Modernising the archive: what the search data says

In late March I migrated three mobile technology sites — All About Symbian, All About Windows Phone, and All About Mobile, 24,000 articles across roughly 145,000 pages — from a dying legacy server and LAMP code base to a modern static archive. I've written about how the migration was done, the optimisation passes that followed, and adding structured data at scale. This post is about what happened next: whether any of that work shows up in Google's treatment of the sites.

What changed, in SERP-signal terms

Preservation was the goal, but the side effect was a full modernisation of everything Google's documentation says it cares about:

• Static delivery. Dynamic CMS pages on ageing PHP became pre-rendered HTML behind a CDN. Time-to-first-byte and reliability improved overnight, and there is no longer an application layer to fall over mid-crawl.
• Core Web Vitals and Lighthouse. The first optimisation pass took a representative page from Performance 71 / Accessibility 66 / Best Practices 73 / SEO 83 to 93 / 91 / 96 / 100. Subsequent passes pushed further: today the All About Symbian homepage audits at 96 / 100 / 100 / 100 on mobile emulation.
• Structured data at scale. From basic Article microdata to comprehensive JSON-LD across six content types — 4,504 review pages with proper Review schema, 81,117 forum threads with DiscussionForumPosting and comment markup, breadcrumbs on 23,000 pages.
• Crawl hygiene. Sitemaps resubmitted, thousands of 404s and deprecated-schema warnings worked through as Search Console flagged them.

None of this changed a word of content. The archive is frozen — no new posts, no updates, no promotion. Which makes it an unusually clean setting for the question: does technical quality alone move search performance?

What Google did

Search Console was set up the day after cutover, so the data starts with the modernised archive's first day — sixteen weeks of it now.

Three things stand out in the All About Symbian data:

1. Clicks ratcheted up and held. From 1,114 in the first full week to a sustained ~1,430 average over the last six weeks — up 29%, with no new content to explain it.
2. Click-through rate rose from 1.06% to 1.26%. When the pages appear in results, more people choose them. That's consistent with what richer snippets and faster pages are supposed to do — the listing itself got better, not just more frequent.
3. The impressions curve tells a re-evaluation story. Google spiked impressions to a peak in early May — around 177,000 in a week, plausibly re-testing the re-crawled pages across many queries — then settled roughly a third lower. Clicks held through that settling. Google showed the archive less, but better.

Same treatment, three outcomes

The section above tells a simple, positive story, but here's the slightly messier reality. All three sites got identical treatment. They did not get identical results.

All About Mobile — the smallest site — grew the most, roughly 1.7× on both clicks and impressions, but I would take that with pinch of salt, given the low volume. All About Symbian gained steadily. All About Windows Phone drifted down 12%, though is now showing upward signs. Same schema scripts, same performance work, same infrastructure. The difference is presumably demand and competitive context per topic: technical quality changes how Google evaluates your pages, not how many people care about Windows Phone in 2026.

Why this matters

For a frozen archive, technical quality is the only lever there is — no new content, no outreach, nothing else to pull. What the data shows is that Google demonstrably re-evaluates: a site that will never publish again can still improve in search, because the ranking systems keep re-reading the signals you send. Twenty years of accumulated authority is an asset, but it pays out through a technical surface that either helps or hinders.

For living sites the implication is sharper. These signals — speed, structured data, crawl hygiene — are the floor under everything else you do. And the cost of getting them right has collapsed: this entire modernisation, 145,000 pages, was days of agentic work rather than a quarter of engineering. The excuse for a slow, schema-less, error-ridden site is gone.

The other quiet enabler is the measurement itself. The charts above come from programmatic Search Console access feeding a weekly analytics digest — the same instrumentation that flagged the trends in the first place. Modernise the site, but modernise the measurement with it.