A low-traffic page hurts the whole site at the point where it is also low-quality, thin, or duplicate, and part of a broader pattern of such pages that drags on your site-wide quality signals. Low traffic alone is not the trigger. A page can get very few visits and still be a perfectly good page, useful to the rare person who needs it, accurate, and properly indexed. That page is not a liability. The harm begins when low traffic coincides with low quality and there are enough of those pages to form a pattern.
The reason the pattern matters more than any single page is that search engines assess sites in aggregate as well as page by page. One thin page among thousands of strong ones is noise. But when a meaningful share of your indexed pages are thin, duplicative, or built for nobody in particular, that mass can shape how the whole site is perceived, lowering the average quality signal and spreading crawl attention across pages that earn none of it. The traffic number is just a symptom that sometimes accompanies the real problem, which is quality. Plenty of harmful pages were never going to get traffic, but they were also never worth indexing.
So the pivot is low-quality-pattern, not low-traffic. The diagnostic question is not “how many visits does this page get,” it is “is this page thin or duplicate, and is it one of many like it.” A useful, accurate, low-traffic page that serves a genuine narrow need is fine and can stay. A thin, purposeless page that happens to also get no traffic is the one to worry about, and it worries you most when it has lots of company. This reflects observed behavior rather than a published rule, so weigh it against your own data, but the line holds.
To put this to use, stop sorting your pages by traffic alone and start judging them by quality and by whether they cluster into a pattern. Identify the thin, duplicate, no-purpose pages, count how widespread they are, and treat that pattern, not the traffic figure, as the thing to fix.