How to Evaluate a Web Agency: A Startup Founder's Checklist + RFP Template

Most founders pick the wrong web agency the same way: a friend's referral, a slick portfolio, one good sales call, and a contract two weeks later. Then the rebuild starts six months in.

After reviewing hundreds of project inquiries from founders — pre-seed solos to post-Series A teams — the pattern is clear: the founders who get great sites all do the same five things before signing. They write a real brief, they vet the team they'll actually work with (not the partner on the call), they read the contract for IP language, they ask past clients the uncomfortable questions, and they share their budget upfront.

This guide is a 30-point checklist plus a copy-paste RFP template you can send to three to five agencies tomorrow. It's written from the agency side: what we screen for, what we wish founders did differently, and the red flags that should kill a deal before you waste another call.

Who this is for: Pre-seed and seed founders, Series A CMOs running their first agency RFP, and technical co-founders asked to vet vendors. Budgets $15K–$150K, timelines 8–24 weeks.

Key Takeaways

• A useful agency vetting process covers 8 dimensions: portfolio depth, technical fit, team structure, process, pricing model, contract terms, references, and post-launch.
• Portfolio screenshots lie. Ask which work was led by the team you'd actually get. Junior-led projects look identical to founder-led ones in a deck.
• Webflow ≠ custom dev ≠ headless. Match the tool to your stage. A $20K Webflow build can outperform a $80K Next.js stack at pre-seed.
• IP and source-code ownership must be explicit in the contract. "Work for hire" + escrowed repo handoff = the safe default.
• Budget transparency cuts your timeline in half. Sharing a range upfront lets agencies design to your constraints instead of pricing blind.
• Use the inline RFP template at the bottom. Copy, fill in, send to three to five agencies. That alone moves you ahead of 80% of founders running this process.

1. Portfolio Audit: What to Look For Beyond the Pretty Screenshots

Every agency portfolio looks great. That's the job. Look past the visuals.

1-1. Match-the-Stage Filter

Sort their portfolio by company stage at time of project, not by your favorite design.

• Pre-seed/seed sites should look fast, opinionated, and shippable. If their seed-stage portfolio is full of $200K productions, they don't know how to constrain scope.
• Series A+ sites should show information density, conversion patterns, and structured content systems. If their Series A work looks like a brochure, they don't ship for product-led growth.

Ask: "Show me three projects done with similar budget and timeline to mine. Who led each?"

1-2. The Lead-vs-Junior Test

Top agencies sell with their senior partner and execute with juniors. This isn't always bad, but you need to know which mode you're buying.

Ask directly: "Who designed and engineered each piece in your portfolio I'm looking at? Will they be on my project?"

A good answer is specific names. A bad answer is "our team."

1-3. Live-Site Stress Test

Open three of their recent client sites in dev tools. Check:

• Lighthouse score (Performance, Accessibility, SEO)
• Mobile rendering at 360px width
• Whether forms still work (try one)
• Whether they're still online a year after launch

If their own portfolio sites have CLS issues and broken contact forms, that's what your site will look like in 12 months.

2. Technical Fit: Webflow, Custom, Headless — Match Tools to Stage

The biggest founder mistake is buying a stack instead of a website.

2-1. The Stage-Stack Map

If an agency proposes a custom Next.js + GraphQL + Contentful stack for your seed-stage marketing site, ask why. Sometimes the answer is good. Often the answer is that's just what they build.

2-2. The "Why This Stack" Question

Make every shortlisted agency explain, in plain English:

• Why this stack for my stage and team
• What I'd hand off to a junior dev or in-house hire later
• What it'll cost to maintain in year two
• What breaks if I want to A/B test next year

If they can't answer cleanly, they're selling their toolkit, not your site.

2-3. Performance + SEO as Defaults

In 2026 these are non-negotiable:

• Core Web Vitals targets (LCP <2.5s, INP <200ms, CLS <0.1) — agreed in writing
• Schema markup, hreflang for multi-locale, sitemap automation
• Image pipeline (WebP/AVIF, responsive images)

For more on the team-model side of this question, see our in-house vs agency vs freelance web guide.

3. Team Structure: Who You Actually Work With

The org chart matters more than the case studies.

