Hiring a web app development agency is one of the highest-stakes decisions a founder or product leader makes. Get it right and you have a technical partner that helps you build something durable. Get it wrong and you lose months and tens of thousands of dollars on code you can’t maintain, a product that doesn’t work, or a relationship that falls apart before you launch.
This guide covers how to evaluate agencies — not in theory, but in practice. The questions to ask, the signals to look for, and the patterns that predict bad outcomes early.
What You’re Actually Hiring
Before you start evaluating agencies, be clear on what you need. The right agency for a two-week marketing site build is not the right agency for a twelve-month multi-tenant SaaS platform. Agencies have different strengths:
Design-led agencies specialize in visual work, brand, and front-end experience. Strong for marketing sites, landing pages, and products where the UX is the product.
Engineering-led agencies specialize in complex back-end systems, integrations, performance, and architecture. Better for data-heavy products, workflow automation, and anything with complex business logic.
Full-stack product agencies claim to do both — and some actually do. These are the right fit for most SaaS MVPs and growth-stage product builds. They’re also where the quality range is widest.
Know which category you need before you talk to anyone. It shapes every question you’ll ask.
The Portfolio: What to Actually Look At
Most agencies show you polished case study PDFs. These are marketing materials. Here’s what to look at instead:
Live production URLs. Ask for actual links to products they’ve built that are live. Click around. Look at load times, UI behavior, error states, mobile responsiveness. A portfolio of screenshots and mockups tells you nothing about how the product performs.
Products similar in complexity to yours. If you’re building a multi-tenant SaaS with billing, user roles, and third-party integrations, ask specifically for examples of projects with those characteristics. General experience with “B2B software” is not the same as experience with the specific technical problems you need solved.
Projects they’d do differently. Ask them about a past project that went poorly and what they learned. Good agencies have honest answers to this. Agencies that claim every project was a success are either lying or haven’t built enough to have real failures.
The Scoping Conversation
The scoping conversation — the initial call where you describe your project and they ask questions — is the most revealing part of the evaluation. What you’re watching for:
Quality of questions. Good agencies ask hard questions early: What have you built so far? What’s the core hypothesis you’re testing? Who is the first user type? What does success look like in six months? Agencies that skip straight to timeline and budget are telling you something about how they work.
Pushback on scope. If you describe a thirty-feature MVP and the agency’s first response is “great, we can do all of that,” that’s a red flag. A good technical partner pushes back on unnecessary scope. They’ve seen what happens when MVPs try to launch with too much — they take twice as long and deliver half as much. They’ll tell you.
Technical opinions. Ask them directly: what tech stack would they recommend for your project, and why? If they give you a thoughtful answer that reflects your specific needs, that’s a good sign. If every project is “we’d use React and Node” regardless of the context, that’s a sign you’re looking at a shop that runs a template, not a team with judgment.
The Proposal
A good proposal is a diagnostic. It shows you how the agency thinks, not just what they’ll charge.
What a good proposal includes:
- A restatement of the problem in their own words (proves they understood it)
- A proposed scope with specific features listed, not vague categories
- Explicit out-of-scope items (what won’t be built)
- A technology recommendation with brief rationale
- Timeline broken down by phase, not just a total number
- A payment schedule tied to milestones, not just calendar dates
- A clear process for handling scope changes
What a bad proposal looks like:
- A number and a timeline with nothing behind them
- Scope defined as “build the app as discussed” — i.e., whatever you implied, without written confirmation
- An estimate that seems suspiciously low — usually because they’ve under-scoped to win the bid, and will re-scope once you’ve signed
- No mention of what happens if requirements change
One thing to specifically look for: does the proposal differentiate between a fixed-price engagement and a time-and-materials engagement? Both models are legitimate, but the right choice depends on how well-defined your requirements are. Fixed-price works when you have a locked spec. Time-and-materials works when you’re still figuring things out. An agency that doesn’t explain the difference probably doesn’t explain most things.
Red Flags
A few patterns that reliably predict problems:
“We’ll figure it out as we go.” This sounds collaborative. It usually means they don’t have a process for handling scope, and you’ll be paying for their ambiguity for the entire engagement.
Offshore team, onshore salesperson. Not inherently bad — offshore development can be excellent. But if the person you’re scoping with isn’t the person who will build it, ask to meet the actual team before signing. Surprises here are among the most common causes of project failure.
Reluctance to show code. If you ask to see example code from a past project and they refuse or can’t produce anything, that’s a serious warning sign. Agencies with good engineering culture are usually proud of their code and happy to share sanitized examples.
Packed timelines with no buffer. A twelve-week estimate with no slack means any complication will push the deadline. Good estimates include explicit buffer. If they don’t, ask where the buffer is. If the answer is “we’re efficient,” start looking elsewhere.
No post-launch support plan. Ask specifically: what happens if a critical bug appears two weeks after launch? Who fixes it, how fast, and what does it cost? If there’s no answer prepared, that tells you something.
Structuring the Engagement to Reduce Risk
Even a good agency can have a project go sideways. Structure the engagement to limit your downside:
Start with a paid discovery phase. A two-to-four week scoping engagement (paid, $5,000–$15,000 depending on complexity) produces a detailed spec, architecture document, and a fixed-price bid for the full build. This is the best $10,000 you can spend — it surfaces problems before you’ve committed to a full contract, and it tells you a lot about how the agency operates under real work conditions.
Milestone-based payments. Never pay the full project cost upfront. Structure payments to milestones: 20–25% on contract signing, 25% on design approval, 25% on working staging environment, the remainder on launch. This gives you leverage if quality problems emerge mid-project.
Code access from day one. Your code should live in a repository you own. The agency should have access to your repo, not the reverse. This ensures you have everything even if the relationship ends.
Regular demos. Weekly or biweekly working demos, not just status updates. Seeing the actual product at each stage is the only reliable way to catch misunderstandings before they become expensive rework.
The Right Question to End With
After you’ve done all of this — reviewed the portfolio, had the scoping conversation, read the proposal, checked references — ask one final question:
Can you show me the last project you delivered on time and on budget, and can I talk to that client?
If they can answer yes and arrange the call, you’re probably talking to someone worth hiring. If they hedge, it doesn’t mean they’re bad — projects slip for all kinds of reasons — but how they explain it will tell you more about how they handle adversity than anything else in the evaluation process.
We’ve been the agency on the other side of this conversation many times. If you want to talk through what your project needs and whether we’re the right fit, let’s get on a call.