When does a white-label AI client pay for itself?

At the wholesale guides published on our services page, a complete Top Tier buildout costs your agency up to $1,400 in setup and up to $460 a month, plus $0.35 a minute on the voice line. Sell it at $1,500 a month with nothing charged upfront and the client is cash-positive on the second invoice. Count what it cost you to find them and it is the third.

That is the whole answer. The rest of this is the arithmetic behind it, the part of the first sixty days where you are exposed, and the two variables that turn a fast payback into a slow one.

What does one client actually cost you in month one?

Two numbers move: the one-time setup fee you pay us, and the monthly. Voice minutes ride on top at $0.35 a minute for builds that take calls.

I am going to use the top of every published range, because planning against the best case is how agencies end up short in month one. Top Tier is a guide of $1,150 to $1,400 setup and $315 to $460 a month. I will take $1,400 and $460. For call volume I will assume 300 minutes a month, which is a small local business with a phone that rings a normal amount, so $105.

Month one, one client:

The $1,965 is setup plus the first monthly plus minutes. After that your cost of delivery is $565 and your margin is $935 a month, every month, without you touching the build.

The $49 platform subscription sits outside this table on purpose. It is your agency workspace, not a per-client cost, so it does not scale with the client count. One client's first month of margin covers it for a year and a half.

If your pricing call lands nearer the bottom of the ranges, at $1,150 setup and $315 monthly, month one is -$70 instead of -$465 and you are positive on the first invoice. Do not plan on it. Plan on the top and be pleased.

The number most people leave out

Nothing above includes what it cost you to get the client.

If you ran ads, bought a list, or paid a setter, that is cash. If you did it yourself, the twelve hours you spent on outreach, the demo call, and two follow-ups are still a cost, they just do not show up on a card statement. Price your own hour at something. At $75 an hour, twelve hours plus $500 of lead cost is $1,400 of acquisition, which is roughly the same size as the wholesale setup fee.

Add it in and the table moves:

• End of month 1: -$1,865
• End of month 2: -$930
• End of month 3: +$5

So the honest answer to when a client pays for itself is the third invoice, not the second. That is still a fast payback by any standard. It is fast because the monthly margin is large relative to the one-time cost, which is the specific thing that makes a wholesale model work and a build-it-yourself model not.

Write it down rather than wave at it. Acquisition cost is the only figure here you control, and the only one that varies tenfold between one agency and the next.

Where are you actually exposed?

The first sixty days.

If the client cancels at the end of month one, you are out $465 on delivery and whatever you spent finding them. If they cancel at the end of month three, you are up $1,405 and you keep it. The risk is not spread evenly across the year, it is concentrated almost entirely in the window between the setup fee clearing and the second or third invoice landing.

Which makes onboarding speed a cash-flow decision, not a service-quality nicety. Every week the build sits waiting on a logo, a calendar login, or a list of services is a week of monthly cost with a client who has not yet seen the thing work. That is the week they get buyer's remorse. We wrote the intake side of this up in what to collect before the build starts, and the short version is that intake happens once, under your brand, attached to the job, or it happens four times over three weeks.

On how likely early cancellation actually is, the useful reference point is what churn looks like by price band. ChartMogul's analysis of monthly customer churn across its customer base found that companies with average revenue per account under $25 a month had a median monthly logo churn of 6.1 percent, and that it fell to 2.2 percent as ARPA rose above $500 a month. Their 2023 retention report, built on data from over 2,100 SaaS businesses across 2022, found the same shape annually: top-quartile customer retention was 85.7 percent for companies with ARPA over $500 a month against 63.1 percent for those under $10.

Read that as a direction, not a forecast for your book. Those are self-serve software companies, not a service sold by a person to a local business owner. The pattern still points the right way. Things that cost real money and were sold by a human stay longer than things that cost $19 and got signed up for at 11pm.

