Redesign Your Site Without Losing Traffic

A redesign is usually in trouble before the launch date reaches the calendar. The assumption is that because the content is moving over, the traffic moves with it. Rankings do not attach to content in the abstract...they attach to specific URLs.

The value on those URLs is carried by what the templates hold, like headings, body copy, internal links and canonical tags. Change the container and you change what search engines see.

The pass I run below starts before a launch date is confirmed...half of it has to happen while the old site is still live. Three things break in project after project, and they break because nobody was assigned to own them.

Owners for the Three Gaps

Designers own the design and developers own the build. Nobody owns URL continuity, content parity or measurement...those three sit in the gap between agency and client until launch week.

• URL Continuity: Every old address still resolves somewhere sensible.
• Content Parity: The new templates carry the same substance.
• Measurement: Events keep firing the morning after launch.

Put one name against each of them, and make it someone who can hold the date...somebody has to be able to say the work is not done.

Redirect Maps From a Crawl

The XML sitemap tells you what the CMS believes it publishes today...it says nothing about what still earns links or where people are still landing. Plenty of the URLs that matter are missing from it, including old campaign pages, tag and category archives, paginated series, PDFs, and pages someone pulled from the navigation but never removed.

So the list comes from several sources, deduped into one file:

• Full Crawl: Production followed link by link from the top.
• Search Console: Pages data over the longest window available.
• Analytics: Landing pages across a full year, catching seasonal traffic.
• Backlink Data: Target addresses other sites already point at.
• Access Logs: Server records surfacing URLs nothing else knows about.

Then one row per old URL, carrying the current address, the status code, the destination and a short reason. The reason column looks like overhead...it is what stops someone from mapping orphaned pages to the homepage at midnight.

The rules themselves are short.

• Closest Equivalent: Map each address to the page that genuinely replaces it, and never blanket redirect the whole lot to the homepage.
• One Hop: No chains, nothing pointing at another redirect.
• Variants Handled: Protocol, host and trailing slash resolve in one redirect instead of three.
• Deliberate Retirement: A thin, retired page can return a clean 404.

If you are weighing a redirect against a canonical, I have written about when each applies. Test the map against staging with a script that requests every old URL and records both the status code and the final destination...then run the same script against production at launch.

Template Parity Before the Launch

New designs are drawn with sample copy...sample copy is usually shorter than the real thing. The intro paragraph gets trimmed to fit the hero. The H1 becomes an image. Body copy that carried internal links turns into a grid of cards with two word labels.

That rarely surfaces in a design review...it surfaces in rankings weeks later. One page per template gets compared, old against new, on what search engines actually consume:

• Head Elements: Title tag, meta description, H1 and heading structure.
• Body Copy: Whether the substance survived, not just whether it looks full.
• Internal Links: How many in the main content, and where they point.
• Index Controls: Canonical tag, robots meta and the robots header.
• Extras: Structured data, image alt text and page breadcrumbs.
• Raw HTML: What survives in the source with JavaScript disabled.

One more catches good teams. Staging was correctly blocked from indexing...the block ships to production with everything else. Check the response header rather than the meta tag alone...a noindex in a header is invisible in page source.

Tracking Rebuilt on Purpose

Analytics on the old site was usually configured by someone no longer involved, and it was never documented. The base tag survives a redesign because somebody remembers to paste it in...everything downstream of it does not.

GA4 uses an event based model...those events were wired to form IDs, button selectors and thank you page URLs the new build has already renamed. Before launch, inventory what is firing and what depends on it, including key events, conversions imported into ad platforms, call tracking number swaps, chat widgets, CRM form handlers, and the hidden fields carrying source data.

Verify each one on staging by completing the action end to end and confirming the record arrives where a human will see it. A form that posts cleanly and delivers nothing is worse than one that visibly errors.

Two details are worth settling early. Verify Search Console through DNS rather than a file or a meta tag the new build can drop. And if a consent banner ships with the redesign, it changes what gets measured...a measurement change rather than a traffic change, and worth saying so before anyone panics.

The First Hour After Launch

Write the sequence down beforehand and work it in order. One person can usually get through it in under an hour.

1. Request the homepage on every protocol and host variant, confirming one hop to canonical.
2. Run the old URL list against production, expecting correct destinations, no chains, no 404s.
3. Confirm robots.txt is the production version, and check robots meta and headers per template.
4. Load a key page with JavaScript disabled and confirm content and links sit in the HTML.

Leads, Reports and the Record

The rest of the list belongs to a person at a keyboard...someone has to use the site the way a customer would.

1. Submit the primary conversion form and confirm the lead lands in the inbox or CRM.
2. Watch realtime reporting for traffic and for events with the names you expect.
3. Read the contact page closely, including phone, address, hours and tracking numbers.
4. Submit the new sitemap and inspect one page from every template.
5. Record the exact launch timestamp and annotate it in your reporting.

What to Do This Week

If your redesign has a date but no named owner for URLs, templates and tracking, fix that this week. Run the crawl and export your Search Console and analytics history while the old site still exists...you cannot crawl a site that has been replaced.

After launch, review crawl errors and indexing daily for two weeks, then weekly. Expect ranking movement, and resist reading every wobble as damage...what you are hunting for is the self inflicted kind, like a mapped URL returning a 404, a noindex that shipped, or a template that lost half its copy.

Those are cheap to fix early...expensive to find in a quarterly report. If you want a second set of eyes on a redesign in flight, get in touch before launch.