Most businesses that commission a website redesign do not know, at the moment the brief is signed, what their current conversion rate is. This is not unusual. It is also the reason most redesigns fail.
A redesign without a baseline is a renovation project without a floor plan. You can change a great deal and still not know, when it is finished, whether anything improved. The new site looks better. The old site looked dated. Both statements can be true while the business is still generating the same number of leads it was six months ago, at the same cost, with nothing meaningfully shifted in the numbers that actually drive revenue. The agency has delivered what was briefed. The business has spent a significant sum and is not sure what it received.
This pattern is visible across the NZ market. The businesses most likely to experience it are not the ones that cut corners on budget or chose the cheapest agency they could find. They are often the businesses that invested properly, worked with a capable team, and still ended up with a site that performs essentially the same as the one it replaced. Understanding why requires looking at where redesign projects go wrong before a single wireframe is drawn.
The brief starts from the wrong question
The most common website brief in New Zealand goes something like this: we want to modernise the look and feel, improve the user experience, and better represent where the brand is today. Sometimes there is a secondary note about mobile usability or page speed. These are reasonable things to want. They are also almost entirely unmeasurable.
What does "modernise" mean in a way that connects to revenue? What does "better represent the brand" achieve commercially? Neither question has an obvious answer, which means neither has an obvious success condition. The project is set up to be evaluated on one question: does this look and feel right to the people in the room. Objective commercial criteria rarely feature.
Agencies respond to the brief they are given. If the brief asks for a better-looking website, a competent agency will deliver a better-looking website. But looking better and performing better are not the same thing. A site can pass every internal review, from design approvals to content revisions to stakeholder sign-offs, and launch to no measurable improvement in enquiry rate, conversion rate or revenue per visitor.
The redesigns that consistently produce measurable results start from a different question. Not "how do we want the site to look" but "what specific commercial problem are we trying to solve." One NZ business we worked with did not come to us with a general brief about modernising their Shopify store. They came with a specific problem: their conversion rate was well below what comparable stores achieve, and their logistics workflow was creating operational friction. We rebuilt around those two problems. Conversion rate increased by 103%. Sales increased by 28% in the six months after launch. The design improved considerably. But the design was the outcome of solving a specific problem, not the starting point.
Content migration is not content strategy
There is a moment in almost every redesign project where the question of content comes up. The answer is usually some version of: we will keep what we have and tidy it up where needed.
This is how the fundamental problems of the old site survive the redesign.
If the old site was not converting, the content on the old site was part of the reason. The way the value proposition was framed. The order in which information appeared on the page. The copy on product pages, service descriptions, contact prompts. These are not peripheral elements of a website. They are the mechanism through which a visitor decides whether to act or leave. Redesigning the visual wrapper around weak content produces a more polished version of a site that does not convert.
The agencies that build sites which perform treat content as an architecture problem, not a production task. Before a wireframe is drawn, there are decisions to be made about what a visitor needs to understand, in what order, and what action they should take next. These decisions shape the page structure, which shapes the design. Reversing that sequence, designing first and dropping content in later, produces pages that look considered but are structured around aesthetic preferences rather than customer behaviour.
The copy itself is a separate problem. Most business website copy is written from the inside out: what the company does, what they offer, how long they have been doing it. The copy that converts is written from the outside in: what the customer is trying to solve, what they are afraid of getting wrong, what they need to understand before they trust you enough to make contact. Porting the old copy into the new design retains all the inside-out thinking that made the old site underperform.
The stakeholder problem
Website redesigns are expensive, which means senior people are involved in approving them. Senior people have strong opinions about how the business should be presented. These opinions are usually shaped by years of internal perspective, not by how customers actually use the website.
Navigation structures get rearranged to reflect the internal organisation of the business rather than the way visitors think about their problems. Hero messaging gets revised to lead with things the leadership team values: awards, years in operation, number of staff. Not the things that stop a visitor from leaving in the first few seconds. Service pages get written to reflect the way the sales team talks about the offering, not the way a customer at the beginning of a buying process would describe their own need.
None of this is deliberate. It is the natural outcome of giving approval authority to people who have absorbed years of internal perspective and have limited visibility into how visitors actually behave on the site. The practical solution is to base structural and copy decisions on evidence: session recordings, heatmaps, enquiry form data, sales call notes. Hard data is a more effective counterweight to strong opinions than judgment alone.
In practice, this means having a direct conversation before the design phase starts. What do we actually know about how visitors use the current site? Where do they drop off? What questions do they ask before they buy or enquire? If the answer is "we don't really know," the first step is finding out. Building a redesign on assumptions about customer behaviour is expensive. Building it on data is not significantly harder and produces considerably better results.
Speed is not optional
Page speed deserves its own section because it is the redesign failure mode that is most consistently underestimated and most consequential once the site is live.
New visual design tends to introduce heavier pages. More animation. Larger image files. Additional third-party scripts for tracking, chat widgets and personalisation tools. Design-led agencies rarely measure Core Web Vitals as part of their delivery criteria. The result is a site that looks substantially better than the old one and loads measurably slower. Google registers this immediately. Rankings that took years to build can erode in a matter of months.
For NZ businesses operating in a relatively small market, often on constrained hosting, competing for a limited pool of high-intent search traffic, page speed matters more than most agencies acknowledge. A half-second improvement in Largest Contentful Paint can move a well-optimised page from the third organic result to the first on a category with meaningful search volume in this market. A full-second regression, caused by an animation library added late in the build that nobody benchmarked, can undo two years of SEO effort before the business notices what happened.
Studios that build sites which hold their rankings treat performance as a design constraint from the beginning. Image format decisions, load sequencing, script management and hosting environment are not afterthoughts addressed in the final week before launch. They are part of the architecture. A site that scores well on Core Web Vitals at launch and maintains that score under real traffic is not a bonus outcome. It is the minimum standard for a redesign that does not undo what preceded it.
What the sites that perform have in common
The redesigns that move the needle do not have dramatically different budgets to the ones that do not. They are not necessarily more complex or more visually ambitious. The consistent difference is in how the project is framed before work begins.
A specific commercial problem. A current baseline: conversion rate, enquiry rate, average order value, whatever metric the business is actually trying to move. A target that defines success in terms both parties can measure, agreed before a wireframe is produced. A content approach built around what the customer needs to understand, not what the business wants to say. And a technical delivery standard that treats page speed and accessibility as first-class requirements rather than items on a checklist that nobody checks.
One NZ business that designs and sells premium saunas came to us with a specific challenge: their existing site was generating enquiries at roughly the rate you would expect for the category, but they knew they were losing visitors who were genuinely interested. The brief was not "redesign the website." It was "fix the conversion rate." The Webflow site we built for them now converts enquiries at 4.6% against an industry benchmark of around 2.7%. The design is strong. The design is also the direct result of solving a specific problem, not the starting point.
The businesses that walk away from expensive redesigns frustrated are, with very few exceptions, the ones whose brief was built around appearance. The businesses that see real commercial improvement from their investment are the ones whose brief was built around performance, who held the agency to that standard throughout the project, and who measured the result against a baseline they had established before work started.
The website you are looking at right now is the brief made visible. If the brief was vague, the site will be too.