3-1. The Five Roles to Identify

Ask the agency to name the human filling each role on your project:

1. Strategist / Account lead — translates your goals into scope
2. Designer — does the actual visual work
3. Engineer(s) — writes the code that ships
4. Project/Producer — owns timeline and unblocks decisions
5. QA / DevOps — last-mile launch + handoff

If two roles collapse into one person on a $50K+ project, scope is at risk. If five roles balloon into eight on a $30K project, your money is going to overhead.

3-2. In-House vs Subcontracted

Some agencies subcontract design or engineering. That's fine if disclosed. Ask: "Are any of these roles subcontracted? Through which firm?"

Subcontracted teams across timezones extend your timeline by 20–40%. Price that in.

3-3. Who Replaces You in the Room

You will not be on every meeting. Ask who runs the project when you're not there, and what their authority is to make decisions on your behalf. A project lead who has to "check with the founder" on every $500 decision will burn weeks.

4. Process & Communication: Async, Standups, Handoff

Agency communication style determines how much of your week the project eats.

4-1. Async vs Synchronous Cadence

Ask:

• Daily Slack/Linear updates, or weekly status calls?
• Decision log shared in Notion / Linear / email?
• Who escalates when something is at risk?

A good answer is specific tools and rituals. A bad answer is "we'll figure out what works."

4-2. Feedback Velocity

Founders kill projects by batching feedback. Agencies kill projects by drowning in opinions. Negotiate this upfront:

• Two scheduled feedback windows per sprint
• Single source of truth for design comments (Figma, not Slack)
• 48-hour decision SLA from the founder side
• Named final approver — usually the founder, sometimes the CMO

4-3. Documentation Quality

Ask to see anonymized examples of:

• A previous project's brief response
• A change log / decision doc
• A handoff README they delivered

If they can't show samples, you'll inherit chaos.

5. Pricing Models: Fixed-Bid vs T&M vs Retainer

The pricing model often matters more than the price.

5-1. Fixed-Bid

Best for tightly scoped projects (marketing site, redesign of existing IA). Risk is on the agency, so they pad. Expect 15–25% buffer. Change orders are inevitable — read the change-order clause carefully.

5-2. Time-and-Materials (T&M)

Best for ambiguous scope (web app, ongoing iteration). Agency bills per hour or per sprint. Risk is on you. Cap with a not-to-exceed number and a weekly burn report.

5-3. Retainer

Best for ongoing relationships post-launch (content scale, growth experiments, A/B testing). Typical: $5K–$25K/month for two to four days of capacity per week.

5-4. Hybrid

Most realistic for startups: fixed-bid for v1 launch, then convert to retainer post-launch. Negotiate the conversion terms upfront.

For full budget context, see our premium website cost budget guide.

6. Contract Red Flags & IP / Code Ownership

This is where founders get burned.

6-1. IP and Source-Code Ownership

Default should be: work-for-hire, all deliverables and source code transferred to the client at final payment, repo escrowed during the project.

Watch for these red-flag clauses:

• "Agency retains rights to reusable components" — vague, can mean half your site
• "License granted to client" instead of full transfer — you're renting your own site
• No mention of source code at all — you'll get the build artifacts, not the repo
• "Hosted on agency infrastructure" with no migration clause — you're locked in

6-2. Termination Clauses

You need a clean exit. Look for:

• Termination for convenience with 30-day notice
• Pro-rated refunds for unused fixed-bid budget
• Repo + asset handoff within 14 days of termination
• No claw-back clauses on already-delivered work

6-3. Scope-Change Mechanics

Every project changes. The contract should specify:

• Change-order threshold (e.g., anything above 4 hours requires a written change order)
• Hourly rate for change orders, capped
• Decision SLA on change orders (agency responds within 3 business days)

6-4. Liability + Confidentiality

• Mutual NDA before discovery
• Liability cap reasonable (typically equal to fees paid)
• No personal guarantee from the founder

7. References & Case Studies: Questions to Ask Past Clients

Most founders skip references. The ones who don't get the best signal in the entire process.

7-1. Ask for Three References, Get Two

Agencies will offer the two most enthusiastic clients. Ask for three — the third is usually less polished and tells you more.

