Redesigning a Website Without Losing What Worked

Redesigns are seductive. The current site feels dated, the team is tired of looking at it, and a fresh start promises to fix everything at once. I've felt that pull many times, and I've also seen what happens when a team gives in to it completely: a beautiful new site launches, and a few weeks later, leads drop, search traffic falls, and nobody can explain why.

The problem is rarely the new design itself. It's that the redesign threw away things that were quietly working — pages that ranked well, phrases that converted, paths that customers relied on — without anyone noticing they were there.

Working on a website redesign for a solar energy company sharpened my approach to this. Here's how I try to redesign confidently without losing what worked.

Start with an audit, not a moodboard#

The natural starting point for a redesign is visual: collecting references, exploring new directions, imagining the new brand expression. That comes later. The first step is understanding what the current site is actually doing.

I look at:

Analytics. Which pages get the most traffic? Where do visitors enter and leave? Which pages lead to inquiries or sales?
Search performance. Which pages rank for valuable search terms? Which queries bring qualified visitors?
Conversion paths. What do people typically view before they contact the company? Which forms and calls to action actually get used?
Sales and support input. What questions do prospects ask before buying? What content do salespeople send to leads?
Customer feedback. What do customers say confused or reassured them when they were deciding?

This audit almost always surprises someone. On the solar project, a fairly plain page explaining how financing and incentives worked was one of the strongest drivers of inquiries. It wasn't pretty, and it wasn't featured prominently, but it answered the biggest question homeowners had: can I actually afford this? That page needed to survive the redesign — and arguably deserved to be promoted.

Separate what's dated from what's broken#

Not everything about an old site is a problem. It helps to sort issues into categories:

Dated — the visual style feels old, but the content and structure work fine.
Broken — things that actively hurt the experience: slow pages, confusing navigation, missing mobile support, inaccessible content.
Missing — things the site should do but doesn't: answer key questions, support a new product line, build trust for a new audience.
Working — things that are performing well, regardless of how they look.

A redesign should fix what's broken, add what's missing, update what's dated, and protect what's working. When teams treat the entire site as "dated," they end up rebuilding things that didn't need rebuilding and losing things that did their job.

Protect your search equity#

Search traffic is one of the easiest things to destroy in a redesign and one of the slowest to rebuild. Pages that have ranked for years carry accumulated value: links from other sites, history, relevance signals.

The basics matter enormously:

Keep URLs where possible. If a page ranks well, keeping its address avoids starting over.
Redirect everything that moves. Every old URL should point to the most relevant new page, not a generic homepage.
Preserve valuable content. Rewriting is fine; deleting substantial content from strong pages often isn't.
Keep the structure meaningful. Headings, page titles, and descriptions should still describe what the page is about.
Check performance. A new design that's slower to load can quietly hurt both rankings and conversions.

I create a simple mapping of every important old page to its new equivalent before any design work begins. It's tedious, but it prevents the most common and costly redesign mistake.

Keep the language that converts#

Copy often gets rewritten wholesale in a redesign, usually to match a new brand voice. That's understandable, but some existing phrases may be doing heavy lifting.

Before rewriting, I look for copy that clearly performs: headlines on high-converting pages, phrases that customers repeat back in sales calls, answers to frequently asked questions. These can be refined and restyled, but their substance should survive. The words customers use to describe their problem are often more effective than the words a brand team would choose.

Design for the actual decision#

Visitors to a solar company's website aren't browsing for fun. They're considering a large, unfamiliar purchase with long-term consequences. They want to know: will this work for my home, how much will it cost, how much will I save, and can I trust this company?

That shaped the new information architecture around the decision journey rather than around the company's internal structure:

Understand — how solar works for a home like mine.
Evaluate — costs, savings, financing, incentives.
Trust — reviews, installations, certifications, warranties.
Act — a clear, low-pressure way to get a quote.

Each stage had clear paths forward. Visitors who arrived ready to act could do so immediately; visitors still learning could explore without being pushed.

Trust is the real product#

For any significant purchase, trust signals matter as much as information. The redesign gave real attention to:

Real project photography instead of generic stock images of smiling families.
Specific testimonials with locations and project details, not anonymous quotes.
Transparent process explanations — what happens after you request a quote, how long installation takes, who you'll talk to.
Credentials and warranties presented clearly rather than buried in the footer.

These elements existed on the old site, but scattered and underemphasized. Bringing them forward was one of the biggest improvements, and it didn't require inventing anything new.

Launch carefully, then watch#

Even with careful planning, redesigns carry risk. A few practices reduce it:

Test before launch — check redirects, forms, tracking, and performance on the staging site.
Launch at a calm time — not right before a major campaign or a seasonal peak.
Monitor closely afterward — traffic, rankings, form submissions, and error logs, daily for the first few weeks.
Have a plan for fixing things fast — because something will be missed.

The first month after launch is part of the project, not an afterthought. It's when you find out whether your assumptions were right.

Respect is the method#

The common thread in all of this is respect for what came before. The existing site was built by people who learned things about the business and its customers. Some of those learnings are embedded in pages, phrases, and structures that look unremarkable but work.

A good redesign doesn't erase that history. It builds on it — keeping what earned its place, fixing what didn't, and presenting the whole thing with a clarity and confidence that the old site couldn't quite reach.

New is exciting. Working is valuable. The goal of a redesign is to end up with both.