Index bloat is a mismatch between how many URLs a search engine has stored and how many pages the site actually needs to rank. A shop with four hundred products can end up with sixty thousand indexed URLs, and none of the extras were written on purpose.
The sources repeat across sites. Filter and sort parameters create a URL per combination. Internal search results pages get crawled and stored. Session identifiers and tracking parameters split one page into many. Tag archives, deep pagination, and staging subdomains contribute their share.
Two costs follow. Crawl budget goes to URLs that will never rank, which slows discovery of the pages that matter on large sites. And a query that could have matched one strong page now has forty weak candidates competing for it.
Cleaning up starts with counting. Compare indexed URLs in Search Console against the sitemap, then group the excess by pattern. Each pattern gets a decision: canonical to the version that should rank, noindex the ones users need but search engines do not, or block in robots.txt where the URLs should never be fetched at all.