Skip to content
One-time build · Source code handed over · No mandatory retainer

Stop renting software. Build once. Own it for good.

Your SaaS stack costs money every month. And at the end none of it is yours. FW Delta builds websites, tracking, automation and internal systems as infrastructure you own: a one-time build, the full source code, no mandatory retainer.

Discuss a project What we build
Up to 60–75 % lower three-year cost in suitable scenarios Source code handed over Hosted on infrastructure you own
up to 60–75 %
Cheaper than SaaS
over three years, in suitable scenarios and depending on the stack you start from
1×
Pay once
no renewals, no seat fees
100 %
Your code
full source code and Git access, handed over when the project closes
None
Seat fees
what remains is the infrastructure, from a few euros a month depending on the system

When renting gets expensive

SaaS is excellent for commodity software. It gets expensive on core processes.

We do not claim that building it yourself is always better. We build exactly the systems where ownership makes more economic sense than renting. These three points decide it.

Year 5

The fee does not end

A subscription costs the same in year five as in year one, usually more. A one-time build mainly produces infrastructure and maintenance cost afterwards instead of recurring licence and seat fees. Whether that adds up depends on term, user count and volume. That is exactly why this site carries a calculator and not a blanket statement.

Export

What is left after cancelling

On many platforms the logic, the configuration and the execution environment stay tied to the vendor. What you can usually export is the data, not the process. For commodity software that is bearable. For a process your business runs on it is a risk somebody should assess.

665+

SaaS products discontinued since 2015

Shut down: Google Reader, Parse, Fabric. Acquired, with consequences for pricing and roadmap: Segment, Tableau. Every tool you depend on is somebody else’s exit strategy. When they pivot, you migrate, at your own cost.

The full SaaS Graveyard Run your own numbers
Three areas

What we build

Every system is a one-time build, run on infrastructure you own and handed over with the full source code.

Measurement

Tracking, attribution and first-party data

When shop, website and ad platforms report different numbers, every platform optimises against a different picture. We build the measurement architecture so that one order stays one order.

  • GA4 and Google Ads
  • Google Tag Manager
  • Shopify and Consent Mode v2
  • Server-side and Conversions API
See measurement

Automation

Recurring processes without task and seat fees

Automation platforms bill per task. At low volume that is cheap; at high volume it dominates the bill. We build the cases where that flips.

  • n8n and custom workers
  • API and webhook integration
  • CRM, ERP and legacy systems
  • Internal approval and reporting flows
See automation

Software

Tools you own instead of permanent platform cost

Off-the-shelf software covers most of the requirements. The remainder, the part that sets you apart, is rarely on somebody else’s roadmap. That remainder is what pays off as your own system.

  • Dashboards and reporting
  • Internal portals and tools
  • Websites on Astro
  • Custom applications
See software
See all services

Technical assessment, not legal advice. Whether a specific setup is GDPR compliant depends on the data processed, the recipients and the legal basis.

Where to start

Start small.

Nicht jedes Unternehmen braucht sofort ein eigenes System. Unser häufigster Where to start ist ein klar abgegrenztes technisches Problem, das sich in einem Auftrag lösen lässt.

Fix the tracking

A clearly scoped engagement with a real test purchase as the acceptance step. Usually done within a few working days.

View

Automate one workflow

A single process, built cleanly and documented, instead of rolling out a platform.

View

Run the numbers on your SaaS

Numbers first, decision second. The calculator uses your user count and your term.

View

From the work

Problem, work, outcome

Three anonymised projects from different industries. What did not work before, what was built, and what was verifiable afterwards. Anything we cannot evidence does not appear here as a number.

Measurement Online fashion retail · DACH · anonymised
Situation
Only about half the shop purchases were reliably visible in analytics. The main site and the shop ran on separate platforms through separate measurement paths, and the order confirmation counted again on every reload.
Work
The measurement plan was rewritten, one container now serves both domains, consent signals and transaction IDs were unified, and deduplication was verified by repeatedly calling the checkout confirmation.
Outcome
Purchases arrive in analytics and in the ad account with value and a unique transaction ID. Acceptance was a real test purchase; documentation was handed over.

No ROAS or CAC figures are published for this case.

See the matching service
Automation B2B software · sales · anonymised
Situation
Enquiries came through a form, a CRM and a shared sales inbox. Duplicates appeared daily and nobody could say which channel had produced an enquiry.
Work
One workflow became the single delivery path: deduplication on a stable identifier, channel origin stored on the record, a log per delivery, and defined retries on failure.
Outcome
Every enquiry lands in the CRM exactly once and can be traced back to its source. Code, configuration and an operating guide sit with the client.

No revenue or close-rate effect is claimed; what was measured is delivery.

