You have been paying somebody every month for account access, and at some point the thought arrives: the margin on this looks good, and you understand the product better than the person selling it to you. The thought is right about the margin and wrong about the product. What follows is what actually sits behind an account business once real clients are on it, what each piece costs to carry, and the route to a share of the same revenue without carrying any of it.
The business is credit and compliance, not accounts
The version in your head is simple. You source accounts, you mark them up, you collect monthly, and the work is a spreadsheet and a chat thread. That version is accurate for about 3 clients. It stops being accurate somewhere around 20, because by then you are not selling accounts at all. You are running a small credit desk with a compliance function bolted onto it, and every genuinely hard part of the job lives in those 2 things rather than in the account.
The account is the only part the client can see, so the account is what he thinks he is buying. What he is actually buying is what happens on the day it stops working. Every expensive thing in this business lands on that day, and the price you charge has to cover all of it out of the months where nothing goes wrong. Price it for the quiet months and the first bad week takes the year with it.
Somebody has to hold client money, and it would be you
On a prefunded structure, client money sits on your side of the line between the top up and the invoice the platform eventually issues for delivery. An advertiser spending $100,000 a month is spending roughly $3,300 a day, so a 7 day buffer parks about $23,000 of his working capital with you. Multiply that by every client you sign. It is not revenue. It is a liability with your name on it, and if you hold it in the same account as operating cash you will spend some of it by accident and only find out in a month you needed it.
Then the day it goes wrong. An account is disabled with a balance still sitting on it. The client wants his money, and the money is behind the same review that took the account down, so none of the timing belongs to you. You will refund some of those out of your own pocket to keep the relationship, before you know whether it ever comes back. Nobody models that line before their first one, and it is usually larger than the fee income for the month it lands in. The same arithmetic is set out in the breakdown of float against percentage fees, just pointed the other way. There it is your working capital parked with somebody else. Here it is everybody else's parked with you.
The trust you would be reselling is not yours to build quickly
Delivery is not decided by how tidy your setup looks. It is decided by history: verification status, a payment record, and spend that has already run and already been paid for without incident. That is why an account issued off a portfolio spending about $50 million a month behaves differently from an identical looking account that has never spent a dollar. Same interface, different priors, and the difference is not something you can configure your way to.
New supply has no history, by definition. You can buy assets sold to you as aged, and you find out 3 batches in how much of that was real, usually when a client steps a campaign from $200 a day to $5,000 a day and the account stops delivering at the number that matters. The mechanism is worth understanding before you spend money on inventory, because why account history changes what the platform lets you do is the entire reason this market has prices at all. Trust of that kind is bought with years of paid invoices at volume, and it belongs to whoever paid them.
Replacements are the product, and they do not keep hours
Strip the pitch back and the client is paying for 1 thing: when the account goes down on a Friday afternoon, another is issued and he is live again before weekend traffic. Delivering that means holding spare accounts that earn nothing while they wait. Reserve inventory is capital sitting still, sized against a ban rate you cannot know until you have run for a year. Undersize it and you break the only promise that matters. Oversize it and the margin you started this for is gone.
The clock is short, because the client measures you against how fast he was live the last time, not against how hard it was to do. What has to happen in the first hour after an asset goes down mid campaign is the standard you would be signing up to meet, at whatever hour it happens. Accounts stop at inconvenient times, and clients spread across timezones do not take turns, so you would be the escalation point at 2am, on a Sunday, on the day you are on a flight. His spend is stopped while he waits and every hour of it is money he can count.
Cover costs money before it saves any. A second person has to be trusted with client balances, has to know the difference between a policy rejection and an account level restriction, and has to be paid in every month where nothing breaks. Until you can carry that salary the business has 1 point of failure, and it is you with a phone. Here is what sits behind the promise once you are the one making it:
- Spare accounts sized for your worst week, not your average one.
- A supply relationship that still answers when your volume triples in a quarter.
- A way to move a pixel, a domain and a page across without restarting the client's learning phase.
- Whatever the disabling did to the client's balance, whether or not the platform ever releases it.
