Most articles about this business tell you what to charge. Almost none tell you what you are going to pay, which is the number you actually need before you can decide anything.
So here it is, with real figures. Some of them are ours, because we run the platform side of this; the rest are published prices anyone can check. Where a number depends on decisions you have not made yet, I have said so rather than inventing an average.
The platform is the cheap part
This is the thing people get wrong in both directions. They either assume a white-label platform costs thousands a year, or they assume that once they have one, everything else is free. Neither is true.
A self-hosted platform licence is bought once. Ours is $25 on CodeCanyon and includes the full source code, lifetime updates and 1,000 AI credits to try the assistant with. Hosted competitors charge monthly instead — reseller tiers at the better-known vendors sit around $599 a month, which is a different shape of business and worth comparing properly rather than by headline price.
Either way, the licence is not where your money goes. It is roughly what you would spend on lunch for the week, and after that it stops mattering.
What you pay to keep it running
A server. The platform needs PHP 8.2, MySQL, OpenSSL and about 2 GB of RAM. That is shared hosting or the smallest VPS at any provider — you do not need a dedicated machine. Budget $5 to $15 a month depending on how much you care about the control panel. You also need two domains or subdomains with SSL certificates, and Let's Encrypt handles those for nothing.
A domain. Ten to fifteen dollars a year. Buy the one your clients will see, not a clever one.
GitHub. Free. The apps' code lives in a repository that the platform creates for you.
Codemagic. This is the one people expect to be expensive and it usually is not. The free tier includes 500 build minutes a month, which in practice covers around 200 app builds. Unless you are rebuilding constantly, you will not reach it in your first year. Above the limit, Codemagic bills you directly at their published rates.
AI credits. Only if your clients use the assistant. Every licence starts with 1,000 credits, a $5 value, which is enough to configure several apps and see whether it earns its place. After that you top up by card when you want to. This is genuinely pay-as-you-go, and the whole platform works without it — every setting the assistant touches can be set by hand.
Running total so far, for a year: the licence once, somewhere around $100 of hosting and a domain, and nothing else mandatory. Call it under $150 to have a working, branded platform on the internet with your prices on it.
Then the store accounts, which are not optional
Here is the part that surprises people, and it has nothing to do with which platform you pick.
Apple Developer Program: $99 a year. Per account, recurring, forever. No way around it — not with a converter, not with native code, not with anything. If anyone tells you otherwise they are describing TestFlight, which is not publishing.
Google Play Developer: $25, once. Cheap, but new personal accounts now have to run a closed test with real testers for a sustained period before they can publish publicly. That is not money, it is calendar time, and it will be the longest thing on your schedule for a first launch. Start it the day you decide to do this.
Now the decision that actually moves your costs: whose accounts are these?
- The client's own accounts. They pay Apple the $99, the app is published under their company name, and if they leave you they keep it. Costs you nothing, and it is the honest arrangement. It also means walking a non-technical client through Apple's enrolment, which takes a phone call and some patience.
- Your accounts. You pay $99 per client per year out of your margin, the apps live under your name, and switching provider is painful for them. Faster to set up and better for retention, but you have just created a recurring cost that grows linearly with your client list, and an awkward conversation for the day somebody wants to leave.
Neither is wrong. But choose deliberately, and price for it — this is one of the things that quietly decides whether your pricing works.
The cost nobody puts in the spreadsheet
Everything above adds up to a few hundred dollars. Your real investment is a few weeks of attention, and it falls in three places.
Learning the publishing pipeline. Not the platform — the stores. Signing keys, provisioning, privacy questionnaires, data-safety forms, age ratings, the reasons Apple rejects things. The first app teaches you all of it badly and the second one takes a fifth of the time. Budget a frustrating first week and stop expecting it to be smooth.
Building the thing you sell with. A demo app in the niche you are going after, with plausible branding, installed on your own phone. This is the single highest return use of your first week, and most people skip it in favour of a website nobody reads.
Support, forever. This is the one that compounds. One client is nothing. Ten clients is a few messages a week. Forty clients generates a steady trickle of "the app won't open", "can you change the colour", "Apple sent me an email", and that trickle does not scale with your prices unless you plan for it. The operational side of this business breaks there, not at the technical level.
A worked first year
One person, three clients, their own store accounts, modest everything:
- Platform licence, once — $25
- Hosting, twelve months — about $90
- Domain — about $12
- GitHub, Codemagic, OneSignal, Let's Encrypt — $0
- AI credits beyond the included 1,000 — $0 to $30, entirely up to you
- Store accounts — $0, the clients hold their own
Under $160 for the year. If you carry the store accounts yourself instead, add $99 per client and the picture changes quickly: three clients takes you to about $460, and ten clients to roughly $1,100 — still small against what ten clients should be paying you, but no longer a rounding error.
Where this gets expensive, and it is not infrastructure
If this business goes wrong financially, it will not be because hosting cost too much. It will be one of these:
- Underpricing the setup. The first app costs you a week. Charging a few hundred for it means working below minimum wage and resenting the client by month two.
- Clients who should never have bought. A business whose customers never return will not renew, and you will have spent the setup cost to acquire churn. There is a two-minute test for this and it is worth running before you quote.
- Unbounded support. "Free support" without a definition is a subscription you pay to your client. Write down what is included.
- Store accounts you took on without pricing them. See above. $99 a year each, forever, is a real line item.
What to check before you spend anything
Three things, in this order, and all of them before the licence:
- Do you have a first client in mind? Not a market — a business whose name you can say out loud. If not, that is the problem to solve first, and no platform fixes it.
- Can your hosting run it? PHP 8.2, MySQL, cron, two subdomains with SSL. Most shared hosting can; check before you buy rather than after.
- Are you willing to learn the store side? This is the honest filter. The platform removes the code, not the paperwork. If Apple's enrolment screens make you want to close the laptop, this business will be a grind regardless of what you build with.
The money is not the barrier here. Under two hundred dollars gets you a working platform with your name on it, and the first client pays that back several times over. The barrier is the two weeks of learning and the willingness to keep answering emails in year two — which is exactly why the people who do push through end up with something worth having.