on it. Across a network with billions in commission-in-flight, that float is meaningful revenue.

The third is batching efficiency. Affiliate payouts are expensive in legacy rails — international wires, ACH, PayPal, country-specific local rails. Holding for a month lets the network batch a hundred commissions into one payout per affiliate, which is one wire fee instead of a hundred.

All three reasons are real. They are also all artifacts of using legacy payment rails for a workload they were never designed for. On-chain settlement attacks each of them.

What on-chain settlement actually costs

A Solana attribution transaction costs roughly $0.00025. A USDC transfer from the campaign escrow to the affiliate wallet costs the same. The whole “release commission to affiliate” operation, end-to-end, is well under a tenth of a cent in chain fees.

That number is the thing that makes the rest of this analysis work. Once the per-transaction cost is below a thousandth of a typical commission, the “batch for a month to amortize wire fees” reason for holding evaporates. There is no fee to amortize. You can pay 100 commissions to 100 different affiliates as 100 separate transactions and the chain cost is still negligible.

Solana is the chain that makes this possible at all. The same workload on a chain with $0.50 per-transaction fees would invert: the chain fee would exceed the commission on small conversions, and the protocol would be forced back into batching.

The structural change: escrow before promotion

Removing the batching reason still leaves reversal risk and float as motives to hold. Njord’s answer to both is to move the holding from after the conversion to before.

A Njord campaign is not real until it has been funded. The campaign owner moves USDC into an on-chain escrow scoped to the campaign at creation time. That escrow is the campaign’s credibility — affiliates do not have to take “the budget is $50,000” on faith because they can see the escrow balance on Solana.

When a conversion is recorded, the contract releases the commission from escrow instantly. There is no separate “wait 30 days, then pay” step because the float window is already over: the company committed the funds when it created the campaign.

The reversal-risk problem is handled by the hold period, which is now a function of affiliate reputation rather than a one-size-fits-all month. A New-tier affiliate has a 7-day hold; a Verified affiliate has 3 days; a Trusted affiliate has 24 hours; an Elite affiliate is real-time. The reputation tier itself is on-chain history plus an optional stake. It is the protocol’s version of “we know who you are because we have processed thousands of clean conversions for you, so we can pay you faster.”

That tier system replaces a KYC desk with a programmable reputation account. The economic effect is that fraud-prone new participants wait while their history accumulates, and trusted long-term participants get paid on the conversion.

What this lets you build that you could not before

Three program shapes become viable at three-second settlement that did not previously make sense:

Micro-affiliate programs at scale. When a commission is $0.40, a 90-day hold is comical and a wire fee would dwarf the payout. With Njord, you can run an affiliate program where every podcast review, every TikTok mention, and every blog backlink earns a small commission that lands in the creator’s wallet that day. The math works because the per-attribution chain cost is $0.00025 and the per-commission infrastructure cost is approximately zero.

Agentic AI referral flows. If you ship an agent that browses, recommends, and purchases on a user’s behalf, you want to be able to record affiliate attribution for the recommendation step and settle the commission to the agent operator’s wallet on conversion. The legacy stack cannot do this because the agent operator is a non-human entity that does not fit any KYC template. The Njord stack does not care: the wallet is the identity, and the tier system replaces the KYC step.

Real-time campaign optimization. When commissions settle within seconds, the company can see ROI on a specific creator within minutes of the campaign going live. The feedback loop on “this creator’s audience converts, this one’s does not” tightens from quarterly to same-day, which changes the kind of campaign-level experimentation that is worth running.

Where on-chain settlement is not the right answer

Honest section, because the protocol does not solve every problem.

Refund-heavy categories. Physical goods with 30-day return windows do not naturally fit a 3-second-payout system. Njord is built for digital products and services — SaaS, apps, subscriptions, digital content, on-chain assets — where there is no return-shipping window. You can run a physical-goods campaign with a longer hold period, but you are eating into the latency advantage.

Disputes that require human review. If a category needs editorial brand-safety vetting or human dispute resolution, the protocol’s automated tier system and stake-backed challenge window are a partial solution at best. Networks like Impact and CJ have real account-management value here.

Reach. The incumbents have already built publisher catalogues of hundreds of thousands of vetted affiliates. Njord does not. Building that catalogue is a separate, slower process from building the protocol; the network effects compound differently for an open protocol than for a managed network, and we are honest that it will take years to match.

The math, summarized

On the legacy stack, the average creator earns commissions today and sees them in a bank account 30–90 days later, after a 15–30% network cut. The affiliate’s effective income is delayed and reduced.

On Njord, the same conversion lands as USDC in the affiliate’s wallet in roughly three seconds, after a 2.5% protocol cut and a 1% bridge cut. The affiliate’s effective income is immediate and 96.5% of the gross commission.

That is the whole pitch. Everything else — the bridge SDK, the tier system, the campaign escrow, the governance parameters — is the apparatus needed to make those two sentences true.

The protocol is open on GitHub. Devnet is live. The mainnet program ID is reserved. If you operate an affiliate program today and any of the math above looks like an unlock for the kind of campaigns you could not previously run, the integration starts with one Anchor program call and one USDC deposit.