Hiring a web developer is a trust fall. You're about to hand several thousand dollars — and the digital front door of your business — to someone whose work you can't fully evaluate until it's done. Get it right and it's the best money you spend all year. Get it wrong and you join the long, sad parade of business owners paying twice: once for the site that didn't work, once for the rescue.
Most guides on this topic are generic checklists written by people who've never hired anyone: "define your needs,Five questions that reveal everything check portfolios, read reviews." Technically true. Practically useless — like advising someone to "buy low, sell high."
This one's different. We're a five-person studio that's shipped 150+ projects since 2022, and our founder hires — and gets hired — on Upwork's front lines. We've sat on both sides of this table hundreds of times. Here's what actually matters — and the exact answers that should make you walk away.
TL;DR — the takeaways
Brief outcomes, not pages. "Visitors book a call" beats "5 pages" — developers can only estimate accurately against outcomes.
Vet with evidence, not charm. Live sites you click yourself, written answers, and a small paid test project filter out 90% of mismatches.
Structure payment in milestones (roughly 20/40/25/15) — never 100% upfront, never "pay when you're happy."
Own everything from day one: domain, hosting, code repository, analytics. The developer gets collaborator access, never sole ownership.
The first three messages predict the whole project. Slow, vague, or evasive during hiring means slow, vague, and evasive during the build.
First: freelancer, agency, or in-house?
An agency telling you when not to hire an agency is either honest or bad at sales. We'd rather you hire right than hire us wrong.
Freelancer
Best for: bounded projects with a clear outcome — a site, a store, a fix, an MVP. Two weeks to four months of work.
Real cost: $500–$8,000 for a typical small-business site; marketplace rate data puts North American mid-level devs around $70–100/hr, with offshore rates much lower.
Biggest strength: you talk to the person writing the code. No telephone game, no markup.
Biggest risk: single point of failure. If they vanish, stall, or turn out to be wrong for the job, you're restarting.
Agency / studio
Best for: multi-skill projects (design + build + content + SEO), ongoing work, or when you want one accountable team.
Real cost: $4,000–$18,000+ for a small-business site (our Sharp Site starts at $4k — prices on the page).
Biggest strength: bench depth — if someone gets sick, the project doesn't stop. Plus design, QA, and project management in one place.
Biggest risk: costs more; small projects can feel like a big machine for a small job.
In-house hire
Best for: permanent, full-time product work — 40hrs/week indefinitely.
Real cost: a loaded US salary of $130k–160k/year. Absurd for "we need a website."
Biggest strength: they're just... there. Every day. Forever.
Biggest risk: you're now managing a developer, which is a job in itself.
Our honest take: well-defined project plus weekly feedback — a good freelancer is often the best value in web development. Multi-specialty project, reliability over price, or nobody on your side to manage it — hire an agency. The website is the business and the work never ends — hire in-house. Most small businesses need the first two, not the third.
Write a brief that gets good quotes
Here's a secret from the hiring side: most failed projects were doomed by the brief, not the developer. "Need a modern website" gets fifty proposals guessing at what you want — and the cheapest guess usually wins.
Write one page covering:
The outcome, measurably. "Visitors book a consultation call — currently 0.4% convert, target 1.5%." Not "5 pages."
What it must do. Booking, payments, logins, integrations (CRM, email, Stripe) — the functional stuff that drives cost.
What you're providing. Logo? Copy? Photos? Product data? Content delays kill more timelines than coding ever does — be honest about what exists.
Budget range and deadline. A range is fine ("$3k–$5k, live by December"). Secrecy about budget doesn't get you a better price; it gets you incomparable quotes.
How you'll decide. "Shortlist 3, paid test task, decide by Friday."
Post the same brief everywhere you look. Comparable briefs produce comparable quotes — and a quote you can't compare is a quote you can't trust.
Where to actually find candidates
Referrals from other business owners. Still the highest-signal channel. Ask specifically: "who built your site, would you hire them again, and what went wrong?" That last question is the useful one.
Upwork and marketplaces. Huge supply, escrow protection, real reviews — and huge noise. A popular posting gets 50+ proposals in 48 hours, most of them templated. Ignore the proposal, click through to the profile, open their live work. Proposals blur together; a live site that loads fast and works on your phone never lies.
