Back to Blog
How to Rescue a Failing Software Project (Before Starting Over)

How to Rescue a Failing Software Project (Before Starting Over)

Dennis Reinkober2 min read
TL;DR

Don't rewrite yet. Freeze scope, secure repo, infrastructure, and domain access, run a two-week assessment, then choose: continue, partial rebuild, or restart. In the rescue projects we've taken over, 40–70% of the code survives — the data model and core logic are usually sound, and the rot lives in delivery, not in the codebase.

We regularly take over projects that another team started. The founder's first question is almost always "do we have to start over?" — and the honest answer is usually no.

How do you know the project is actually failing?

Slipping deadlines alone don't mean failure. These signs do:

Warning signSeverity
The demo shows things that don't exist in the deployed productCritical
Nothing has shipped to production in weeksHigh
One person holds all the knowledge — and all the accessHigh
Every estimate is "two more weeks", every timeHigh
No tests at allMedium — fixable, but it slows everything else

Two or more of the top four, and it's time to intervene.

The four-step triage

1. Freeze scope. No new features while you assess. Every addition widens the hole you're standing in.

2. Get the code and the access. Can you, without the current team, clone the repo, run the app locally, and deploy it? If the answer is no, that's the first fix — not architecture, not features. Repo, CI, cloud accounts, domains, third-party dashboards: in your hands.

3. Run a two-week assessment. A senior engineer (internal or external) sorts the system into three buckets: what demonstrably works in production, what's salvageable with focused effort, and what's rot. Two weeks is enough for a decision-grade picture; two months of analysis is procrastination.

4. Decide. Continue with the current base, rebuild the rotten parts behind stable interfaces, or restart. Make it an explicit decision with a date — the default of "keep going and hope" is the most expensive option of the three.

Do you really need a full rewrite?

Rarely. Across our rescues, 40–70% of the code survives. The data model and the core business logic are usually worth keeping; what gets rebuilt is typically the deployment setup, auth, and the messiest UI layer. A full restart only wins when the data model is fundamentally wrong or the stack is one nobody can hire for. Founders overestimate rot because the experience of the project was bad — the code is usually better than the process that produced it.

Switching agencies? Secure these before the relationship ends

The moment you announce a switch, cooperation drops. Before that conversation, make sure you hold: the repository in an org you own, root and billing access to every cloud account, domains and DNS, and whatever documentation and credentials exist. We've written about why handoffs go wrong in the agency handoff problem — the short version: ownership from day one is cheaper than recovery at the end.

Stuck project? We do the two-week assessment as a fixed-scope engagement and tell you honestly which bucket you're in — see our MVP development service.

Similar Posts

Let's Talk

Automation, AI, infrastructure or a software idea? Leave your phone number and we will call to discuss what you need.

Prefer to contact us directly? Write to us or give us a call.

We use your phone number to handle your callback request.

Or call: +49 (0)172 7593488