Most personal injury websites are mostly noise, and I mean that as a measurement, not an insult. In my working experience across this vertical, a typical PI site more than two years old runs a Worth-to-Index ratio in the single digits: of every hundred URLs Google is spending resources on, only a handful serve a query that could ever produce a case. This piece explains the ratio, the crawl economics behind it from Google's own patents, how noise accumulates on law firm sites specifically, and the prune-merge-keep discipline that fixes it.
Worth-to-Index is the share of a site's indexed URLs that do real commercial or supporting work: pages that either serve a query an injured person actually types or structurally support a page that does. Everything else is carried weight: tag archives, paginated fragments, parameter duplicates, image attachment pages, author archives for authors who wrote twice in 2019, press releases nobody will ever search for, and the doorway fleets covered earlier in this series.
The number matters because the systems reading your site do not have infinite patience, and more importantly, because what they crawl and store shapes what they believe your site is. A site that is five percent signal teaches the machine, a hundred times a day, every day, that it is mostly noise, and the machine is a diligent student.
Crawling costs Google money, so Google engineered frugality, and the engineering is public.
The crawl-scheduling patents, US 8868541 and the crawler scheduler family behind US 10621241, describe the budgeting directly: finite crawl resources allocated across an expanding web, scheduled per document by importance and by how often content actually changes. Pages earn visits; they are not owed them.
One member of that family even publishes a formula worth reading twice. A document's crawl priority can be computed as its PageRank squared, multiplied by its URL change frequency. Importance, squared, times freshness of change. Read what that means for the noise on a law firm site: a thin archive page with no authority and no changes earns a priority near zero, and it STILL gets crawled sometimes, which means it is still spending budget the money pages could have used. Multiply by two thousand junk URLs and the practice pages that sign cases are sharing their oxygen with dead weight.
The quieter cost compounds it: the site-quality machinery forms its judgment from what it sees. When ninety-plus percent of what it sees is scaffolding, the prior the whole site starts from is a scaffolding prior. The noise does not just waste crawl. It testifies.
Nobody plans a five percent ratio. It accrues through perfectly normal decisions.
The CMS ships it: every image an attachment URL, every tag an archive, every author a page, pagination forever. Nobody turns the defaults off because nobody knew they were on.
Vendors add it: a blog program measured in posts per month produces hundreds of thin pages aimed at nothing; a city-page initiative adds forty templates; a redesign leaves the old URLs alive in parallel because deleting felt risky.
And nothing removes it, because removal is nobody's job and feels like destroying work that was paid for. Sites only ever grow. The ratio only ever falls. Ten years of normal operation produces exactly what I keep finding: a thin layer of real pages buried in an index of accumulated exhaust, with nobody at the firm able to say when the burying happened, because it never happened on any particular day.
The Worth-to-Index audit is unglamorous and decisive. Pull every indexed URL. Classify each one into exactly four buckets: commercial (serves a case-producing query), supporting (earns its place by helping a commercial page: real guides, real FAQs, genuine local pages), structural (necessary plumbing: contact, about, policies), and noise (everything else, honestly labeled).
Compute the ratio. Then assign every noise URL one of three verdicts, and this trio is the whole method: prune, merge, or keep. Prune what serves nothing: deindex or delete, with redirects where anything external ever pointed at it. Merge what has fragments of value: fold three thin posts into the one real resource they should have been, redirect the husks. Keep only what earns its place, and write down why.
Two cautions from doing this on live firms. Never mass-delete on the first pass; check what quietly earns links or entrances first, because every site has two surprises. And sequence redirects like surgery, mapped before cut, because the crawl machinery treats URL changes as events and sloppy ones leak history.