Six weeks after launch, the phone call always sounds the same. The new site looks far better than the old one, everyone internally is pleased with it, and organic traffic has fallen off a cliff. A website redesign is one of the few marketing projects that can make a business look more professional and earn less money at the same time, and the reasons are almost always predictable in advance.

Why a website redesign loses traffic

Most of the damage happens in the plumbing rather than the design. URLs change and nobody maps the old ones to the new ones, so every page that had ranking history returns a 404. Page titles and meta descriptions get rewritten by whoever was closest to the CMS. Content gets trimmed for visual balance, which quietly removes the paragraphs that were actually answering search queries. Internal links vanish when a tidy new navigation replaces the messy old one.

The fix is unglamorous. Before anything is built, crawl the existing site and export every URL with its traffic and backlinks. Decide page by page what it becomes: kept, merged, redirected or deleted. Any URL that changes needs a permanent 301 redirect to its closest equivalent, not to the homepage, because a homepage redirect is treated as a soft 404 and passes almost nothing.

What a website redesign cost actually covers

Quotes for the same brief routinely differ by a factor of ten, which makes the market feel arbitrary. It is not. A template build with light customisation is a few thousand, because most of the work has already been done by whoever made the template. A custom design on a standard CMS with a handful of templates and a content migration usually lands in the mid five figures. Anything involving custom functionality, integrations with a CRM or an ERP, multiple languages or a genuine design system moves into six.

The three line items people underestimate are content, migration and QA. Writing or rewriting fifty pages is a real project with a real cost. Moving a thousand existing pages cleanly is engineering, not copy and paste. Testing across devices, browsers and assistive technology takes longer than the design phase in most builds. When a quote looks suspiciously cheap, one of those three is missing.

Speed is a design decision

Performance gets treated as a developer problem at the end of the process, when most of it was decided in the design file. A hero video, four custom fonts, a carousel and an animation library are all design choices with a measurable cost in load time. Core web vitals optimization is far cheaper when it starts with a brief that says the largest element on the page must be an image, not a video, and that the layout must not shift while fonts load.

Google's own documentation on Core Web Vitals lays out the three metrics that matter and the thresholds they are measured against. Read it before the design is signed off rather than after, when changing it means paying twice.

Design for the two pages that matter

Homepages absorb most of the argument in a redesign and most of the budget, and they are rarely where the money is made. For a service business the entry point is usually a service page or a blog post, and the conversion happens on a form that nobody spent much time on. Applying landing page best practices to those templates will do more for revenue than another round of homepage revisions.

Keep one clear action per page. Put the proof, whether that is client names, numbers or a case study, above the point where people stop scrolling. Cut form fields to what you genuinely need, because every optional field costs completions. None of this is new, and it is still where most redesigns leave money behind.

If the site is going multilingual

Redesigning and localising at the same time is efficient, provided the language versions are planned rather than bolted on. That means real URL structures for each language, correct hreflang, and translated navigation, forms, error messages and metadata rather than only body copy. A machine-translated page with an English contact form converts about as well as no page at all, and the practical steps in this guide to building localized landing pages that work cover what usually gets missed.

Measure it honestly

Record a baseline before launch: organic sessions, rankings for your top twenty terms, conversion rate per template and average load time. Compare like for like at thirty and ninety days. A dip in the first fortnight while search engines recrawl is normal. A dip that persists past six weeks is a technical problem, and it is nearly always redirects. The history of web design is full of beautiful relaunches that quietly cost their owners a year of traffic, and almost none of them failed for aesthetic reasons.