Build or license a white-label client portal? The maths, honestly
What a client portal costs to build once you count the parts nobody sees, a worksheet to run the numbers with your own salaries, the eight questions that separate a real white-label licence from a themed reseller account, and the cases where building it yourself is the right answer.
- The visible half of a client portal, logins and file upload, is a few weeks of work. The half that decides whether it is safe to put customers on is the part nobody sees, and it is where first-time builds lose their year.
- Run the numbers with your own salaries, not ours. On the worksheet below, even a deliberately cheap team lands well above $15,000, and the build has a maintenance bill that never ends.
- "White-label" is not one thing. Eight questions, listed below, tell a real rebrand with no trace of the vendor from a theme with a footer credit.
- Build it yourself if the portal is the product you are trying to make your name on. Licence it if the portal is the thing your customers need so you can sell them something else.
A client asks for a private portal. You scope it on a call, someone says "it is just logins and a folder", and the quote comes back at a number that makes the whole idea look like a mistake. So it either gets built cheaply and badly, or it does not get built, and the client stays on email attachments and a shared drive.
This is a way to decide that properly: what the build actually contains, a worksheet to price it with your own numbers, how to tell a genuine white-label licence from a themed reseller account, and the cases where you should close this tab and build it yourself.
The part of the iceberg under the water
Everything a customer sees in a portal is straightforward. A login page, a list of projects, a place to open a file, a message box. A competent developer can put that on a screen quickly, and a demo of it looks finished, which is exactly what makes the estimate go wrong.
The cost is in the parts that only matter when something goes wrong, and in a product that holds other people's documents and money, something eventually does. None of these is a differentiator, and every one of them has to be right before you can put a paying customer on the system.
| The part | Why it is easy to get wrong | What it looks like when it fails |
|---|---|---|
| Tenant isolation | A filter in the interface is not isolation. It has to hold against a request someone crafts by hand, which means enforcing it below the screen, in the data layer. | One customer reads another's files. You find out from a customer. |
| File delivery | A public URL to a private file is a link that never dies. Files need short-lived signed links that expire and can be revoked. | A document shared in March is still open to anyone holding the link in December. |
| Payment webhooks | A browser redirect after checkout proves nothing. Status has to be settled by a signature-verified server event, handled so a replayed event cannot act twice. | A customer abandons a payment and still gets the files, or pays twice and is charged twice. |
| Versioned documents | "Which is the current one?" needs an answer the system gives, not one a person remembers. | Two people work from different revisions and each has a reasonable claim to be right. |
| An audit trail | Nobody wants it until a dispute, and it cannot be added after the fact: the events you did not record are gone. | You are asked who opened a file and when, and the honest answer is that you do not know. |
| Email on your domain | Invitations and receipts that land in spam, or arrive from someone else's address, undo the branded experience in one message. | Customers never receive their invitation, then blame your product. |
| Upkeep | Dependencies, security patches, browser changes and the gateway's own API changes continue for as long as the product exists. | A feature nobody touched for a year stops working on a Tuesday. |
Each row is a solved problem. Solving it again is not wasted effort so much as effort that earns you nothing a customer can see.
The figure we quote on the licensing page is nine to fourteen months of engineering before a first customer can safely log in. That is our estimate, not a measured industry statistic, and your team may be faster or slower, which is why the next section gives you a way to produce your own number instead of borrowing ours.
A worksheet you can fill in yourself
Costing a build with only the salary line is how the number comes out too small. These are the lines to include. Use real quotes from your own team or your usual contractors wherever you have them.
