How pagination works
When a list is too long for one screen — a blog archive, a product category, a directory of resources — it is split into a series, and each part is given its own address, usually a numbered path or a query parameter. The first part carries the newest or most important items; the rest hold everything else.
Those later addresses are real pages. A crawler can request them, read them and follow the links inside them to the individual items they list. That is pagination’s actual job in search: it is the route by which deep items are discovered. Anything that breaks the route — links that appear only after a click, an address that never changes — leaves those items reachable by luck rather than by design.
Why pagination matters
On a large catalogue or an established blog, most of your content sits beyond the first part of a listing. If the series is crawlable, those items are found, refreshed when they change and re-crawled when you update them. If it is not, they can quietly drop out of the index while nothing on the site appears broken.
It matters for customers too. Someone comparing options wants to see the full range, keep their place, and send a colleague a link to what they were looking at. A series of stable addresses does all three. A listing that resets to the top when you use the back button does none of them.
Where pagination goes wrong
The most damaging mistake is pointing every part of the series at the first one with a canonical tag. It looks tidy, and it tells search engines that the later pages are duplicates not worth keeping — along with the only links to the items listed on them. Each page in a series should carry a canonical pointing at itself.
Two others are common. Blocking later pages from the index while still expecting them to pass value onward: a page excluded for long enough tends to have its links followed less, and the items below it go quiet. And leaving every page in the series with an identical title and description, which produces a wall of near-identical rows in your reports and gives nobody, human or machine, a way to tell one from another.
Getting it right
Give every page in the series a self-referencing canonical, a title that says which part of the series it is, and ordinary crawlable links to the parts either side. Keep the item links in the HTML rather than loading them only after an interaction, and make sure sorting and filtering controls do not silently spawn a separate series for every combination a visitor can pick.
Decide, too, what belongs in a listing at all. If the real value sits in the individual items, a shorter series supported by strong internal links from relevant articles beats an endless one nobody scrolls through. And if you are considering replacing the series with infinite scroll, keep the paginated addresses underneath it rather than removing them, so the discovery route survives the interface change.