Vetted networks (Toptal, Lemon.io, Arc). Pre-screened seniors at $80–$200/hr. Worth it for critical rebuilds and MVPs; overkill for a brochure site.
Direct / small studios. You talk to the builder, no middleman markup. This is our lane — our shipped work is public, and you can go poke the live sites yourself.
Platform expert directories. Shopify, Webflow, and WordPress all certify experts. Useful when you're locked into one ecosystem.
The 10-minute vetting funnel
Vetting is a funnel: filter fast on evidence, spend real time only on the final three. The whole thing should take days, not weeks.
1. Portfolio triage (10 minutes per candidate)
Open two live links — not screenshots, not mockups, not "coming soon." Click around on your phone. Does it load fast? Does the contact form work? Does the checkout flow make sense? Dead links or mockup-only portfolios get rejected immediately. Then run one of their sites through Google's PageSpeed test. A developer whose own portfolio scores poorly is telling you exactly how much they care about performance.
2. The written screen
Before any call, ask three questions in writing: what stack did you use on your last project like mine? What was the hardest bug you hit on it? How did handover work? Written answers reveal communication quality — and communication quality is the single best predictor of remote project success. The candidate who writes two clear paragraphs beats the smooth talker every time.
3. The paid test project (2–4 hours)
This is the highest-ROI step in all of hiring and almost nobody does it. Give finalists a small, real, paid task from your actual project — "build this hero section from the design, responsive" or "fix this one broken thing on our staging site." Pay their hourly rate. A $150–$300 test filters out the vast majority of mismatches before you've committed thousands. Interviews predict likability; a paid task predicts delivery.
4. The logistics check
One 10-minute call with a past client: did they ship on time, respond within a day, document the handover? And confirm availability in writing — "20 hours/week fore availability becomes a vague timeline.Five questions that reveal everything
On the call, skip the riddles and ask questions tied to your project. Strong candidates answer with specifics — file names, tools, timelines. Weak ones answer with adjectives.
"Walk me through the last project like mine you shipped. What broke at launch?"
Listen for concrete detail and honesty about problems. "Checkout failed because payment webhooks retried during deploy; I added idempotency keys and it held through 400 orders on Friday" beats "I'm passionate about quality code" by a mile. Everyone hits bugs; pros can describe theirs."Where will my code live, and what exactly do I get on the last day?"
Correct answer: your repository (GitHub/GitLab), on your hosting account, with admin credentials you own, deployment docs, and a short handover walkthrough. Hesitation here is a red flag the size of a billboard."How do you handle a change request mid-project?"
Good answer: a written scope update with the time and cost impact before anyone codes. Bad answer: "sure, no problem!" — the three most expensive words in web development."What happens if you get sick or disappear for two weeks?"
Expect: code pushed regularly to your repo, docs sufficient for someone else to take over, maybe a backup contact. "That won't happen" is not a plan."How do you make pages fast?"
Expect specifics: image optimization, caching, font strategy — and verify it by testing their past sites yourself. If their answer is vague, their sites will be slow.
Red flags: the 2026 edition
AI-generated portfolio. New this era: portfolios full of stunning mockups that were never built, or "case studies" describing sites that don't exist. If you can't click it live, it doesn't count.
Yes to everything. "Can you do the design, the copy, the custom booking system, and the mobile app?" — "Yes!" Nobody is excellent at all of those. The best candidates tell you what they'd not do themselves.
100% upfront — or "pay when you're happy." Both are traps. Full upfront removes all your leverage; "pay when happy" means no defined done-ness and endless revision loops. Milestones (below) or nothing.
They register everything in their name. Domain, hosting, analytics under the developer's email means you rent your own website. Create the accounts first, invite them second. Non-negotiable.
No questions about your business. A developer who never asks who your customers are, what action visitors should take, or how you'll measure success isn't building a business tool. They're typing.
Ghosting between messages. Days to reply during the sales process — when they're trying to impress you — means weeks during the project. Communication during hiring is the trailer; the project is the movie.
Proprietary everything. A custom CMS only they understand, hosting only they control, "our special system." Insist on standard platforms and tools any competent developer can take over.
Money: structure it so nobody can get hurt
A one-page agreement beats a handshake, and beats a 20-page template nobody reads. Six clauses cover 95% of freelance web builds:
Milestones tied to payments. Something like 20% kickoff, 40% on staging demo approval, 25% at launch, 15% retained 14–30 days for bug fixes. Every launch has bugs — the retained slice is what gets them fixed instead of invoiced.