See the matching service
Dashboards Technical wholesale · reporting · anonymised
Situation
Reporting ran on a per-seat model. Anyone who only wanted to read the numbers needed a licence, so most people read forwarded spreadsheets in varying states instead.
Work
An application of their own directly on the existing database, with roles for reading and editing, defined metrics and a visible data cut-off on every view.
Outcome
Everyone sees the same number at the same cut-off. The per-person licence is gone; in return the business carries the server and the maintenance.

Whether that is cheaper depends on the user count. The calculation is shown openly in the Tableau comparison.

See the matching service
More project examples: four tracking cases

SaaS disappears when you stop paying. Code you own stays.

Code you own can become a long-term technological asset of your company. No single vendor then controls the price, the roadmap or your access to a core system.

The full source code

A complete Git repository at handover. Every line, every configuration, every deployment script.

Balance sheet treatment

Depending on the accounting standard and the specific use, certain development costs may meet the criteria for capitalisation as an intangible asset (IAS 38). SaaS fees are operating expenses. The accounting treatment has to be confirmed with your accountants or tax advisers.

Free to change

Your team, any agency, any freelancer can change it without asking us. No vendor approval, no API limits, no platform tie to us.

Less vendor dependency

No single SaaS vendor controls the price, the roadmap or your access to a core system. Dependencies on cloud, frameworks and operating systems remain, and maintaining the stack afterwards sits with you or with us.

How we work

From assessmentto handover.

Fixed price before the project starts. Timeline in the contract.

01
30–60 min

Technical assessment

We look at where things stand, name the actual problem and say so plainly when an off-the-shelf product is the better answer.

Technical proposal with a cost breakdown
02
2 working days

Fixed price

One number before the project starts. No hourly rates, no scope creep. The payment plan follows the project size and is set out in the contract.

Signed contract with a timeline
03
depends on scope

Build and test

Weekly updates and full staging access. You see every commit, every change, every decision.

A working staging environment
04
3–5 days

Handover

Production deployment, complete code handover, a walkthrough for the team and 90 days of support.

Production system, code and documentation
Research

We publish our methodology.

Original reports on SaaS economics, lock-in and AI engineering. Public sources, disclosed methodology, corrections documented. No sponsored content.

All reports
Research August 2026

Tracking Engineering

Web Measurement Reliability & Recoverability Report 2026

40 sources cited
Research July 2026

AI Engineering

AI-Assisted Software Delivery Evidence Review 2026

17 sources cited
Research June 2026

Software Supply Risk

Source-Available License & Governance Risk Report 2026

21 sources cited
Methodology and data basis SaaS Graveyard: discontinued products

Frequently asked

Five questions that come up before every project.

What does a system of your own cost?

It depends on the scope, and we state the number as a fixed price before the project starts. A scoped tracking engagement is a different order of magnitude from an internal portal. The comparison pages carry calculated scenarios with disclosed assumptions so you can judge the range yourself before you talk to us.

When is a system of your own worth it instead of SaaS?

When the process is permanent, becomes business critical, or the fee grows with user count and volume. For commodity software that many companies use in the same way, SaaS is usually the better answer, and we say so. We build the systems where ownership makes more economic sense than renting.

Do we own the source code outright?

Yes. The full source code, complete Git access, all documentation and the infrastructure credentials. You can change the system, extend it or pass it on. There is no licence from us that you keep paying for.

Can our existing team take the system over?

That is what it is built for. The stack is deliberately mainstream and documented, not proprietary. Handover includes a walkthrough and an operating guide. After that you can hire us, hire somebody else, or run it yourselves without needing our approval.

How long does a typical project take?

A scoped tracking engagement takes a few working days. Websites and dashboards run to several weeks, custom software to several months. The timeline is in the contract, together with what happens if it slips.

Further questions are answered on the service pages.

Free. No commitment.

Your next system
does not have to be another subscription.

A 30 minute technical assessment. We look at where things stand, name the actual problem and say so plainly when an off-the-shelf product is the better answer.

Discuss a project Run the numbers yourself
30 minute technical session Open cost breakdown A recommendation against us when that fits No sales pitch
Hetzner
ISO/IEC 27001:2022 certified data centres
GDPR-oriented
Data stored in the EU
Hetzner
Data centres in Germany (Falkenstein, Nuremberg)
TLS 1.3 + AES-256
In transit and at rest
Newsletter

Research for technical decisions

New reports, benchmarks and technical analyses on SaaS economics, AI engineering and owned infrastructure.

Original research Public sources No sales mail

By subscribing you receive new analyses and updates from FW Delta by email. You can withdraw your consent at any time. Further information is available in the privacy policy.