7-2. The Five Reference Questions

When you get them on the phone, ask:

1. "What did the agency do better than expected?" — calibrates strengths
2. "Where did the project go sideways, and how did they handle it?" — calibrates failure mode
3. "Were the people on your project the same as the ones in the sales pitch?" — catches bait-and-switch
4. "What did you wish you'd known before signing?" — golden question
5. "Would you hire them again, and for what kind of project?" — net-promoter, with nuance

If a reference dodges question 2, the project went badly and they don't want to say.

7-3. Post-Launch Reality Check

Ask references: "How responsive were they 90 days after launch?" Many agencies vanish post-payment. You want one that doesn't.

8. Post-Launch: SLA, Maintenance, Handoff to In-House

The launch is the easy part. Year two is where the value lives.

8-1. Maintenance SLA

Specify:

• Critical bug response time (target: <4 business hours)
• Non-critical bug fix window (target: <5 business days)
• Patching cadence for dependencies (target: monthly)
• Uptime monitoring + on-call rotation (or explicit "no on-call")

8-2. Handoff to In-House

When you eventually hire your first internal dev, the handoff cost determines whether the agency-built site is an asset or a liability. Demand:

• Architecture README (under 5 pages, written for a senior dev who's seen it before)
• Loom or recorded walkthrough of the codebase (under 30 minutes)
• Up-to-date dependencies, no abandoned packages
• Clean commit history on a public-style branch model
• All credentials and accounts transferred (DNS, hosting, CMS, analytics, error reporting)

8-3. Content + Growth Capacity

If your agency vanishes after launch, you're left without the team that knows how to extend the site. Either:

• Negotiate a retainer with capped capacity
• Or get an explicit "we'll respond within 5 days for paid scope expansions" clause
• Or train your in-house team during the project so they can extend it themselves

9. The Founder's RFP Template (Copy-Paste)

Below is a copy-paste RFP. Fill in the brackets, send to three to five agencies, give them two to three weeks to respond.

Do this: Save your filled-in version as a Notion doc or PDF. Send a single shared link, not an attachment. Track who opens it.

Web Agency RFP — [Project Name]

1. About Us

• Company: [Name]
• Stage: [Pre-seed / Seed / Series A / Series B]
• Funding raised: [Amount + date of last round]
• Team size: [N people]
• ICP: [Who buys from you]
• Current site: [URL or "founder-built Webflow"]

2. Project Type

• New marketing site
• Marketing site refresh / rebrand
• Web app dashboard / product UI
• Landing page system for ABM / paid
• E-commerce / pricing portal
• Other: [describe]

3. Goals + Success Metrics

• Primary objective: [e.g., 2x demo requests in 6 months]
• Secondary: [e.g., signal scale to enterprise prospects]
• Anti-goal: [what you do NOT want, e.g., generic SaaS template look]
• KPIs: [conversion rate, signups, demo bookings, time-on-site]
• "Good" 6 months post-launch looks like: [describe]

4. Scope

• Pages / sections required: [Home, Pricing, Product, Use cases, About, Blog, etc.]
• Major features: [search, gated content, CMS for marketing team, multi-locale, A/B testing]
• Integrations: [HubSpot, Segment, Stripe, Auth0, Algolia, etc.]
• Out of scope: [be explicit — apps, internal tools, etc.]

5. Brand Assets

• Brand guidelines exist: [Yes / No / Partially]
• Logo + typography locked: [Yes / No]
• Brand refresh required as part of project: [Yes / No]
• Photography / illustration owned: [Yes / No]

6. Technical Constraints

• Preferred stack: [Webflow / Next.js / Astro / no preference]
• Required: [hosting on Vercel / supports our SOC2 vendor list / etc.]
• Forbidden: [WordPress, jQuery, anything that breaks our SOC2]
• Existing systems to integrate: [name + docs link]
• Performance targets: [LCP <2.5s, INP <200ms, CLS <0.1]
• Accessibility target: [WCAG 2.1 AA]

7. Timeline

• Target launch: [Date]
• Hard deadlines + reasons: [e.g., demo day, fundraise close]
• Internal review windows available: [days/week we can give feedback]

