Hiring Web Help Without Getting Burned
The nervous buyer's protection page: what web work fairly costs, the red flags that predict pain, the ownership questions to settle before any work starts, and how to compare quotes that look nothing alike.
Buying web help is a trust purchase made in a fog: quotes for "a website" arrive at $800 and $18,000 for reasons nobody explains, every provider speaks a different dialect of jargon, and the industry's incentives — retainers that outlive their usefulness, platforms that resist exit, deliverables that are hard to evaluate until long after payment — do not all point your way. Practices know how this feels from the other side of their own desk: it's how their clients feel. This page is the protection a good advisor would give a friend.
The four protections, before any work starts: ownership settled in writing (domain in your registrar account, site and content yours on exit, all accounts under the practice — the foundations rules, made contractual); scope you can point at (pages, functions, and the finish line described concretely enough that "done" is checkable); price shape understood (what's fixed, what recurs, what triggers extras); and the exit priced while entering ("if we part ways in two years, what leaves with me, and what does leaving cost?"). A provider comfortable with all four is probably one you can trust; discomfort with any is the forecast.
What things fairly cost
Honest ranges are wide because the work genuinely varies — but the shape of fairness is describable. A practice website's cost lives in three layers: the build (from around a thousand for a competent templated site, to mid four figures for custom design and conversion-grade pages, to five figures where real strategy, content, and integrations ride along — what moves the number is scope and seniority, and a quote should say which); the recurring (hosting and upkeep from trivial to a few hundred monthly depending on who carries updates — recurring fees deserve a named deliverable, not vibes); and the optional ongoing (marketing retainers — judged on evidence, always separable from the build). The comparison discipline for wildly different quotes: make each one answer the same scope sheet — what's included, what's excluded, what's extra, what recurs — and much of the mystery difference evaporates into included-versus-not.
The red flags, from the field
Patterns that reliably precede the hostage stories and the abandoned rebuilds: "we'll register the domain for you" (the classic — always in your account, them invited in); proprietary lock-in unmentioned — ask "what happens to the site if we leave?" and grade the squirm; guarantees of rankings or AI recommendations (nobody honest sells certainty); pressure pricing ("this quote expires Friday"); no questions about your business — a provider who quotes before asking what the site must do is selling inventory, not solving problems; and a portfolio without reachable clients — checkable proof applies to vendors most of all: two references, called, asking "how was the exit, or the fee dispute, or the deadline miss?" — the questions references can actually answer honestly.
And one green flag worth naming, because it's the inverse of the industry's incentives: the provider who tells you what not to buy. "You don't need that yet" from a person you're trying to hand money to is the most credible sentence in commerce.
Questions practices actually ask
Freelancer or agency? Structure matters less than the four protections: a solo professional with clean ownership habits beats an agency with a platform trap, and vice versa. The real trade is capacity and continuity (agencies) versus cost and directness (freelancers) — the exit questions apply identically.
Is the cheap option ever the right option? Genuinely, sometimes: a modest, owned, current site that does its five jobs beats an expensive one that doesn't. Cheap becomes expensive only through lock-in, staleness, or rebuilds — all of which the protections screen for regardless of price.
Should I sign a monthly retainer? For named recurring work (updates, maintenance, agreed improvements): reasonable. For "ongoing optimisation" unattached to deliverables: run the pause test after a quarter and let the numbers decide.
What if I've already got a provider and never asked any of this? Ask now, framed as housekeeping — every question is routine to a clean operation. The answers either confirm you're fine or start the untangling early, and both are better than finding out during a falling-out.
Part of The Foundations — the ground everything stands on.