What crawl demand measures
Crawling has two halves. One is how much your server can take, which shows up as crawl rate. The other is how much the search engine actually wants your pages, and that is crawl demand. A site can have plenty of spare capacity and still be crawled rarely, simply because nothing about it suggests there is anything new worth collecting.
Demand is built from a few judgements the engine makes on its own. How popular a URL appears to be, judged from links pointing at it and from interest in it. How often the content has genuinely changed in the past, because a page that never changes does not need re-reading. How important the URL looks within your own structure, which is why pages linked from the homepage and main navigation get visited more than pages buried at the end of a pagination chain. And whether something site-wide has happened, such as a migration or a change of address, which temporarily raises demand across everything.
Why crawl demand matters
Because for most sites this, and not server capacity, is the real constraint. Owners of large catalogues often assume Google is being held back by hosting. Far more often Google has simply decided that most of those URLs are not worth returning to, and the fix is editorial rather than technical.
It also explains the pattern almost every business notices eventually: important pages get re-crawled quickly while older, deeper pages sit untouched for long stretches. That is demand doing its job. If a page you care about is in the slow group, something about how it is linked, how it changes and who visits it is telling the engine it does not matter.
Where crawl demand goes wrong
Mass-producing URLs is the usual way to destroy it. Filter and sort combinations, session parameters, internal search results, near-identical location or service pages spun from a template. Each one is a URL the engine may have to consider, and a site full of pages that say almost nothing teaches the engine that the site is not worth checking often.
Faking freshness is the second. Bulk-updating dates, republishing old posts with no real change, or re-saving every page to bump the last-modified stamp all get discounted quickly, because the engine compares what it fetched with what it had. What survives that comparison is content that actually changed.
The third is orphaning. A page nothing links to and nobody visits has almost nothing to generate demand from, however good it is.
What to do about it
Start by cutting the URLs that add nothing. Consolidate near-duplicates, stop generating parameter variants that nobody would ever want indexed, and keep thin archives out of the index. Reducing index bloat concentrates attention on the pages that earn it.
Then feed the pages you care about. Link them from pages that already get crawled often, mention them in new content, and give them a route from the navigation rather than only from a sitemap. Update pages when there is something real to update, and leave the rest alone.
Finally, judge results by what gets re-crawled, not by how much. In the crawl stats report, watch whether requests are going to money pages or to noise. Shifting that mix is the point of the exercise, and it is usually the most valuable part of a technical SEO project on a large site.