8. Budget

• Range: $[low] – $[high]
• Includes: design, build, content migration, [post-launch month of support]
• Excludes: [stock photo licenses, ongoing retainer]

9. References We Like (Anti-References Too)

• Sites we'd love to look like: [3 URLs + 1-line each on why]
• Sites we don't want: [2 URLs + 1-line on what to avoid]

10. Stakeholders + Decision Process

• Final approver: [Name, title]
• Day-to-day point of contact: [Name, title]
• Decision deadline: [Date by which we'll choose]

11. Submission

• Deadline: [Date]
• Format: [PDF / Notion link / loom + doc]
• Send to: [email]
• Questions: [email + Slack option if any]

10. Conclusion

The founders who hire well do five things differently: a real brief, named-team vetting, contract-level diligence, structured references, and budget transparency. None of it is hard. All of it is uncommon.

Pick three to five agencies. Send the RFP. Run the same checklist on every response. The agency that wins on the checklist is rarely the agency with the flashiest deck — it's the one that answers your scope back to you with sharper questions.

For your first shortlist, browse our top creative agencies in Asia (2026) or our creative web studios in Osaka directories, and pair this checklist with the web budget guide before you send.

11. About Utsubo

Utsubo is a creative studio in Osaka, Japan. We design and build production-grade web experiences for SaaS, hospitality, cultural, and consumer brands — including marketing sites, product UI, and interactive web installations using Three.js and WebGPU. We've shipped sites and digital installations for clients including JR West and projects connected to Expo 2025.

We're often on the receiving end of RFPs like the one above. We wrote this guide so the founders we don't work with also run a clean process.

12. Let's Talk

Building a marketing site or web product where the design has to do real work — convert, scale, signal seriousness for a fundraise? We work with founders and CMOs on web projects from pre-seed marketing sites through Series B platforms.

If you're exploring a partnership, let's discuss your project:

• What you're building and the constraints you're working with
• Which technical approach makes sense for your stage and budget
• Whether we're the right fit to help you execute

Book a project discussion

Prefer email? contact@utsubo.co

13. Pre-Submission Checklist

Before sending your RFP, verify:

• Stated stage, funding, and team size honestly
• Defined primary goal as a measurable outcome
• Listed scope as specific pages and features (not vibes)
• Called out integrations and existing systems
• Set a budget range — not a ceiling
• Named a final approver and a day-to-day contact
• Attached or linked brand assets if they exist
• Listed 3 reference sites you like and 2 you don't
• Set a submission deadline at least 2 weeks out
• Sent to no more than 5 agencies
• Set a decision deadline within 4 weeks of the submission deadline

FAQs

How many agencies should we send the RFP to? Three to five. Fewer than three limits your options; more than five dilutes the quality of responses, and good agencies can tell when they're one of fifteen and deprioritize.

Should we share our budget upfront? Yes — as a range. Agencies quote blind without it, which produces proposals spanning 4–5x for the same brief. Sharing a range like "$40K–$80K" lets agencies design to your real constraints. It's not a ceiling; serious shops still propose what fits the goal.

What if we don't have brand assets yet? Say so. Some agencies do brand + site as one engagement, some only execute on locked brand. The honest answer narrows your shortlist correctly.

Webflow or custom dev for a seed-stage marketing site? Default Webflow unless you have a specific reason not to. Speed to ship, low maintenance overhead, easy CMS for non-technical marketers. Move to custom (Astro / Next.js) when SEO scale, A/B testing infrastructure, or product-marketing integration justifies the operating cost.

How long should we give agencies to respond? Two to three weeks for a thoughtful response. One week is enough time only for templated proposals — that's a self-defeating filter.

What's a fair budget for a Series A marketing site refresh? Most fall in the $40K–$120K range for a 6–10 page site with a CMS, content migration, and a month of post-launch support. Above $120K usually buys brand refresh + photography + complex integrations. Below $40K usually means scope cuts that show.

What's the single biggest red flag? Vagueness on IP and source-code ownership in the contract. Anything other than "all deliverables and source code transferred at final payment, work-for-hire" should trigger a rewrite request. If they push back, walk.