Scope with acceptance criteria. Not "build website" but "homepage loads in under 2.5 seconds on mobile, contact form delivers to info@company.com, client can publish a blog post via the CMS unassisted." Testable beats "looks good."
IP assignment on payment. One sentence: all code, designs, and content produced under this agreement become the client's property upon payment. Without it, ownership defaults vary by jurisdiction. Get it in writing.
Your accounts from day one. Domain, hosting, repo, analytics registered to your email; the developer gets collaborator access.
A bug-fix window. 14–30 days of fixes included after launch. Contract it now or pay emergency rates later — typically 1.5–2x standard rates.
A termination clause. Either party exits with 7 days' notice; you pay for approved work delivered, with code pushed to your repo. This single paragraph prevents most hostage situations.
Treat any fixed bid sitting 50% below the cluster of other quotes as a risk flag, not a bargain — it usually means misunderstood scope, not efficiency. And if you're hiring through a marketplace, remember the platform fees (Upwork charges clients around 5% on top) when comparing against direct quotes.
Be a client developers want to work with
Uncomfortable truth: plenty of failed projects fail from the client side. The best developers quietly rank their clients. Here's how to get the A-team effort:
One decision-maker. "Let me run it by my partner / committee / cousin" repeated six times is how 3-week projects become 3-month projects. Nominate one person with the authority to say yes.
Fast, consolidated feedback. When work is shared for review, respond within a couple of days with everything at once — not a drip of contradictory notes over three weeks.
Content on time. The number-one timeline killer across our 150+ projects isn't coding. It's "we'll send the copy next week." Have your words and photos ready before the build starts, or pay the developer to create them.
Say the uncomfortable thing early. Something feels off in week two? Say it in week two. Small problems addressed early stay small; problems nursed in silence become blowups. And when they give technical advice, listen — it comes from pattern recognition across dozens of projects, not ego.
When it goes wrong anyway: the rescue playbook
Sometimes you did everything right and it still went sideways — the developer vanished mid-build, or delivered something unusable. (A growing share of our work is exactly this: finishing abandoned projects and AI-built prototypes properly.) Don't panic-hire the replacement. Instead:
Freeze and secure. Change passwords on hosting, domain, and admin accounts. Make sure you can access everything — this is when account-ownership chickens come home to roost.
Inventory what exists. What's done, what works, what's half-built? Get the code into a repository you control, even if it's messy.
Document, don't vent. Write a factual timeline: what was agreed, what was delivered, what was paid. You'll need it whether you negotiate, dispute, or just brief the next developer.
Hire the takeover specialist, not the cheapest bidder. Rescue work is a distinct skill — reading someone else's code, stabilizing it, then finishing. Ask candidates specifically about takeovers they've done. Expect it to cost more per hour than a fresh build; untangling is harder than building.
Quick answers
How much does it cost to hire a web developer?
For a typical small-business site: $500–$8,000 with a freelancer, $4,000–$18,000+ with a small studio or agency. Hourly, marketplace data puts North American mid-level developers around $70–100/hr and seniors $100–160/hr, with offshore rates substantially lower. Price tracks complexity and seniority far more than job title.
Freelancer or agency — which is better?
Neither, categorically. Freelancers win on price and direct communication for well-defined projects; agencies win on reliability, bench depth, and multi-skill projects. Match the hire to the job: bounded project with clear outcomes goes freelance; complex, ongoing, or "I cangt manage this myself" goes agency.
What's the safest way to pay a web developer?
Milestones tied to deliverables — e.g., 20% kickoff, 40% staging approval, 25% launch, 15% retained for bug fixes — with IP assignment on payment, all accounts in your name, and a written termination clause. Never 100% upfront.
Or skip the hiring lottery
Vetting developers is a skill — ours. If you'd rather hand the whole thing to a team that's shipped 150+ projects with a 100% on-time record and puts its prices on the page like a menu, book a free call. No pitch, just an honest take on what you'd need and what it would cost.
And if you already have a site and you're wondering whether the problem is the developer or the site itself — get a free roast. Real scan, honest score, no pitch. Sometimes the answer is "hire better." Sometimes it's "fix three things." We'll tell you which.
