Most agency disasters are visible in the sales process if you know what to look for. The agency that ghosts after the kickoff invoice, the one that triples the timeline at week three, the one that hands you a pixel-perfect Figma file and zero working code. Every one of those outcomes leaves a trail of red flags during the discovery call. This is the checklist I wish every prospect I have lost to a bigger shop had run before signing.
Twelve questions. Score one point for every red flag the agency throws. If they hit five or more, walk. If they hit three to four, push hard on those before signing anything. If they hit zero to two, you probably have a real partner.
1. Will they show you a finished project end-to-end?
Not a Dribbble shot. Not a hero screenshot. A real, live URL you can click around. If the agency cannot show you three production sites in your category, you are paying them to learn on your project. Bonus red flag: the case studies on the site lead to dead links or "coming soon" pages.
What to ask: “Can I see three live URLs of projects similar to mine, and can I email one of those clients?”
2. Is the price tied to scope or to hours?
Hourly billing is a profit model, not a value model. The agency wins when the project takes longer. You win when it ships. A serious shop will quote a fixed price for a defined scope, with explicit change-order rules for additions. If the proposal says “estimated 240-320 hours at $175/hr” you are not buying a product, you are buying time.
What to ask: “Will you fix the price for the scope we just agreed on?” If the answer is no, ask why.
3. Who is actually building it?
Big agencies sell you their senior team in pitch meetings and assign your project to a junior. Medium agencies sell you a US-based principal and outsource to a contractor in another timezone. Solo studios are the only model where the salesperson and the engineer are the same human, and that consistency is undervalued.
What to ask: “Who specifically will write the code? What is their LinkedIn? Can I see their GitHub?”
4. How long until I see something working?
If the answer is “four weeks of design phase before any code,” that is a process built for billing, not for shipping. Modern delivery has working previews on day three. Real preview links every few days. Production-shaped staging environments before the design is even locked. Long discovery cycles correlate strongly with projects that ship late or never.
What to ask: “When will I see a clickable preview, and what URL will it live at?”
5. What happens when scope changes mid-project?
The honest answer is: a written change order with a price, a timeline impact, and a yes/no decision before any work happens. The dishonest answer is: “we just absorb small changes” which means they are pulling time from your other priorities to chase shiny additions. The worst answer is no answer, which means there is no process and the final invoice will surprise you.
What to ask: “Show me the change-order template you use.”
6. Do they own the code, or do you?
Some agencies retain ownership of the code base and license it back to you under a perpetual maintenance retainer. This is the single largest hidden lock-in in the industry. The contract should explicitly state that all source code, design files, and account credentials transfer to you on final payment, with no ongoing license requirement.
What to ask: “Show me the IP transfer clause in your standard contract.”
7. Can you leave without losing everything?
Even with code ownership, lock-in shows up in the platform layer. If they built your site on a proprietary CMS only their developers can edit, or hosted it on infrastructure with non-standard access patterns, you cannot fire them. Push for standard tooling: git, mainstream cloud providers, documented deployment processes.
What to ask: “If I take this code to another developer next year, what will they need to do to deploy it?”
8. What is the post-launch story?
Three real options after launch: hire them on retainer, hire someone else, or maintain it yourself. The agency answer is usually only the first one, because the retainer is where the margin lives. The honest answer is all three, with a clear handoff process for the second two. If post-launch is a black box, the build cost is not the real cost.
What to ask: “What does month one look like after launch if I do not sign a retainer?”
9. How do they handle bug reports?
Within the warranty window, bugs that came from the original scope should be fixed at no charge. Outside the warranty window, fixes are billed against a maintenance retainer or a clearly priced support tier. If they cannot tell you what the warranty is, they are negotiating it case by case, which means you lose.
What to ask: “What is the warranty window for bug fixes after launch, and what counts as a bug versus a change request?”
10. Have they shipped your stack before?
An agency that has shipped one Next.js project will charge you to learn the second. An agency that has shipped fifty will recognize the gotchas in your scope before you do. Ask for stack-specific case studies and read the technical posts on their blog. If their blog is full of design awards and zero engineering content, the engineering capability may be a sales claim.
What to ask: “How many projects have you shipped using the exact stack we are talking about?”
11. What is their average actual delivery vs. estimated delivery?
The most useful number in the entire engagement, and the one agencies will least want to share. A shop that consistently ships on time has data and will share it. A shop that slips by 30 percent on average will dodge the question. A shop that has never tracked the data is telling you they do not run the engagement as a delivery operation, they run it as a billing operation.
What to ask: “Across your last ten projects, what was the median ratio of actual to estimated delivery?”
12. How does the discovery call feel?
Trust the room. If the discovery call is a sales process disguised as a conversation, the engagement will be a sales process disguised as a partnership. The agencies you actually want to work with will push back on parts of your scope, suggest cheaper alternatives where they fit, and be willing to say “you do not need us for this.” If every answer is “yes, we can absolutely do that,” you are not on a discovery call, you are on a closing call.
What to ask: Yourself. “Did this person tell me anything I did not want to hear?”
The Honest Disclosure
I run a solo studio. The questions above are calibrated against the kind of agency engagements clients describe to me when they have already been burned and are looking for an alternative. They are not neutral journalism. They are the pattern I see across roughly forty inbound conversations a year, where the buyer arrived after a previous agency missed the timeline, padded the budget, or held the code hostage.
Solo studios are not always the right answer. For enterprise procurement, multi-platform mobile apps, or scope larger than one person can ship in twelve weeks, an agency is the correct model. The point of the checklist is not to route every project to a solo studio. The point is to make the failure modes of the agency model visible during the sales process, before the contract is signed and the leverage flips.
What to do with the answers
Score the agency. Five or more red flags means walk away regardless of how polished the proposal looks. Three to four means renegotiate before signing. Zero to two means you probably found a real partner.
If you are still deciding whether you even need an agency, run the scope through the no-code vs custom decision tree first. The cheapest agency engagement is the one that turns out to be a Squarespace site and a Stripe checkout link. If you have already decided to build custom, get a fixed-price baseline before you talk to anyone, so you can spot inflated quotes when they arrive.
For a deeper read on the cost differences between agency, freelancer, and solo studio, the 2026 hiring guide walks through the real numbers from shipped projects.