What Is Index Bloat?
Index bloat is a state where Google holds far more of your URLs than your site has useful pages, with the surplus earning no impressions and no clicks.
Index bloat is a state where Google holds far more of your URLs than your site has useful pages, with the surplus earning no impressions and no clicks. It is not a penalty, and a large indexed count is not automatically a case of it. Bloat is measured by the dead share, never by the total.
What index bloat actually costs you
The first cost is attention. Crawl requests spent on filter combinations and empty tag archives are requests not spent on the pages you publish deliberately, and on a large site that gap shows up as slow discovery of anything new.
The second cost is competition inside your own domain. When six near-identical URLs cover one topic, internal links and external links spread across all six, and Google picks one to show. It often picks the one you would not have chosen. The third cost is that your reporting stops working: coverage totals, average position, and index counts all average across pages that were never meant to exist. Checking which pages still pull real visits separates the working part of your site from the residue, and that split is the whole diagnosis.
Where the extra URLs come from
Almost nobody publishes bloat on purpose. It accumulates from defaults, from features nobody switched off, and from URL variants the site generates while doing exactly what it was built to do.
- CMS archives left on. Tag, category, author, and date archives are generated automatically. A publisher can end up with thousands of archive pages holding one post each.
- Attachment and media pages. Some platforms give every uploaded image its own indexable URL with no content on it beyond the image.
- Parameter variants. Sorting, filtering, and tracking parameters produce distinct addresses for content a visitor would call identical.
- Internal search results. If query URLs are crawlable and anything links to one, the whole result space becomes eligible.
- Protocol and host duplicates. HTTP alongside HTTPS, www alongside bare domain, trailing slash alongside no slash, each served with a 200 rather than redirected.
- Expired content nobody retired. Sold properties, filled vacancies, and discontinued products that stay live because no one owns the deletion decision.
Every one of these passes a technical audit, because each URL is reachable, returns 200, and carries no blocking directive. The question is not whether a URL is eligible for indexing, but whether it should have been. Those are different tests, and only one of them is automated.
| What the data shows | What to do about it |
|---|---|
| Indexed, has impressions, no clicks | Leave it indexed and improve the page |
| Indexed, zero impressions in 90 days, no traffic value | Noindex or remove |
| Indexed, duplicate of a better URL | Redirect or canonicalise to the version you want |
| Indexed, needed by users but not by search | Noindex and keep it linked internally |
| Crawled, currently not indexed | Google already made the call; deletion changes nothing |
The blunt version
A large indexed count is not a problem by itself, and anyone opening an audit with the raw number is selling you the pruning project before the diagnosis. The number that matters is indexed-with-zero-impressions.
Work it out yourself. Take the indexed count from the Pages report in Search Console. Then open the performance report, set the range to 90 days, filter to pages, and count the URLs with at least one impression. Subtract. Say a site shows 20,000 indexed and 3,200 URLs with any impression at all: roughly 17,000 URLs exist in Google and have never been shown to a human in three months. That subtraction is the diagnosis, and it takes ten minutes.
Run it in reverse and it stays honest. A catalogue with 400,000 indexed URLs where most earn impressions is working as designed, and pruning it would remove revenue. A 600-page site where 400 pages have never been shown has a real problem, at a scale nobody would headline in a proposal. Size is not the signal. Silence is. The same logic applies when traffic drops without explanation, and someone reaches for the index count as the cause.
Example
Say a publisher with 2,000 articles finds 61,000 URLs indexed. The gap is not mysterious once it is broken down: an attachment page for every uploaded image, a tag archive for every tag an editor ever typed, and a date archive for every month since launch. Of those 61,000 URLs, around 2,400 have earned an impression in the last quarter. The fix is boring and mostly configuration: disable attachment pages, noindex tag and date archives, and keep category archives that genuinely aggregate. Traffic does not jump the following week. What changes is that new articles get picked up in days rather than weeks, and the coverage report becomes readable again.
FAQ
How many indexed pages count as index bloat?
There is no threshold, because the count on its own carries no information. Compare indexed URLs against URLs with at least one impression in 90 days. If most of your index has never been shown to anyone, you have bloat, whether the total is 900 or 900,000.
Should I noindex these pages or delete them?
Noindex the ones people still need, such as filtered views and internal search. Return 410 for content that should no longer exist at all. Do not block them in robots.txt first: a URL that cannot be crawled cannot be read, so the noindex tag on it is never seen and the URL stays put.
Does index bloat trigger a penalty?
No. Nothing gets applied to your site as punishment for holding too many URLs. The damage is mechanical rather than punitive: crawl attention goes to pages that cannot earn anything, ranking signals split across duplicates, and every report you rely on gets harder to read.
Related terms
- TTFB — the server response time that decides how much crawling your site can absorb.
- Faceted Navigation — the single most common source of URLs you never intended to publish.
- Indexability — whether a URL is allowed into the index, which is a separate question from whether it earns anything.
If an audit quotes your indexed page count without quoting how many of those pages earned an impression, it has told you nothing. Ask for the second number before you approve a single deletion.