The redesign that erased 62% of organic traffic
Ridgeway Cycles sells bikes and accessories from a single shop and a 12-year-old website with 480 indexed pages. Organic search used to deliver 3,100 sessions a month. Six months ago the owner moved to a new platform because the old one "looked dated". The developer kept the same text but changed every URL to a cleaner pattern and redirected only the 40 most-visited pages. The blog, which held 90 advice articles, was dropped entirely because "nobody read it". Six weeks after launch, organic sessions are down 62%. Branded searches still reach the homepage, but category and product pages no longer appear in results. Paid search spend is unchanged and total revenue is down 19%.
1. Separate the evidence in the setup into symptoms of measurement failure and symptoms of genuine loss of search visibility. What would you verify first, and why does the order matter?
Read the worked answer
Symptoms of visibility loss (most likely):
- Category and product pages missing from results - these are exactly the pages whose URLs changed and which were not redirected. An unredirected URL returns a 404, so the ranking signal and the traffic go nowhere.
- The 62% drop is concentrated in non-branded organic, while branded traffic still works - consistent with lost landing pages, not with a site-wide penalty.
- 90 blog articles deleted - those URLs are also gone unless redirected.
Symptoms that could be measurement, not marketing:
- Revenue down 19% is smaller than the 62% session drop. That gap suggests either that organic sessions were low-value, or that some conversions are no longer being recorded (new platform = new tracking tags, new checkout, possibly a new consent banner blocking analytics).
- "Branded searches still land on the homepage" may simply mean the homepage kept its URL and is the only page with a working redirect.
Order of checks:
- Crawl the old URL list against the new site and count live 404s. This is fast, cheap and definitive.
- Confirm analytics and conversion tracking fire on the new platform (compare a week of orders in the shop backend with recorded conversions).
- Only then look at search console coverage, canonicals and sitemaps.
Order matters because the 404 issue is both the most probable cause and the most reversible; diagnosing tracking first would waste weeks while the redirect damage compounds.
2. The owner says "just redirect everything again". Describe a defensible recovery plan and the order you would execute it in.
Read the worked answer
1. Build the URL map before touching anything. Export the old URL list (sitemap archive, crawl backup, analytics landing-page report for the last 12 months). For each old URL, choose the closest new equivalent: product → same product, category → same category, blog article → closest surviving article or category hub. Do not blanket-redirect everything to the homepage; that is treated as a soft 404 and helps nobody.
2. Redirect at page level with 301s, one hop only, and verify no chains or loops.
3. Recover the deleted content. Restore the 90 articles, or at minimum the ones with backlinks or historical traffic, and redirect their old URLs.
4. Resubmit the sitemap, request re-indexing for priority categories, and monitor search console coverage daily for a fortnight.
5. Communicate the timeline. Expect weeks to months, not days, for rankings to return; set that expectation with the owner in writing so the plan is not abandoned halfway.
6. Fix the process failure. Add a redirect checklist and a pre-launch crawl to any future migration, otherwise the same thing recurs.
3. The developer argues the cleaner URL structure is better for users and wants to keep it. How do you decide between keeping it and restoring the old URLs?
Read the worked answer
Judge it on evidence, not aesthetics.
Questions that decide it:
- Do the old URLs have external links and years of accumulated ranking? A crawl of referring domains per URL answers this. If 300 backlinks point at old product URLs, those URLs have equity the new ones do not.
- Are the old URLs actually user-hostile (parameters, session IDs, cryptic IDs) or merely plain? "/product/bike-1234" is not a user problem worth a 62% traffic loss.
- How recent is the migration? Within weeks, restoring is cheap. Six months in, the new URLs may already have picked up links of their own and a second migration adds risk.
Pragmatic answer in most cases: keep the new structure for new content, but redirect old URLs to the matching new ones and keep those redirects permanently. Users get the clean URL; search engines get the mapping. Removing redirects later "because they look untidy" is exactly how this failure repeats.
Decision rule: the URL pattern is a cosmetic preference; the redirect map is the asset. Never trade the second for the first.