How to Evaluate a Software Development Partner (Without Getting Burned)

We've inherited projects from over 30 agencies and freelancers. Stalled builds. Spaghetti codebases with zero tests. 'Almost done' projects that were 40% complete. Security vulnerabilities that would make a compliance officer cry. And in every case, the client made reasonable decisions based on the information they had - they just didn't know the right questions to ask.

This article is those questions. The patterns of what went wrong are remarkably consistent, and the red flags were almost always visible during the sales process. Here's how to spot them before you sign anything.

Red flag #1: They agree with everything you say. The biggest red flag in agency sales is unanimous agreement. 'Yes, we can build that.' 'Sure, 6 weeks is realistic.' 'Absolutely, that's within budget.' A good development partner will push back. They'll tell you feature X isn't worth the complexity. They'll say your timeline is aggressive and here's why. They'll suggest removing 40% of your spec to ship something meaningful faster. If they're not challenging your assumptions, they're either not experienced enough to know the risks, or they're saying whatever it takes to close the deal.

Red flag #2: Fixed price without scoping. Any agency that gives you a fixed price after a single discovery call is either padding the estimate by 200-300% to cover unknowns, or planning to cut corners when the actual work exceeds their assumptions. Both outcomes are bad for you. A proper scoping phase - typically 1-2 weeks, costing $3,000-$8,000 - produces detailed technical specifications, architecture decisions, and realistic estimates. Yes, it's an upfront cost. But it saves 10-50x in avoided miscommunication, scope creep, and rework.

Red flag #3: You can't talk to the engineers. If the agency keeps engineering behind a sales wall - 'our team will review your requirements and get back to you' - that tells you how your project will be managed. Layers of telephone between you and the people writing your code. Decisions made by project managers who don't understand the technical implications. Features lost in translation. The people building your product should be in the room from the first meeting, asking technical questions and offering implementation alternatives.

Green flag: They have a structured process. Good agencies have a methodology you can inspect: how they run discovery, how they structure sprints, how they handle change requests, how they communicate progress, how they manage deployments, how they do QA. Ask them to walk you through their process in detail. If the answer is vague ('we're agile, we adapt') - that usually means they don't have a process at all.

Green flag: They show you failures, not just wins. Every agency has beautiful case studies. Ask them about a project that went sideways. How did they handle it? Did they communicate the problem proactively or did the client discover it? Did they eat the cost of their mistakes? Did they bring in additional resources? What did they learn? A partner who can discuss failures openly is a partner who's honest about risks - and that honesty is worth more than any portfolio.

The 12 questions you should ask (and most companies don't). 1) Can I meet the specific engineers who'll work on my project? 2) What happens if my project is behind schedule - who pays? 3) Can you share a recent project that went wrong and how you handled it? 4) How do you handle change requests mid-sprint? 5) What's your testing strategy and what code coverage do you target? 6) Who owns the IP and source code? 7) What happens after launch - do you offer support? 8) How do you handle developer turnover mid-project? 9) Can I speak to a client whose project had problems? 10) What's your security and compliance track record? 11) How do you document technical decisions? 12) What would you cut from my spec to ship faster?

The cost of 'cheap' - real numbers from real projects. In the past 18 months, we've rescued 8 projects from other agencies. Average original budget: $35,000. Average amount spent with the original agency before the client pulled the plug: $28,000. Average cost for us to rebuild properly: $52,000. Total cost to the client: $80,000. If they'd hired a competent agency from the start, the project would have cost $45,000-55,000 and been done 4-6 months earlier. The cheapest bid is almost always the most expensive decision.

Our recommendation: invest in scoping before committing. Before signing a development contract, invest in a paid scoping engagement. This should produce: a detailed technical specification, architecture recommendations, a realistic timeline with milestones, a cost estimate broken down by feature, risk identification with mitigation plans, and team composition recommendations. Use this document to compare agencies on substance, not sales pitches. The agency that produces the most honest, detailed scoping document is usually the right choice - even if their estimate is higher than the competition.