Warning Signs Your Website Build Is Going Sideways

Week seven of a twelve-week build, and there is still nothing to click. There have been calls. There was a mood board in February and a revised one in March. A PDF arrived last week with two homepage directions on it, both attractive, neither loading in a browser. What does not exist is a URL where the thing you are paying for can be seen on a phone.

That is the shape almost every troubled website project takes. The problem is not that the work is bad — nobody outside the studio can assess the work, because there is nothing to assess. Seven weeks of budget has become artefacts about the website rather than the website.

Here is a claim most studios would rather not see in writing: a website project that is going to fail is usually identifiable by week four, and the tell is never the quality of the design. It is the shape of the communication. Healthy builds leak evidence constantly — links, half-finished pages, questions about edge cases you had not thought of. Unhealthy builds produce presentations. We have covered the questions to ask before you sign; this is the other half, where the money is partly spent and the thing is drifting.

Normal mess versus real trouble

Custom builds are messy in the middle, and treating that mess as a warning sign makes you a difficult client for no benefit. Normal is week two running long because the content you promised arrived in week three — usually the client's fault, and the most common cause of a late launch. Normal is a section built, looked at and thrown away, or an API turning out worse than advertised. Normal is a date moving with somebody explaining why before you ask. Our post on how long a custom website actually takes covers where the weeks genuinely go.

Real trouble differs in kind. It is a widening gap between what you are told and what you can verify yourself. Dates move without explanation, or without being announced at all. Questions get answered adjacent to the question. Progress is described rather than shown. That last one is the master signal, and everything below is a case of it.

The five signals worth acting on

No staging link by week three

A staging site is a private, password-protected copy of your website on the internet, updated as work happens. It is not a deliverable or a reward for good behaviour — it is where the work lives, and on a normal build something exists at a URL well inside the first month. It will be ugly: placeholder text, the wrong logo, navigation that goes nowhere. The point is that the project has a location you can visit unannounced.

So ask directly: what is the staging URL? The answers that should worry you are "we'll share it once it's ready", "it wouldn't make sense to you yet", and "we work locally until sign-off". The first two protect you from information you have paid for. The third is a real workflow for one developer on a small job, but on a full build it means nothing has been deployed — and deployment is where the ugly surprises live. Meeting them in week ten rather than week three is how launches slip a month.

The design reveal instead of the checkpoint

The reveal is theatre: weeks of silence, then a polished presentation of a near-final design you are invited to react to. It feels professional. It protects the studio rather than serving you, because it collapses dozens of decisions into one.

You are shown a homepage and asked what you think, when what actually happened was forty choices about hierarchy, navigation labels, how the enquiry flow works, what the service pages inherit. If one is wrong, unpicking it means unpicking the composition it was baked into. So you say the polite thing, or ask whether the blue can be warmer, because that is the only size of feedback the format accepts.

The healthy alternative is small, ugly, frequent checkpoints. Sitemap agreed before anything is designed. One page reviewed before the other twenty. Wireframes — grey boxes, real headings, no colour — signed off before visual design begins. Each is cheap to reverse.

A scope question answered with an invoice

Watch what happens the first time you ask for something small. The healthy response is a conversation: here is what that changes, here is what it costs, here is what we would drop to fit it, your call.

The unhealthy responses are a variation order landing in your inbox with no discussion, or a cheerful "no problem, we'll add it" with no mention of time or money. Invoice-first usually means the studio is underwater on the fixed price and recovering margin through changes. Yes-to-everything is worse, because unrecorded scope comes out of the deadline, or out of the parts nobody is watching.

Radio silence between milestones

Silence is the most reliable predictor on this list, partly because it is the easiest to rationalise. They are heads-down. You do not want to be the client who emails twice a week.

Be that client anyway. A weekly written update — three lines, what moved, what is next, what is blocked — costs a studio ten minutes and is standard on any well-run project. If you have to chase for it, the project is not being managed; it is being remembered, and a project being remembered is one bad fortnight from being forgotten. Watch especially for calls rescheduled twice, then a Friday deadline passing without acknowledgement. Apologetic, then quiet, is the front end of what to do when your web developer disappears, and the window while they still answer emails is the useful one.

