Most half-built MVPs are salvageable. If you have access to the code, it runs, and there is a real product idea underneath the mess, the problem is usually fixable rather than fatal. The honest exception is rare: a codebase you cannot obtain, or one so thin it was never really built. This checklist helps you tell which situation you are in.
The short answer
- Salvageable in most cases: you have the source code, it runs somewhere, and the core feature does something real.
- Fixable but slow: the code is messy or half-finished, but the shape of the product is sound.
- Better to rebuild parts: specific areas are broken, but the rest is worth keeping.
- Genuinely stuck: you cannot get the code, or there is almost nothing there. This is rarer than founders fear.
The word "salvageable" frightens people more than it should. In practice, finishing a started MVP is nearly always cheaper and faster than starting again, because the thinking, the design decisions, and the early feedback are already paid for.
Can you actually get the code?
Everything starts here. Before you judge quality, confirm ownership and access.
- Do you have the login to the code repository, usually on GitHub, GitLab, or Bitbucket?
- Can you, or someone you trust, download the whole thing today?
- Do you own the accounts for hosting, the database, and any third-party services?
If a previous developer holds the keys, your first job is recovering access, not fixing bugs. This is common and usually resolvable with a calm, professional request. If you reach a genuine dead end, a fresh start may be forced on you. But do not assume the worst until you have asked plainly and given it a week.
A quiet point of principle: any team you bring in should leave you owning everything. At Shpatik Studio you always keep all your code, accounts, and infrastructure. That is the baseline, not a favour.
Does it run?
An MVP that starts up, even badly, is a much better position than a pile of files nobody can launch.
- Can someone get it running on a laptop from the code alone?
- Is there a version live on the internet, even a rough one?
- When it runs, does the main thing it was built to do actually happen?
If a developer can clone the project and see it working within a day or two, you are in good shape. If it takes a week just to get it to start, that is a signal about how it was built, but still not a reason to abandon it. Startup trouble is a known, bounded problem. It tells you the finishing work will take longer, not that the work is impossible.
How much of the core actually works?
Founders often describe an MVP as "nearly done" when what they mean is "most of the screens exist". Those are different things. Look for working behaviour, not the appearance of it.
- Pick the single most important action in your product. Does it work end to end, for real, with real data?
- Can a stranger sign up, do that one thing, and get a result without you standing over them?
- How many features are genuinely finished versus started and abandoned?
A blunt guide. If the core action works and the rest is polish, you are close. If the core action does not work yet but everything around it exists, you are further out but the foundation is real. If nothing connects to anything, that is the hardest case, and worth a second opinion before you decide.
What is the state of the code underneath?
You do not need to read code to sense its health. You need to ask the right questions and watch how they are answered.
- Is there any automated testing, or is every change a gamble?
- When something is fixed, do three other things break?
- Can a new developer understand the project, or does it only make sense to the person who left?
- Was it partly generated by AI tools without anyone checking the result?
That last point matters more each year. A lot of MVPs are now vibe-coded: built quickly with AI assistance, working on the surface, fragile underneath. This is very fixable, but it needs an honest look rather than more of the same. If that describes your project, our vibe-code cleanup work exists precisely for it, and a proper code audit will tell you how deep the tidying needs to go.
What do the questions to your last developer reveal?
If the person who built your MVP is still reachable, a short conversation tells you a great deal. Ask:
- What is finished, what is half-done, and what was never started?
- Where are the parts you were least happy with?
- If you had two more weeks, what would you fix first?
A developer who answers openly is giving you a map. A developer who is vague or defensive is telling you something too. Neither answer condemns the project. Both help you plan the finishing.
Key takeaways for your decision
- Access beats quality. Owning the code and accounts matters more than how tidy the code is.
- Running beats pretty. A rough live version is worth more than a beautiful design nobody can launch.
- Core beats count. One feature that truly works is worth more than ten that half-exist.
- Honesty beats hope. A clear, unflattering assessment now saves months later.
If most of these point the right way, your MVP is salvageable, and finishing it will almost certainly cost less than the fresh start you may be dreading. Rebuilds are sometimes right, but they are the exception, and they should be a deliberate choice rather than a panic.
A calm way to get a straight answer
You should not have to guess at this alone, and you should not have to take a sales pitch on faith either. The most useful thing is an outside engineer looking at the actual code and telling you plainly what they see.
That is what our fix a broken MVP service is built around: taking a stalled or shaky product and getting it finished, without throwing away the work already done. Where the situation is bigger than a single MVP, our broader software project rescue work covers the same ground with more scope. If you are weighing up an investment or an acquisition rather than your own build, technical due diligence answers the same question from the buyer's side.
We are a small team of senior engineers based in Moldova, working in EU time zones, GDPR-aware, at roughly 40 percent below typical Western rates. We have shipped and run live products, including GazetAI, CarVinVin, and Bank-Lift, so the assessment you get comes from people who finish things, not just critique them. You can see more of that on our portfolio, and a plain account of how we work.
If you would like a second opinion before you decide anything, start with a free 15-minute triage call. Send us a short note about where your MVP is stuck through our contact page, and we will tell you honestly whether it is worth finishing. Often, the answer is a quiet yes.