AMA — six years running payments for offers nobody else will touch. Ask.

I've been doing the payments side of high-risk offers since 2020. Not as an affiliate — as the person the affiliates call when the money stops. I've opened and buried something in the region of forty merchant relationships across crypto gateways, card-to-crypto rails, and two acquirers I will not name.

Why now: I've written a lot of long posts on this board that were really answers to questions nobody had asked yet. Easier to do it the other way round for once.

What I'll answer

• Anything about how a processor actually behaves once you're inside it
• Freezes, reserves, underwriting, why accounts die, how to make them die less
• Numbers, where I have them, with the caveat attached
• Mistakes. I have a lot of these

What I won't answer

• Which specific vertical I work in, or for whom
• Anything that amounts to "give me an account" — I don't sell access
• Requests to evaluate a specific offer's legality. Not my job and not my competence

I'll be answering in batches rather than live, so expect a day or two of lag. This will probably run for a while.

Pinned for the month. Keep the account-buying requests out of it.

Forty merchant relationships in six years. What's the median lifespan?

Median is about seven months. That's a bad number and I want to be honest that it's bad.

The distribution is what's interesting though. It's not a bell curve, it's two humps. A third of them died inside eight weeks — those are the ones where the underwriting was going to say no and just took a while to get round to it. Another cluster sits between fourteen months and three years. Almost nothing dies between month nine and month thirteen.

If an account is alive at nine months it's usually alive at two years, and if it's dead it usually died before you'd finished integrating. The middle is empty. I've never had a satisfying explanation for that shape and I've stopped trying to construct one.

seven months and you kept doing it for six years

The alternative was zero months. Yes.

What's the single biggest mistake you see people make?

Leaving money in the account.

It's not close. It isn't the exciting answer and everyone nods at it and then doesn't do it, so let me put a number on it. Across everything I've had frozen, held, reserved or lost over six years, the total is somewhere around 190k. The amount of that which was still recoverable was maybe 40%. And the amount which would never have been at risk at all if the balance had been swept the same day it landed is, by my reckoning, over 80% of it.

Everybody's mental model is "will this provider freeze me". The right model is "when this provider freezes me, how much is inside". You control the second one completely and the first one not at all.

The hosted-balance gateways make this worse by design, because parking is the default and sweeping is a setting. Cryptomus, NowPayments, all of them — the dashboard shows you a balance because a balance is a good thing for them.

Sweep on a schedule, not on a feeling. That's the whole answer.

This is correct and i learn it with 3,100 dollar. Cheap lesson compare to yours.

Sorry if this is a beginner question but — at what point does someone actually need to think about any of this? I do maybe 2k a month. Is this all irrelevant to me?

Mostly irrelevant, with one exception.

At 2k a month you are below the threshold where anybody is going to spend a human's time on your account, and the sensible thing is to pick whatever is easiest and get on with the product. Don't build a two-rail architecture for two thousand dollars.

The exception is the sweep habit, because that costs nothing to start now and is genuinely hard to retrofit later. Whatever you're using, set up the auto-forward to a wallet you control, today, at 2k. Then when you're at 40k you already do it and you never have the conversation the rest of this thread is having.

The other thing I'd do at your size is run the self-hosted option once, on a cheap box, just to have done it. Not as your main rail. As proof to yourself that you could.

Question you'll dislike: how much of your six years is transferable, and how much is just that you got lucky with which regulators were paying attention in which years?

I don't dislike it, it's the right question and my honest answer is that I don't know the split.

Some of it is obviously environmental. The 2021-2022 window was easier than anything since, and people who started then and quit in 2023 will tell you a story about their own judgement that is really a story about timing. I'd include myself in that for at least the first two years.

What I think is transferable: the sweep discipline, reading a terms page for the reserve clause rather than the fee, and understanding that the review happens on their schedule not yours. Those would have been true in 2015 and I expect them to be true in 2030.

What isn't transferable: any specific provider recommendation with a shelf life longer than about nine months, and every instinct I had about which jurisdictions were quiet. That second one has been wrong twice.

Do you have an actual number for what a rolling reserve costs you against the headline rate? Everyone quotes 10%/180 days and nobody annualises it.

I do, and annualising it is the correct instinct because the way it's quoted is designed to stop you doing that.

Ten percent held for 180 days on steady monthly volume means that in the steady state you are permanently short six months of ten percent — that's 60% of one month's revenue, sitting in someone else's account, forever, until you stop trading. On 50k a month that's 30k of your money that you will never see again while the relationship exists.

The cost of that isn't the 30k, it's what the 30k would have done. If your working capital turns over even twice a year at any real margin, the reserve is costing you more than the entire processing fee. I've run this on my own numbers three times and every time the reserve came out between 1.4x and 2.6x the headline rate as an effective cost.

And the release is the part nobody warns you about: it's rolling, so month seven you start getting month one back — but only while you're still processing. The day the account closes, that whole balance sits under whatever the exit terms say, which is usually another 90 to 180 days on top.

Caveat: this is all on merchant accounts with real reserve clauses. Most crypto gateways don't have one, which is a genuine argument for them that nobody makes.