| Line | How to estimate it |
|---|---|
| Engineering | People × months × fully-loaded monthly cost. Fully loaded means salary plus the overhead of employing them, not the figure in the offer letter. |
| Design and front-end | A portal has many screens: login, onboarding, dashboards, documents, invoices, messages, settings, empty states and error states. Count them. |
| Security review | An independent look at isolation and file delivery before launch. Price it as a line, not as something you will get to. |
| Payment integration and testing | Gateway onboarding, webhooks, refunds, failed and duplicate payments, and test mode through to live. |
| Time to first revenue | Months of building multiplied by what a launched portal would have earned in each of them. This is the line most worksheets leave out and often the largest. |
| Maintenance, every year | A fraction of one engineer, indefinitely, plus the cost of being the person who is called when it breaks. |
| Opportunity cost | What those engineers would have shipped for paying clients instead. |
An example, with every assumption showing
These are assumptions chosen to be cautious, not market data, and you should replace every one of them. Take a three-person team, deliberately the low end of what a portal needs, building for nine months, which is the bottom of our own range.
| Assumption | Monthly cost per person | Build cost | $15,000 licence as a share of it |
|---|---|---|---|
| Higher-cost team | $5,000 | $135,000 | About 11% |
| Deliberately cheap team | $2,000 | $54,000 | About 28% |
3 people × monthly cost × 9 months. This leaves out the security review, the cost of the months before launch and every year of maintenance afterwards, all of which push the build figure up rather than down.
Even the cheap row costs several times the licence before you have paid anyone to maintain it. Add half an engineer for upkeep (6 months of a $2,000 salary is $12,000 a year, 6 months of a $5,000 salary is $30,000) and the gap widens every year. The licence fee is $15,000, one time, with 12 months of updates and support included.
Eight questions that separate white-label from themed
"White-label" is applied to everything from a full rebrand to a logo upload with a footer credit. The word tells you nothing. These questions do, and they are worth asking of any vendor, including us.
- Is there any attribution at all? A "powered by" line, a credit in the footer, a name in the page source or the meta tags. In a real white-label there is none, and your customers cannot tell the platform did not originate with you.
- Does it run on your domain, with SSL? Not a subdomain of the vendor and not a redirect through them.
- Does email send from your address? Invitations, password resets and receipts should carry your domain, your signature and your reply-to.
- Whose payment account takes the money? Your own gateway account, settling to you, or a vendor wallet that pays out later on their schedule. The second makes the vendor a party to your cash flow.
- Is there a revenue share, a per-seat fee or a commission? A low monthly price with a percentage attached is a different product from a fee you pay once. Do the sum at ten times your expected volume.
- Who holds the data, and where? Hosted on your infrastructure means customer data never passes through the vendor. Ask about data residency if your market has rules on where records are kept.
- What do you get if the vendor disappears? Ask what the licence says about continued use, and whether source access exists, and on what terms. It should be a straight answer with a price, not a promise.
- What is and is not in the price? Custom features, managed hosting, a second deployment and priority cover are the usual extras. You want them named before you sign, not discovered afterwards.
For reference, here is how the $15,000 licence answers the ones above. There is no attribution anywhere, your domain and email are configured as part of the deployment, payments run through your own gateway account, and there is no revenue share. It is hosted on your own infrastructure by default, with managed hosting available as an add-on. Source access is available but agreed at signing rather than assumed. The licence does not expire; what renews, optionally and annually, is the updates and support. All of it is set out on the white-label licensing page, including what is quoted separately.
What licensing costs you
The case for licensing is easy to make and easy to over-make, so here are the costs, stated plainly.
- You do not own the roadmap. The platform does what it does. If a customer needs something outside it, that is custom work, scoped and quoted separately, and it is not covered by the licence fee.
- You are depending on a vendor. For updates, for fixes and for the answer when something breaks. That is a real dependency, and it is the reason the questions above about support terms and source access matter.
- You are not building a technology asset. If part of your plan is to hold a codebase you wrote, a licence is the wrong route and you should say so out loud.
- You still carry the customer relationship. You are the one your customer calls. A licence gives you a product to sell, not a way to avoid supporting it.
When you should build it yourself
There are cases where building is plainly right, and a vendor who tells you otherwise is selling.
- The portal is the product. If you are a software company and this is the thing you are trying to make your name on, you want to own it.
