SaaS: launch cheap, learn fast

Not every SaaS idea takes off — most do not. My job is to make your first launch cheap enough that you can afford to be wrong, and fast enough that you find out in weeks, with real users, not in a pitch deck.

With me you start fast, because the foundation already exists: marketing pages and CMS, languages and SEO, authentication and roles, billing and payments, an admin back office, notifications, integrations and a Docker deployment that runs on a small server. Fluss and a set of proven frameworks provide the base; my network provides the first testers and customers. A “six-month MVP” becomes a few weeks, and a hypothesis becomes a number.

How fast have I taken products online?

What is already there on day one?

• Marketing site and CMS with languages, SEO, forms and lead capture
• Authentication, roles, permissions and multi-tenancy
• Billing: subscriptions, trials, renewals, dunning; payment gateways
• Admin back office with dashboards, notifications (e-mail, Telegram) and audit trails
• Integration plug-ins for payments, accounting, delivery, registries and AI providers
• Docker deployment, TLS and monitoring on a server that costs less than a lunch
• A network of founders, agencies and first users to put the product in front of

B2B and B2C: both in production

Fluss is B2B SaaS: tenants, plans, per-customer pricing, marketplace, an agricultural B2B marketplace with state-registry integrations running on it. Jobsucher and Polymarket Automation Tool are B2C: subscriptions, onboarding, usage limits, real-time data. The playbook is the same — the base is reused, the hypothesis is what changes.

How would we work?

• Week 1 — hypothesis, scope lock, the one metric that decides
• Weeks 2–4 — launch on the reused base: real domain, real payments, real users
• Weeks 5–8 — measure: activation, retention, unit economics, cost to serve
• Then decide: double down, pivot, or stop — and I will say “stop” when the numbers do

Which SaaS ideas should not be built?

The ones whose first version cannot be tested in eight weeks with real payments. If the hypothesis needs a year of development before a single customer can say yes or no, it is not a SaaS bet, it is a product company, and it needs different money. I will say that on the first call rather than after the first invoice.

What happens after launch?

Weeks five to eight are measurement: activation, retention, cost to serve per customer, and whether the one metric from week one moved. Then a written decision: double down, pivot, or stop. If we stop, you own a codebase and a data set that the next attempt can reuse, and you spent weeks, not quarters.

What does a first version cost?

A quote after a call, because it depends on how much of the reused base fits and how much is new. What I can say blind: the infrastructure runs on a server that costs less than a lunch a month, there is no platform licence, and the first invoice covers a launch, not a plan.

Which stack, and why?

Laravel for the API and the back office, Vue rendered on the server for the storefront, MySQL or PostgreSQL, Docker for the whole thing, Elasticsearch or Meilisearch where search matters. It is not the newest stack and that is the point: every problem in it has been solved publicly, hiring is easy in every European city, and a first version does not depend on a framework that may not exist in two years. Fluss, Jobsucher and this site run on it.