There is no fixed number, because bloat drags by proportion, not by count. What matters is the share of your indexed site made up of low-value pages, not the raw total. A few thin pages sitting on an otherwise strong site are noise the search engine can absorb without re-rating the whole domain. A site that is mostly thin, duplicate, or templated pages has a quality problem that shows up in how the entire site is judged. Treat this as working judgment rather than a measured threshold, because the line lives in the ratio.
The reason a count misleads is that the same number means opposite things on different sites. Fifty thin pages on a domain of fifty thousand strong ones is a rounding error. Fifty thin pages on a domain of eighty pages total is the site’s defining characteristic. The search engine forms an impression of overall quality from the body of work it has indexed, and that impression is driven by what the typical page looks like, which is a proportion question, not an absolute one.
So the useful diagnostic is the make-up of your index, not its size. If the bulk of what you have indexed adds genuine value and a small minority is weak, you are in noise territory and the weak pages are a tidy-up, not an emergency. If a large fraction of your indexed pages are thin, duplicative, or auto-generated padding, that share is what shifts the site’s quality signal downward and holds your good pages back. The same five hundred weak pages can be harmless or fatal depending entirely on what they sit beside.
So when you ask whether bloat is dragging you, stop counting and start measuring share. Look at what proportion of your indexed pages would survive an honest “does this earn its place” test. Judge the bloat by how much of the index it occupies, and act when that fraction is large enough to define the site, not when a counter crosses some imagined number.