How to redesign a SaaS website without losing your rankings
Why redesigns lose traffic
A redesign changes three things at once — the URLs, the content, and the code that delivers it. Each can independently cost you rankings, and because they ship together it’s hard to tell afterwards which one did the damage.
The pattern is depressingly consistent. Traffic holds for a week or two while the old pages are still indexed, then falls over the following month as Google recrawls and finds redirects that chain, pages that no longer contain the content that ranked, or a slower experience. By the time anyone notices, the launch is three weeks old and nobody wants to relitigate it.
The process
Step 1
Audit what's actually earning
Pull twelve months of Search Console data and sort by clicks. In most SaaS sites, a handful of pages account for the large majority of organic traffic — and they’re rarely the homepage. They’re usually a comparison page, a specific feature page, and two blog posts written three years ago.
Those pages are assets. Anything that changes about them is a risk that needs a reason.
Separately, pull your backlink profile. A page with fifty referring domains carries authority that transfers through the site’s internal linking. Delete it and you lose that, even if the page itself was ugly.
Step 2
Decide what survives
For each URL earning traffic or links, pick one:
- Keep as-is. Same URL, restyled. The safest option, and the right one for anything with strong rankings.
- Keep and improve. Same URL, better content. Adding depth to a page that already ranks is the highest-return SEO work there is.
- Merge. Two thin pages competing for one query become one strong page, with a redirect from the loser.
- Retire. Only when the page earns nothing and has no links. Redirect it to the closest relevant page, not the homepage.
Step 3
Design the system, not the pages
This is the point where a redesign either becomes maintainable or becomes the next redesign. Design a set of reusable sections — heroes, feature blocks, proof, pricing, FAQ, CTA — rather than a handful of bespoke page layouts.
Two reasons. First, your team can then build new landing pages without a designer, which is what lets you actually execute an SEO content plan. Second, consistency across pages is what makes a site feel considered rather than assembled.
While designing, keep four SEO constraints in view:
- One h1 per page, real h2s for sections — not styled divs.
- Meaningful text above the fold, not baked into a hero image.
- An image budget. Decide the maximum weight for a hero before you design one.
- Internal linking surfaces — a footer and a related-content block are design decisions.
Step 4
Build server-rendered
If your new site renders content client-side, you’re betting your indexation on the crawler executing your JavaScript promptly and consistently. Sometimes it does. Often it does so late, and inconsistently across pages.
Server rendering with static generation where the content allows removes the bet entirely. The HTML that arrives already contains the content, and interactivity layers on top.
Step 5
Map every redirect before launch
Build a spreadsheet with every old URL and its new destination. Every one — not just the ones with traffic, because links point at pages nobody visits.
Rules that matter:
- Use 301s, not 302s.
- No chains. Old → new directly, never old → interim → new.
- Test the whole map on staging before launch, not after.
- Keep the redirects in place permanently. There is no expiry date on an old link.
Step 6
Measure for ninety days
Rankings move for a few weeks after any significant change. Don’t panic in week one and don’t declare victory in week two.
Watch: indexed page count, organic clicks by page, average position for your top queries, Core Web Vitals, and conversion rate by page. Compare like-for-like against the same period last year, not against last month, so seasonality doesn’t fool you.
If a specific page drops and doesn’t recover in six weeks, diff it against the archived version and look for what left — usually content that was doing more work than anyone realised.
What good looks like afterwards
A redesign that worked shows flat-to-up organic traffic within two months, better Core Web Vitals, a higher conversion rate, and — the part people forget to check — a marketing team that can publish a new page without filing a ticket.
Frequently asked questions
If you can avoid it, yes — separate them. When both ship together and traffic drops, you cannot tell which caused it. If they must ship together, be extra rigorous about the redirect map.
Six to ten weeks for a mid-sized site including audit, design system, build, and redirect mapping. Faster is usually skipping the audit.
Yes, and it's lower risk. Ship the section system and a few pages, verify nothing broke, then migrate the rest.
Keep the URLs. Blog posts are often the majority of a SaaS site's organic traffic and changing their paths is the single most common cause of a traffic cliff.