Most website rebuilds change the design and nothing else.
A rebuild is worth paying for when it changes what the site does. XBA rebuilds to a fixed prescription: one outcome stated above the fold, one page for each thing you sell, proof sitting beside the claim it supports, and speed and accessibility measured rather than promised.
Personal reply from Andrew within one business day.
What does a website rebuild actually change?
A website rebuild is a restructure of what a site says, in what order, and across how many pages — not a re-skin of the same content in a newer template. The distinction decides whether anything measurable changes afterwards.
Most rebuild quotes are priced as design work, which is a reasonable thing for a design studio to sell and is also why most rebuilds change the design and leave everything else where it was. The new site launches, it looks considerably better, and six months later the enquiry volume is what it was before. Nobody was dishonest; the work simply never touched the thing that was failing.
Four structural failures account for most of it, and they are all visible from the outside in about two minutes. The homepage describes the company rather than the problem, so a visitor who has a problem has to translate. A single Services page carries every offering, so no page owns any search term and the page reads as a list rather than an argument. Proof is filed under a Work or Gallery tab that nothing links to from the claims it would support. And five different calls to action compete on the same screen, which reliably produces fewer of all five than one would have produced on its own.
Answer-first structure matters more than it used to. A page that opens with a direct, self-contained answer to the question it targets is easier for a person to skim and dramatically easier for a language model to lift accurately. Burying the answer three paragraphs down, after the scene-setting, is now a real cost rather than a stylistic preference — the assistant summarising your page will take whatever is at the top and present it as what you do. That is the mechanism the SEO and AEO half of this pillar exists to work with, and the two are usually done together for that reason.
Speed is not a score. It is whether the thing a visitor came for has appeared before they give up, on the connection and the device they actually have, which in New Zealand is frequently a phone with two bars somewhere. Most of the weight on a slow site is not the design — it is script that runs before anything paints, fonts loaded from somewhere else, and images shipped at four times the size they display at. Those are structural decisions too, which is why they belong in a rebuild rather than in a tuning pass afterwards.
Accessibility is the same argument. Colour contrast, real heading order, focus states that exist, and text that stays readable when someone zooms are not a compliance overlay applied at the end; they are consequences of decisions made at the start. This site pins every colour token to a measured contrast ratio against the darkest background it is used on, and the deploy fails if a page regresses below the threshold. That is not unusual rigour — it is just cheaper than fixing it later.
What does the restructure look like?
The same content, in a different number of pages and a different order. Almost nothing here is a design decision, which is the point.
Six pages, and no page that a specific buyer question can be answered from. The Services page carries every term the business would want to rank for, which means it competes with itself for all of them.
More pages, each narrower. The homepage stops trying to be the whole site and becomes a router; every offering gets a URL it can own; and the number of things a visitor is asked to do drops to one.
The counter-intuitive part is that the rebuild usually produces more pages rather than fewer, and that more of them are narrow. That runs against the instinct to simplify, and it is the single change that does most of the work: a page about one thing can answer one question completely, rank for one term, and be quoted for one claim. A page about nine things does none of those, no matter how well written it is.
The second thing worth noticing is that proof moves. On the left it lives in a gallery that is only reachable from the navigation, which means it is seen by people already convinced enough to go looking. On the right it sits directly under the claim it evidences, where it is read by the people who are not yet convinced — which is the only audience for whom proof does anything.
What does the prescription look like on a real site?
A recent XBA rebuild for a New Zealand professional services firm, described by its structure. The client is not named and no identifying detail appears here — what is worth showing is the shape of the decisions, not whose site it is.
-
01
The headline states the outcome the buyer wants
Not the firm, not the disciplines, not the founding year. The sentence a customer would use for the thing they are trying to get done, with one primary action beside it.
-
02
The facts buyers check are given, not hidden
Coverage area, disciplines, how long the first conversation takes and what it costs — set out as a scannable panel near the top rather than scattered through prose on three different pages.
-
03
Proof carries the technical reasoning
Each project entry states the constraint, the decision taken, and what happened — not a photograph with a caption. This is what makes a portfolio persuasive to a technical buyer and quotable to an assistant.
-
04
The audience router replaces a generic pitch
A business serving several distinct customer types cannot address them all in one paragraph. Letting a visitor pick the description that sounds like them is more useful than trying to write for the average of them.
-
05
Objections handled on the page
The four-step process section and the plain-English FAQ exist to answer the questions that otherwise become the first phone call. Answering them in public also makes them available to an assistant.
No before-and-after figures are published here, deliberately. A fair comparison on a rebuild needs a full season of data either side and a question set that did not change in between, and quoting anything sooner is marketing rather than measurement. When numbers exist and are honest they belong on a case study with the method attached, the way the field service case study does it — not as a percentage on a services page.
What can be said without qualification is what the structure does. Every offering has a page that owns its term. Every claim on the homepage has proof reachable from directly beside it. Every page answers its own question in its first paragraph, which is what makes the site quotable rather than merely present. That is the whole prescription, and it is verifiable by reading the page you are on — this one was built to it too.
How much of a rebuild do you actually need?
