How to hire a web design agency: the questions that separate the good ones

Most people hire a web design agency two or three times in a career. The agency has done this hundreds of times. That asymmetry — not price, not talent — is why website projects go wrong.

You can close most of it with about a dozen questions. Not technical questions; you don't need to evaluate a framework. Questions about ownership, scope, process and what happens when something changes, because that's where the money actually goes.

This is the list we'd want a client to bring to us, including the answers that should worry you.

Decide what you're buying first

Before comparing anyone, be clear which of these you actually need. Agencies are usually good at one and adequate at the rest, and mismatching here causes more disappointment than any skill gap.

Before any of this, be sure a new site is what you need at all — quite often the honest answer is a repair rather than a rebuild. The most common expensive mistake is hiring for the first and expecting the second. A beautiful five-page site does not rank, has never ranked, and no amount of "SEO-friendly" in the proposal changes that. Ranking is a content and structure commitment, not a build feature.

The questions that actually separate agencies

Twelve questions. The answers matter more than the portfolio.

Question 12 is the one we'd weight most heavily. An agency that agrees with every idea in the brief is either not listening or planning to bill for all of it. The ability to say "you don't need that, and here's what to do instead" is the clearest signal you're buying judgement rather than production.

Question 11 has a specific trick in it. Ask for a client from more than a year ago. Recent clients are still in the honeymoon; the year-old client knows whether the site was maintainable, whether support calls got answered, and whether the thing actually did what it was sold to do.

Ownership is the clause that matters most

This is where firms get genuinely trapped, and it is entirely preventable with one conversation before signing.

The failure mode is rarely malicious. An agency sets up a domain, hosting and analytics on their own accounts because it's faster during the build. Nobody writes down what happens later. Two years on, the relationship ends and the client discovers they cannot move without the agency's cooperation — and if the agency has gone quiet, folded, or simply become slow, the client's website, email and historical analytics are hostage to an account nobody can access.

Non-negotiables: the domain registrar, DNS, hosting, analytics property and any ad accounts should be registered in the client's name, on the client's billing, with the agency added as a user. That's it. Any agency that resists this arrangement is telling you something useful.

Source code is a separate conversation and worth having explicitly. Ask directly: at the end of this engagement, do we get the code, and in what form? Design source files are frequently excluded by default — not unreasonably, but you should know before rather than after.

How agencies charge, and what each model hides

None of these is dishonest. They move risk to different places, and knowing which risk you're holding is the point.

Fixed price puts the risk on the agency, so the price includes a margin for that risk. In exchange you get a number you can budget against and, more valuable, a forced conversation about scope before work starts. The cost is rigidity: anything not in the scope becomes a change order, and if the scope was written badly you'll be paying for change orders all project.

Time and materials puts the risk on you. It suits genuinely exploratory work where nobody can scope it honestly up front. It requires trust and active oversight — ask for a weekly burn report, and agree a not-to-exceed figure.

Retainer buys continuity. It's the right structure for ongoing work and the wrong one for a single build. The failure mode is paying for months where nothing needed doing, or the reverse — quiet scope expansion until the retainer covers three times what it was priced for. A retainer needs a written scope as much as a project does.

A note on cheap quotes. When one is dramatically lower than the others, it usually means a smaller scope rather than a better deal — fewer templates, stock copy, no migration, no testing, no training. Compare what's included before comparing the numbers, and if you can't tell what's included, that's the finding.

How to run the process without wasting three months

The selection process itself is where a lot of goodwill and time gets burned, usually because it's modelled on procurement rather than on hiring.

Skip the long RFP. A twenty-page request for proposal produces twenty-page responses written by people whose job is writing responses. It selects for proposal-writing capacity, which correlates with agency size and almost nothing else. A two-page brief that states the business problem, the constraints, the rough budget range and the decision timeline will get you better answers from better firms.

Name a budget range. The most common objection is that naming a number means everyone quotes to it. In practice, withholding it wastes everyone's time: agencies either guess high and get eliminated, or guess low and scope something you didn't want. A range lets a good agency tell you what's achievable inside it — or tell you honestly that it isn't, which is information you want early rather than after three rounds.

Shortlist to three. Beyond three, you're not comparing, you're collecting. Each additional conversation costs you hours and costs them days, and past three the differences stop being legible.

Consider paying for discovery. For anything substantial, a paid discovery engagement — a week or two of research, architecture and a scoped plan — is the single best de-risking move available. You get a real plan you own, produced by people doing actual work rather than sales. You can then take that plan to whoever you like, including the agency that wrote it. Firms that resist this are telling you they'd rather sell a scope than define one.

Talk to the person who'll do the work. Not only the account lead. Fifteen minutes with the designer or developer who'll actually be on it tells you more than an hour of agency positioning.

Decide who decides. Internally. Before you start. Website projects stall most often because three stakeholders have opinions and none has authority, and no agency can fix that from outside.