- Your workflow is unusual enough that a standard platform would bend it out of shape. A shape built around clients, documents, approvals and payments fits a great deal. If yours is not that shape, believe it.
- You have the engineers and the portal is a good use of them. The worksheet above will tell you. If it comes out cheaper than a licence and the team has spare capacity, build.
- You need to own the intellectual property outright. A licence is not an assignment, and for some businesses that settles the question.
If none of those describes you, the more common situation is a business that already has the customers and the credibility, and is missing a way to serve them that does not need a year of engineering first. That is the case a licence is for.
How to test a platform before you pay
The nastiest build-versus-buy surprise is discovering the hard parts in month seven. The cheapest way to avoid it with a licence is to see the finished product before anything is signed. Some of what to check:
Ask for a live walkthrough, not a slide deck
Watch someone create a client, upload a document set, attach an invoice and see the files stay locked. If it cannot be shown working, it is a roadmap.
Try to break the isolation
Ask how one customer is kept from another's files and whether that is enforced below the interface. A good answer names where, not just that it is.
Follow the money
Ask where a customer's payment settles. If the answer includes a wallet, a hold period or a payout schedule, you have learnt something about the relationship before signing.
Read what is not included
A fee that means the same thing for every licensee is one with named exclusions. Ask for them in writing.
Get the timeline and the scope on paper
Scope, territory, branding assets and a delivery timeline written down and signed, before the clock starts.
If you would like to see the platform running, the licensing call is a live walkthrough, and we will tell you plainly if it is not a fit. If it is, the companion piece, selling a white-label client portal, works through packaging, pricing and how quickly a one-time fee pays back. If you serve architecture practices specifically, what a studio actually needs from a portal is the clearest account of the problem you would be selling a solution to.
See it running under your brand
A thirty-minute call: a live walkthrough, a written scope and timeline before anything is signed, and direct answers on source access, hosting and data residency. The licence is $15,000, one time, with no revenue share.
Questions, answered
The questions agencies and consultancies ask before they license a portal, in plain language.
Ask us anythingIt depends almost entirely on who builds it and how long it takes, so the useful move is to run the sum with your own numbers: people, months and fully-loaded monthly cost, plus a security review, payment integration and testing, the revenue you forgo while it is unfinished, and a fraction of an engineer every year for upkeep. As an illustration, three people at $2,000 a month for nine months is $54,000, and at $5,000 a month it is $135,000, before any of the other lines. Those are assumptions, not market data, so replace them.
A client portal you sell to your own customers under your own brand. In a genuine white-label, your logo, colours and typography replace the vendor's, the platform runs on your domain, email sends from your address, invoices carry your letterhead, and there is no attribution of any kind. Plenty of products use the word for something less than that, such as a logo upload with a credit in the footer, which is why it is worth checking each of those points rather than trusting the label.
Build it if the portal is the product you are trying to make your name on, if your workflow is unusual enough that a standard platform would distort it, if you have spare engineering capacity and your own worksheet comes out cheaper than a licence, or if you need to own the intellectual property outright. Licence it if the portal is the thing your customers need so that you can sell them something else, and you would rather launch in weeks than spend a year on parts no customer can see.
The AtelierLab licence is $15,000 as a one-time fee. There is no revenue share, no per-seat charge and no commission on what your customers pay you. It includes the complete platform, your branding applied throughout, deployment on your domain, handover and training, and 12 months of updates and support. The licence does not expire. After the first year, updates and support renew optionally and are priced annually. Managed hosting, custom feature work and a second deployment are quoted separately.
Source access is available and is agreed at signing rather than assumed, because it changes the shape of the licence and what support looks like afterwards. Raise it on the first call: the right answer from any vendor is a straight yes or no with a price, not a vague reassurance.
Two to three weeks from signed agreement to launch is typical, and the pace usually depends on how quickly your brand assets and domain access arrive. A more complex deployment, or one that needs a payment gateway in a new market, can take longer, and that is better said on the first call than discovered after you sign.