- A written replacement policy, because the first argument you have will be about what counts as a replacement.
You would be aggregating other people's compliance
Every offer your clients run becomes your exposure. A buyer pushing hard claims on a supplement page does not only affect his own account, because enforcement reads patterns across the things that connect to each other: payment instruments, domains, profiles, the people listed as admins. So you vet what somebody intends to run before you issue anything, you watch what he actually runs afterwards, and you make the call where you tell a paying client to change his page or leave. Most people who like the idea of this business do not like that call.
Vetting is judgment work and it does not get easier by hiring. The loud affiliate with the aggressive landing page is priced in. The quiet brand that changes its checkout flow in month 4 without telling you is not. What actually keeps a restricted offer running becomes knowledge you hold on behalf of everybody you sell to, and you answer for it in a way you never do when it is only your own money on the line.
Run the arithmetic before you buy inventory
Price the whole thing against the flat monthly rates the honest end of this market already publishes, because those are the numbers your prospects will hold you next to. The published tiers and the monthly spend each one is capped at take about a minute to read, and that is the whole document you would be competing against. 10 clients at $299 a month is $2,990 of gross revenue. 10 at $995 is $9,950. Now put the costs underneath it: idle inventory, the balances you eat, supply, cover for the hours you cannot work, and the accounts you write off entirely. The number that decides the business is your true cost per client after replacements, and you do not know it until you have replaced the same client 3 times.
Then run the test people skip. If you stopped answering messages for 14 days, what would be left at the end of it? Clients go quiet and stock goes stale, which means the revenue is attached to your attendance rather than to anything that keeps running without you. Put your own hours in at the rate you currently charge for media buying and see whether the margin survives contact with them. Often the honest answer is that you would be buying yourself a harder job and funding the deposit on it with your own working capital.
Who it works for, and the ceiling they still cannot lift
Some people should run it, and it is worth being specific about who, because the conditions are narrower than the pitch suggests. It works when the infrastructure is already a cost you carry, so clients are marginal load on something that exists rather than a new function you are standing up from nothing.
- You already spend across a lot of accounts yourself, so supply and monitoring exist whether or not you sell anything.
- You have capital that can sit still for 12 months, and a stuck balance does not change what you do that month.
- Somebody other than you can answer at 3am, and has been doing it long enough to be trusted with a judgment call.
- You can write off a client's balance without it deciding whether you make payroll.
The version that pays without the carry
Notice what is missing from that list. Even with all 4 conditions in place, the thing that decides whether the accounts deliver is not on it, because history sits on a portfolio somebody else has been paying invoices into for years. You can build the client facing half of this in a quarter. The half that makes the accounts work is not for sale on that timescale at any price, which is why nearly every reseller here is reselling somebody else's supply with a markup and a support burden attached. So the question changes. It stops being how to build this and becomes how much of the work you want to own for the share you would get.
The honest trade off first. Referring operators pays less per client than running the accounts yourself, and it should, because whoever carries the float, the idle inventory and the 3am escalation takes the larger share. You do not set the pricing, you do not write the terms, and you are not the one the client thanks when a replacement is issued in 1 to 2 hours. If the point was to hold the entire margin, this is not that, and dressing it up does not make it that.
What you give up in share you give up in exposure at the same time. No client money on your balance sheet. No accounts sitting idle waiting for a bad week. No conversation about a balance stuck behind a review, because it is not your review and not your balance. You introduce operators you already know, they get issued an account off a portfolio that already carries the history, and the recurring part is billed by whoever absorbs all of it. The partner programme and what it pays on sets the terms out in writing rather than on a call, which is the thing to read before you decide either way.
No structure makes an account immune to enforcement, and anybody telling you otherwise is selling you something. What a structure changes is who absorbs the cost on the day enforcement arrives: the idle stock, the stuck balance, the hour of the night it gets found, and the answer the advertiser is owed by morning. In the business you were sketching out, all of it lands on you. Decide whether you want to be paid for carrying that, or paid for pointing at somebody who already does.