What a good proposal actually contains

Proposals vary enormously in length and hardly at all in what matters. These are the sections that indicate someone has thought about your project rather than reformatted a template:

A restatement of the problem in their words. If the first page describes your business back to you and gets it right, they were listening. If it opens with the agency's history and awards, you're reading a brochure.

Scope expressed in countable things. Number of unique page templates, number of pages, number of revision rounds, how many practice areas or product categories. "A modern, responsive website" is not a scope; "eight unique templates across twenty-two pages, two rounds of revision" is.

An explicit exclusions list. The strongest signal in any proposal. It means someone has thought about the boundary, and it protects both sides.

Client dependencies, named. Good proposals say what they need from you and by when — photography, content approval, access to accounts, a decision-maker. Projects overrun because clients are slow far more often than because agencies are, and a proposal that ignores this is planning to blame you later.

Assumptions written down. "Assumes existing brand assets are available in vector format." "Assumes fewer than 50 existing URLs to migrate." Assumptions are where fixed prices quietly become variable ones, so seeing them listed is a good sign, not a worrying one.

What happens after launch. Support scope, response times, what's billable, who to call.

A price, not a range. For fixed-price work, if the proposal ends in a range, the scope isn't settled — and an unsettled scope is the actual thing you're being asked to sign.

If a proposal is missing three or more of these, the gap isn't a formatting preference. It's the part of the project nobody has thought about yet, and it will surface later as a change order.

Red flags, ranked

Not all warning signs are equal. In rough order of how much trouble they predict:

1. Ownership of accounts stays with the agency. The one that can genuinely trap you.
2. No written scope of what's excluded. Guarantees a change-order relationship.
3. Guaranteed rankings. Nobody can guarantee a position on a search engine they don't own. This is either ignorance or a lie, and both disqualify.
4. Support is undefined. "We're always here" is not an agreement.
5. The pitch team isn't the delivery team, and nobody will say who is.
6. No questions about your business. An agency that doesn't ask who your customers are is selling production, not judgement.
7. A portfolio you can't visit. Ask for live URLs. Sites that no longer exist, or were quietly rebuilt by someone else, are common.
8. Proposal is all deliverables and no outcomes. Twelve pages of what you get, nothing about what it's supposed to achieve.
9. Pressure to sign this week. Discounts that expire are a sales tactic, not a scheduling reality.

None of these are automatically fatal — small agencies are often loose on documentation and excellent at the work. But each one should be raised and answered before signing, and how they handle the question tells you as much as the answer.

Green flags worth paying more for

The signals that reliably predict a good engagement:

• They push back on part of your brief, with a reason.
• They ask what happens after launch — who publishes, who maintains, who measures — before they talk about design.
• They can describe a project that went badly and what changed as a result. Everyone has one. Only some will tell you.
• They want to talk to your customers, or at least to your salespeople.
• The proposal contains a section on what's out of scope, written plainly.
• They tell you a smaller version would be better. Extremely rare, disproportionately predictive.
• They ask which platform you're on and why, rather than arriving with one — the right answer depends on your constraints, not their preference.

Before you sign

A short checklist for the contract itself:

The exit clause is the one people skip and later wish they hadn't. You are not being pessimistic by asking how the relationship ends; you are asking a professional question, and professionals answer it comfortably.

Agencies, freelancers and in-house: which one

Worth answering plainly, because the honest answer isn't always "hire an agency."

A freelancer is often the right call for a small brochure site with a clear brief. You get the person who does the work, usually at a lower rate, without agency overhead. The risks are capacity and continuity: one person gets ill, takes another contract, or moves on, and there is no bench. Mitigate it by insisting on your own accounts, your own repository, and a documented handover — the same list as above, which matters more with a single practitioner, not less.

An agency is worth the overhead when the work spans disciplines — strategy, content, design, build, measurement — or when continuity matters more than rate. You're paying for a team that survives an individual leaving, and for someone whose job is making sure the project actually progresses. If the project is genuinely one discipline, you may be paying for coordination you don't need.

In-house makes sense once the work is continuous rather than a project: constant publishing, frequent campaigns, a product that changes weekly. The trap is hiring one generalist and expecting design, development, SEO and content from them. That person exists but is rare and expensive, and burning one out is a common and costly failure.

The hybrid that usually works best for small and mid-size businesses: an external partner for the build and the structural work, someone internal who owns publishing and can make small changes without raising a ticket. It keeps the expensive expertise external and the everyday capability internal, which is the split most organisations can actually sustain.

If you take one thing from this

Hire for judgement, not for pixels. Portfolios are easy to assemble and mostly show what a client approved, not what the agency recommended. The questions above are designed to surface whether someone will tell you the truth when it costs them a line item — because that's the part you can't evaluate from a gallery, and it's the part you're actually paying for.

If you'd like a second opinion on a proposal you've already received, send it over. We'll tell you what's missing from it, including the cases where the honest answer is that it looks fine and you should sign it.