The same two pages, polished forever

By month two you can recite the homepage from memory. Six revisions in, it is genuinely getting better. Meanwhile the services pages do not exist, the contact form has never been tested, and nobody has mentioned the sixty-page blog archive that has to be migrated.

This is scope avoidance, and it is rarely deliberate. Refining a homepage is pleasant, visible and easy to show a client. Building a location page template, wiring a form to a CRM and migrating content are none of those things, so the pleasant work expands to fill the calendar. The diagnostic is a page inventory: every page with a status against it — not started, in progress, built, content loaded, reviewed. If that list does not exist, that is your finding. If most of it reads "not started" in week eight of twelve, so is that.

What you should have in hand at each stage

None of those signals mean much alone. What makes them legible is knowing what a build normally produces as it goes, so you can tell missing from merely unglamorous.

How to intervene without blowing the relationship up

Most troubled projects are recoverable, and the intervention that works is boring: convert vague concern into a specific, dated, written request. Not a complaint — a request. Send one email asking for four things by a named date. The staging URL. The page inventory. A revised delivery date with the reasons it moved. A standing fifteen-minute call in the same slot each week until launch.

That email does two jobs. If the project is healthy and just badly communicated — which it often is — you will get all four within days and the relationship improves, because most studios would rather be asked clearly than guessed about. If it is genuinely in trouble, the response tells you that too: defensiveness, a counter-proposal to discuss it on a call instead, or three of the four arriving with the staging link conspicuously missing. Keep it in writing and keep it calm; a record of reasonable requests met with vague replies is worth more later than an upset phone call.

The sunk cost maths, done honestly

If that lands badly, the question becomes whether to continue — and people get it wrong in a specific direction. They count what they have already spent, which is the one number that should not enter the decision. That money is gone either way. The only comparison that matters is cost to finish here versus cost to finish elsewhere, and the second is lower than people assume.

Work it on your own figures. Say a $16,000 build, sixty per cent invoiced, so $9,600 spent. Continuing costs the remaining $6,400 plus however many weeks of drift are left — and if the last four weeks produced nothing verifiable, the honest estimate for the next four is also nothing. Starting again costs a fresh build, but not from zero: the sitemap, the content, the photography and the decisions about what the site must do all transfer, and those are a meaningful slice of any build. Do that arithmetic on paper with your numbers. It usually lands closer than the fear suggests.

The other input is time. If there is a hard date — a rebrand launch, a trade show, a seasonal peak — a project that has missed two internal deadlines will miss that one too, and missing it with a new studio starting cleanly often beats missing it with the current one.

Leaving with assets you can actually use

What you take determines whether the next build is a fresh start or a salvage job. Ask for these before you signal that you are leaving, not after.

The design files in editable form. The code repository, or a full export of it. Any content written for the project, including drafts. Photography and illustration with the licence terms attached. And administrative access, in your own name, to every account created on your behalf — domain registrar first, then hosting, analytics and any third-party service wired into the site. Check that domain now regardless: registered in a studio's name rather than yours, it is the most expensive thing on the list to unwind. Our website handover checklist covers what should change hands and in what order.

Two things about contracts. Ownership of code and design usually transfers only on final payment unless your agreement says otherwise, so what you are owed mid-build is whatever the contract says — read it before the conversation, not during it. And the ACCC administers an unfair contract terms regime covering standard-form small business contracts, under which including an unfair term can carry penalties. We are not lawyers and this is not legal advice, but a clause that looks outrageous is not automatically enforceable, and an hour with a solicitor is cheap against a write-off.

What to do this week

Being paranoid about a website build is not the goal — plenty of projects that feel wobbly at week six launch fine. The useful posture is refusing to accept description in place of evidence, early, while the fix is cheap.

So: find the staging URL, or find out why there isn't one. Ask for the page inventory. Get the next delivery date in writing with the reason it changed. Confirm the domain is registered in your business's name. Four things, one email, an afternoon at most. If the answers come back clean, you have bought a better-run second half. If they don't, you have found out in week six rather than week fourteen.

If you are already past that point — mid-build, stalled, holding a half-finished site and a full invoice — it is more common than it feels and more recoverable than it looks. We take on that kind of rescue as part of our custom web design work, and the first step is an honest read on what exists. Book twenty minutes or call us on 0421 933 907, bring the staging link if you have one and the contract if you don't, and we'll tell you whether it is worth finishing.