Three genuinely different pieces of work get called a rebuild. They cost different amounts, fix different things, and the wrong one is a common and expensive mistake.
| Template refresh | Structural rebuild | Rebuild + presence programme | |
|---|---|---|---|
| What actually changes | Appearance. Same pages, same order, same words, newer design. | The page set, the order, and what each page is for. Most copy is rewritten. | All of that, plus the technical, content and entity work that decides who arrives. |
| Who it suits | A site whose structure is already right and whose design has simply dated. | A site that gets traffic but converts poorly, or cannot rank because no page owns a term. | A business whose problem is that too few people arrive at all. |
| What it fixes | Credibility on first impression. Real, and narrower than it sounds. | What the site says, who it says it to, and what it asks them to do. | The whole of stage one and stage two — being found, then being convincing. |
| What it will not fix | Anything to do with being found. A better-looking invisible site is still invisible. | Traffic. A well-structured site nobody links to or searches for stays quiet. | A business that cannot yet say in one sentence what it sells and to whom. |
| Where the effort goes | Design and implementation. Predictable, and the shortest of the three. | Deciding what each page is for, and writing it. The building is the easy part. | The above, plus the ongoing half — publishing and measuring over months. |
| How you tell it worked | Subjectively, mostly. Be honest that this is the weakest signal of the three. | Enquiry quality and conversion on the pages themselves. Visible within weeks. | Rankings, assistant citations and enquiry volume, measured monthly against a fixed set. |
The common expensive mistake is buying the first column while describing the third column’s problem. It happens because a refresh is the easiest thing to quote and the easiest thing to approve, and because the failing thing — nobody arrives — is invisible in a way that a dated design is not. The audit exists partly to catch exactly this, and it is free precisely so that catching it costs nothing.
When is a rebuild not worth it?
More often than a web page selling rebuilds would normally admit. Four situations come up repeatedly, and in each of them the money is better spent somewhere else or not at all.
The structure is already fine. If the pages are the right pages, each one owns its term, and the problem is simply that not enough people arrive, a rebuild will change very little. That is a search and citation problem, and it belongs in the other half of this pillar. Rebuilding first is a common and satisfying way to spend a budget without moving the number that was bothering you.
You cannot yet say what you sell in one sentence. The rebuild will surface this within about an hour of starting, because the first question is what the homepage headline should say and there will not be an agreed answer. That is a positioning problem and it is worth solving before anyone is paid to write pages against it — otherwise the site becomes a permanent record of an argument that was never settled.
The business is about to change what it sells. A site built around the current offering, mid-way through a pivot, gets rebuilt again in a year. Waiting is frequently the cheapest advice available, and it is given more often than the commercial incentive here would suggest.
Nobody will maintain it. A site is not finished at launch; offerings change, proof accumulates, and pages that were true stop being true. If there is no one who will update it, the structure decays back toward the left-hand column of the diagram above within a couple of years. Building something simple enough to be maintained is a better outcome than building something impressive that nobody touches.
Not sure whether the problem is the site or the traffic? The audit is free and the answer is sometimes that the site is fine.
Book a free process auditQuestions about a website rebuild.
What is a website rebuild, as opposed to a redesign?
A redesign changes how a site looks. A rebuild changes what it says, in what order, and across how many pages — the page set is decided again from scratch, most copy is rewritten, and the design follows those decisions rather than driving them. The distinction matters because a redesign will not move enquiry volume or search position, and it is frequently sold to businesses whose actual problem is one of those two.
Will a rebuild lose the search rankings we already have?
Not if the redirects are done properly, and this is the part that most often goes wrong. Every URL that currently exists and has any inbound value gets a permanent redirect to its nearest equivalent on the new structure, mapped before launch rather than discovered afterwards from an error report. Where a rebuild splits one page into four, the old URL points at whichever new page best answers what it used to answer. Rankings usually wobble for a few weeks regardless, because the pages have genuinely changed; a redirect map is what stops that being permanent.
Do we get a CMS we can edit ourselves?
It depends on who is actually going to edit it, and that question is worth answering honestly before the build rather than assuming the answer is yes. A site whose content changes weekly needs an editing interface. A site whose content changes twice a year often does not, and a content system added for its own sake becomes an unpatched dependency and a monthly fee attached to a login nobody remembers. Either way the code and the hosting are yours, and moving elsewhere later does not require a conversation with XBA.
How many pages should a small business site have?
One for each thing you sell, one for each significant question buyers ask before they enquire, and one piece of proof for each major claim — which for most small businesses lands somewhere well above the four or five pages they currently have. The wrong instinct is to consolidate. A page covering nine offerings cannot rank for any of them or answer any single question completely, so the consolidation that feels tidy is usually the thing suppressing the site.
Should the rebuild and the SEO work happen together?
Usually yes, because they are the same decisions made once rather than twice. The page set a rebuild produces is exactly what a keyword-to-page map produces, and doing them separately means either the map is drawn against pages that are about to change or the pages are built without knowing what they should own. Where they are genuinely separable is when the existing structure is already sound — in that case the rebuild is skippable and the money belongs entirely in the SEO and AEO half.