The two things that break this math

Selling at a thin markup. Every number above assumes $1,500 retail against a delivery cost of $565. Sell the same build at $900 and your margin is $335 a month. The $1,400 setup now takes four months to clear, and eight or nine once acquisition cost is in. A client who leaves in month five and you have worked for nothing. The payback period is not really a function of the wholesale price, it is a function of the spread, and the spread is set on your sales call. The full margin breakdown by tier goes through what is left at each one.

Underpricing the minutes. The $0.35 is per minute, so it is the one line in your cost base that moves with something you do not control. Three hundred minutes is $105. An HVAC line in August that runs 1,200 minutes is $420, and if you sold a flat retainer, that $315 difference comes out of your margin in the month you can least afford it.

Worth knowing what you are comparing against here. Raw voice infrastructure is cheaper than $0.35. Vapi's own pricing page shows a hosting fee of $0.05 a minute with models passed through at cost, and its calculator estimates $82 to $129 a month for 1,000 minutes on default providers, on top of a success package that starts at $29 a month. Retell publishes a range of $0.07 to $0.31 a minute depending on which model and voice you pick, with its own infrastructure line at $0.055.

Those are the numbers for a platform account you configure, monitor, and debug yourself. At $0.35 you are also buying the build, the call flows, the transcripts, and someone else owning it when the agent starts fumbling transfers. That is worth money. It is not worth unlimited money, so on a client whose phone never stops, quote the minutes through at cost plus a margin or put a band in the retainer. Do not eat them.

Who should not run this model?

If you cannot put roughly $2,000 of cash behind a signed client and wait sixty days, this model will squeeze you before it pays you. Not because the margin is bad, but because it arrives after the cost does. Sign three clients in the same month and that is close to $5,900 out the door before the first invoices land, which is exactly the point where agencies that looked like they were winning stall out.

Two ways through it. Charge the client a setup fee that covers some or all of the wholesale one, which moves payback to day one at the cost of a slower close. Or start slower and let each client fund the next. We went through the case for and against a client-facing build fee in this post, and it is a real trade, not a right answer.

One piece of demand-side context while you decide how fast to push. The Census Bureau's Business Trends and Outlook Survey found that as of May 3, 2026, 19.8 percent of US businesses reported using AI, under 20 percent among firms of four or fewer employees against 37 percent of firms with at least 250. Most of the market you sell into has not bought anything yet. That is the opportunity, and it is why your acquisition cost is real. You are usually the first person explaining the category, not the third quote.

FAQ

What do I actually pay Lumina in month one for one Top Tier client? The published guide is $1,150 to $1,400 in setup plus $315 to $460 for the first month, plus $0.35 a minute for voice. Your $49 platform subscription is separate and does not change with client count. We agree every figure with you on a 30-minute pricing call before anything gets built, so you know the number before you owe it.

Does the platform subscription go up as I add clients? No. The $49 covers the agency workspace, demo site, client portal, fulfillment access, and onboarding help. Fulfillment is ordered and billed per client build.

What happens if my client cancels in month two? At the numbers in this post you are roughly $470 up on delivery and down whatever you spent acquiring them, so probably net negative. At month three you are ahead. The first sixty days are the exposure, which is the argument for getting the build live fast rather than perfect.

Are the wholesale prices fixed? They are guides. Where scope is standard the guide is one number, where it moves with the job it is a range, and custom builds are quoted on the call. We lock the final figures on your pricing call before the build starts. The services page is the live source, and it changes.

Should I charge my client a setup fee to cover yours? It depends on whether your constraint is cash or close rate. A setup fee removes the month-one hole entirely and slows some deals down. Monthly-only closes faster and asks you to front the build.

What to do with this

Before you send your next proposal, write down three numbers on one line: the wholesale setup, the wholesale monthly plus your honest guess at minutes, and the retail you are about to quote. If the gap between column two and column three is under $700 a month, the payback math in this post does not hold for that deal and you should either raise the retail or drop a service from the scope.

The first two numbers stop being guesses on the pricing call. If you want to run the exercise against a real deal, start an agency workspace at $49 a month, build the order in Fulfillment, and book the call. The FAQ covers what happens between the order and delivery.