A remodeling company called us last spring, about six weeks after their website redesign went live. The old site was ugly. Genuinely ugly, built in 2016, and the owner had been apologizing for it to customers for years. The new one looked great.
It was also producing 11 form fills a month, down from 28. Same ad spend, same market, same crews.
The design wasn’t the problem. Everything nobody thought to check was the problem.
The new site was better in almost every measurable way
I want to be fair to the design team here, because the build itself was solid work.
Load time went from 4.1 seconds to 1.3. The old site failed Core Web Vitals on mobile and the new one passed comfortably. Photography was professional instead of phone snapshots. Navigation made sense. On a laptop, on a phone, on any device you handed it to, the new site was the better website.
So we weren’t looking at a bad build. We were looking at a build that nobody had connected to the business running behind it.
That distinction matters, because most redesign horror stories get blamed on the designer’s taste. This one had nothing to do with taste.
What actually broke
It took about two hours in Google Search Console, GA4, and the Google Ads account to find it. Three separate failures, none of them visible if you were only looking at the site itself.
Nobody mapped the old URLs
The old site had pages like /services/kitchen-remodeling-charlotte. The new site had /kitchen-remodel. Forty-three URLs changed, and not one of them had a 301 redirect pointing the old address to the new one.
Every link Google had indexed for four years now landed on a 404 page.
Organic sessions dropped 44% over three weeks. In Search Console, the Pages report showed the whole story under “Not found (404)”, a clean vertical wall of newly broken URLs starting the day of launch. Nobody had opened that report, because why would you? The site looked fine.
Redirects are boring, unglamorous, and worth more than the design in the first six months after launch. A 301 tells Google the page moved and passes along most of the ranking authority the old URL had built. Skip it and you’re launching a brand-new website with no history, which is exactly what Google treated it as.
The forms were rebuilt, and the tracking wasn’t
The new site used a different form plugin. Reasonable choice. But the Google Ads conversion action was still firing on a thank-you page URL that no longer existed, and the GA4 event had been set up against the old form’s ID.
Google Ads had been reporting zero conversions for five straight weeks.
Here’s why that turned a reporting problem into a revenue problem. The campaigns were running Maximize Conversions bidding. Feed a smart bidding algorithm five weeks of zeroes and it does what it’s designed to do, which is chase whatever looks cheapest while it hunts for signal. Cost per click fell. Click volume rose. Actual booked jobs fell off a cliff, and the dashboard looked busier than ever.
Ten pages of content quietly went missing
The old site had individual pages for ten service and city combinations. Bathroom remodeling in one suburb, basement finishing in another. Thin pages, honestly, written years earlier and never touched since.
They were also responsible for roughly 60% of the company’s non-brand organic traffic.
During the redesign they got folded into a single “Services” page for a cleaner navigation menu. Cleaner it was. It also removed the only content on the site that had ever ranked for anything other than the company name.
This is the failure I see most often, and it’s rarely anyone’s fault in an obvious way. The designer wasn’t given traffic data. The owner didn’t know which pages were doing work. Everyone agreed the nav felt cluttered.
So how do you redesign a website without losing what’s working?
The website redesign process that protects your leads has one thing in common everywhere I’ve seen it done well. It starts with an inventory, not a mood board.
Export a baseline before anyone touches anything. Pull the Search Console Pages report for the last 12 months, sorted by clicks. Pull GA4 landing pages by sessions and by conversions. Then crawl the current site with Screaming Frog (the free version handles 500 URLs, which covers almost every service business) and export the full URL list. You now have three files that tell you which pages are actually earning something.
Sort every URL into keep, merge, or kill. Anything with traffic or conversions keeps its exact URL, or it gets a 301 to the closest equivalent page. Merging is fine. Merging without a redirect is not.
Write the redirect map before the build starts. Two columns in a spreadsheet, old URL and new URL, every row filled in. This should be a named deliverable in the contract. If a designer won’t produce one, that tells you what kind of project you’re about to have.
Rebuild the tracking as part of launch, not after it. Submit a real form the morning it goes live. Check that it lands in GA4, in Google Ads, and in whatever CRM or inbox it’s supposed to reach. Call the tracking number from your cell phone.
Watch Search Console daily for two weeks. Coverage errors and impression drops show up fast. Six weeks is too late, which is the lesson the remodeler paid for.
The recovery, in case you’re wondering, took about nine weeks. We rebuilt the redirect map, restored eight of the ten service pages as proper standalone content, and fixed conversion tracking in an afternoon. Form fills came back to 31 a month, slightly above where they started, mostly because the new site converted better once the traffic returned.
If you’re running paid ads during the rebuild, stage the launch
This part gets skipped constantly, and it’s the most expensive thing on the list.
Your ads point at final URLs. Those URLs break in a redesign exactly the way organic URLs do, except the damage arrives within hours instead of weeks. Google will disapprove ads whose destination returns a 404, and Meta will keep happily spending budget sending people to a broken page without telling you anything is wrong.
Dedicated landing pages are the bigger risk. They usually live outside the main navigation, which means they’re invisible during a design review and easy to leave off a sitemap. I’ve watched a landing page that was converting at 14% get deleted in a redesign because nobody on the project knew it existed.
Two things fix almost all of this. Keep ad landing pages explicitly out of scope on the first pass, so the rebuild can’t touch them. And pause campaigns for the first few hours after launch, verify every destination URL loads and every conversion fires, then turn them back on.
Also, don’t launch on a Friday. Whoever needs to fix something will be unreachable until Monday, and Monday is three days of wasted spend away.
How much does a website redesign cost, and how often do you actually need one?
Both questions come up in every discovery call, so here’s the honest version.
For a small service business, a template-based redesign on WordPress typically runs somewhere between $3,000 and $8,000. Custom design and development on a larger site more often lands in the $10,000 to $30,000 range. Where you fall depends less on page count than on how much custom functionality you need and whether anyone is writing new content. Content is usually the line item people forget to budget for, and it’s usually the line item that determines whether the thing performs.
As for how often, the “every two to three years” rule you’ll read everywhere was invented by people who sell websites. A site that loads fast, works on phones, and converts well doesn’t have an expiration date.
Redesign when there’s a business reason. Real ones look like this. You can’t update it yourself without calling someone. Mobile experience is genuinely bad, not just dated-looking. You’ve added or dropped services and the site no longer describes what you sell. Speed scores are dragging down your ad Quality Score and your rankings at the same time.
“I’m tired of looking at it” is a real feeling, but it’s an expensive reason to rebuild on its own.
Before you sign anything, pull one report
Open Search Console, go to the Pages report, set the date range to the last 12 months, and sort by clicks. Print it if you want to be dramatic about it.
That list is what you’re actually protecting. Hand a copy to whoever is doing your redesign on day one and ask them a direct question about how each of those URLs will be handled at launch.
The answer you get will tell you a lot before a single mockup exists.
