Industry
The features shipped. The multi-tenancy, billing and permissions did not.
The unglamorous middle of a SaaS product is where builds stall: tenant isolation, subscription states nobody mapped, and roles that were a boolean until the first enterprise deal.
Failure patterns
What goes wrong here specifically
Not generic software problems. These are the ones that recur in this sector.
Multi-tenancy decided too late
Tenant isolation is a schema decision. Retrofitting it means touching every query in the application, and it usually surfaces the week a customer asks a security question.
Billing states nobody modelled
Trials, upgrades, downgrades mid-cycle, failed payments, dunning, refunds. Each is a state, and the ones nobody modelled become support tickets and revenue leaks.
Roles that started as a boolean
is_admin works until the first customer wants three permission levels and an auditor. Then it is a migration across the whole product.
Onboarding is where churn happens
Most SaaS teams instrument the funnel and not the first session. The first ten minutes decide retention and are usually the least examined part of the product.
What we build
Systems we have shipped in this space
- Multi-tenant architecture with isolation you can demonstrate to a buyer
- Subscription and billing flows with the awkward states modelled
- Role and permission systems that survive an enterprise contract
- Usage metering and quota enforcement
- Admin tooling so support does not need an engineer
- Onboarding instrumented properly, so churn has a cause
Constraints
The rules this sector plays by
Compliance, data handling and integration reality. These are architecture decisions, not a checklist at the end.
Data isolation you can prove
Enterprise buyers ask how tenants are separated and expect a specific answer. "Every query filters by tenant" is not one, because it is a convention rather than a guarantee.
Security questionnaires
They arrive with the first serious deal and are half documentation of controls you already have. Knowing which half is real is the useful part.
FAQ
Questions people actually ask
We are pre-revenue. Is multi-tenancy premature?
The schema decision is not, and it is nearly free to make correctly on day one. The tooling around it can wait. Getting this backwards is one of the most expensive avoidable mistakes in early SaaS.
Can you take over billing from Stripe Checkout?
Yes. The usual work is not the payment call, it is modelling the states around it — proration, failed payments, dunning, plan changes mid-cycle — which is where the revenue leaks are.
Do you have SaaS work we can look at?
Yes — Alula, a care coordination platform, and ProLyrics, an AI songwriting product with a real organic traffic figure attached. Both case studies are on the work page.
How long to get enterprise-ready?
It depends what the first enterprise buyer asks for, which is usually SSO, audit logs, role granularity and a security questionnaire. Assessed in an audit, that scope is typically four to eight weeks.