DEV Community

Simaila
Simaila

Posted on • Originally published at techvinta.com

How to Choose a Software Development Company: 15-Point Checklist (From Someone Who's Been Burned) 🔥

I once wrote a $40,000 check to a development agency. They ghosted me three months in. Half-built product. No documentation. No source code access.

That experience — and years of being on BOTH sides of the table (as a founder hiring agencies, and as the head of an agency being hired) — taught me exactly what to look for and what to run from.

Here's the 15-point checklist I wish I had back then.


## The 15-Point Checklist

### 1. 🔍 They show you LIVE products, not just screenshots

Visit their portfolio links. Click through them. Try to break them. I once reviewed a portfolio that looked stunning — until I visited the actual apps. Two were offline, one had a broken login.

### 2. 🛠️ Their tech stack matches YOUR needs

Beware the agency that recommends the same stack for every project. If they suggest microservices + Kubernetes for your simple MVP, they're either padding the bill or only know one way to build.

### 3. 📋 They have a defined discovery process

Run from anyone who quotes you after a 30-minute call. Real estimates need real analysis — user stories, architecture diagram, phased roadmap.

### 4. 👥 You can meet the actual developers

The sales call features their A-team. Then your project gets handed to juniors. Always ask: "Can I meet the developers who will write my code?"

### 5. 📢 They own their communication

Daily standups. Weekly demos. Shared project board. The best teams I've worked with sent a short video walkthrough every Friday.

### 6. 🔄 They talk about maintenance BEFORE you ask

Building is 30% of lifetime cost. The other 70% is maintaining it. If they only discuss the build phase, they're planning to disappear after launch.

### 7. 💰 They explain pricing without squirming

Fixed-price, T&M, or hybrid — doesn't matter. What matters is they can explain exactly what you're paying for and what happens when scope changes.

### 8. 📞 They have references you can CALL

Clutch reviews are good. But what you really want is a phone call with a past client whose project hit rough patches.

### 9. 🚫 They push back on your ideas

If they agree with every feature, they're not your partner — they're your order taker. The best teams say "no" more than "yes."

### 10. 🏥 They understand your industry (or admit they don't)

An honest generalist who asks smart questions > a fake specialist who builds the wrong thing confidently.

### 11. ✅ They write tests and can prove it

Ask for an actual test coverage report. Not a number on a slide deck — a real report.

### 12. 🔀 They use version control properly

Git with proper branching, pull request reviews, CI/CD. If deployment involves FTP and crossed fingers, walk away.

### 13. 🔒 Security is built in from day one

Authentication, encryption, input validation — they should bring this up before you do.

### 14. 📄 Clear contract with exit clause

You own the code. Period. Full repo access. All IP transfers to you. You can leave with your codebase anytime.

### 15. ❤️ They care about your BUSINESS, not just your project

The best partners ask about your runway, go-to-market strategy, and revenue model.


## 🚩 Red Flags — Walk Away Immediately

  • No live products in portfolio — screenshots with no links = they didn't build them
  • Fixed price without discovery — they're pulling numbers from thin air
  • Developers silent during technical calls — you're not evaluating the builders
  • "We can build your marketplace in 4 weeks" — no, they can't. Not well.
  • They never ask about your budget — they'll surprise you later when you're emotionally invested

## 🤝 How to Run a Trial Project

Before signing a $50K+ contract, run a small paid trial ($3K-$8K):

  1. Pick a real feature from your roadmap (not throwaway work)
  2. Define clear success criteria in writing
  3. Evaluate the PROCESS, not just the output — communication, code quality, testing, documentation
  4. Run trials in parallel with 2 companies if budget allows — the comparison will be eye-opening

As Toptal's hiring guide puts it: the cost of a bad hire far exceeds a thorough evaluation.


## ⚖️ Agency vs Freelancer vs In-House

Freelancer ($3K-$15K/mo) — Cheapest but highest risk. Great for small tasks. Terrible for building your core product. Single point of failure.

Agency ($8K-$50K/mo) — Full team (PM + designers + devs + QA). Process-driven. Accountable. Best for MVPs and full product builds.

In-House ($25K-$80K/mo) — Full control. Deep knowledge. But takes 2-4 months just to hire, and you're paying salaries before writing any code.

My recommendation: Agency for the initial build → transition to in-house for ongoing development once you have product-market fit.


## 🎯 The Bottom Line

The right development partner isn't the cheapest, the fastest, or the most experienced. It's the one that treats your project like their own product.

Switching partners mid-project costs 2-3x more than choosing the right one upfront. Measure twice, cut once.


Originally published at techvinta.com

Looking for a development partner that checks all 15 boxes? Get a free project estimate from our team.

Top comments (1)

Collapse
 
jon_at_backboardio profile image
Jonathan Murray

The $40k ghosting story is unfortunately common, and the "both sides of the table" perspective makes this more credible than most checklist articles. The asymmetry in that relationship is significant — clients usually have one experience with agencies, agencies deal with dozens of clients and know exactly where the leverage gaps are.

The source code access point is the most important one on any list. Requiring repo access from day one (not day thirty) is the single fastest filter for agency quality because any agency that hesitates on that is already telling you something about how they plan to handle the relationship. Same with documentation — asking to see an example of their handoff docs from a previous project before signing is a low-effort ask that separates professional operations from shops that are winging it.

What's your take on milestone payment structures vs time-and-materials for de-risking scope creep? That seems like one of the highest-leverage contract decisions most clients get wrong.