According to Search Engine Journal, Googlers Gary Illyes and Cherry Prommawin have described two problems that can follow a move to a CDN (content delivery network, a service that stores copies of pages closer to visitors). Both can hurt how a site appears in Google Search.
The first is a crawl surge. Google's crawlers treat faster server responses as a sign the site can take more requests, so they slowly raise their crawl speed. A new CDN usually makes responses faster. But its cache starts empty. The CDN can only store a page after someone requests it, and that first request has to go to the origin server, which is the server where the site actually lives. Illyes said the first request fills the cache and later requests benefit from it. Until then, the origin server may still take the full load, even though the CDN is in place. Illyes compared it to taking the valve off a hose so crawlers drink from a fire hose.
The second problem comes from panic. If an admin sees traffic spike and assumes bots are scraping the site, they may switch on a CDN setting that blocks bots or challenges them with a CAPTCHA. Illyes noted that Google's crawlers cannot pass such a test. If Googlebot stays blocked for long, Google will start dropping pages from its index, according to the article.
Google's suggested fixes are practical. Prommawin said admins should set the CDN to allow known and trusted crawlers. The speakers also recommend rate limiting Googlebot at the CDN level if origin requests cause downtime that affects real visitors. Illyes advised pre-warming the cache, which means fetching your top URLs through the CDN by script before search engines arrive, so the cache fills in a controlled way.
Why it matters
A CDN is often seen as a safe speed upgrade. Google's comments show that the switch itself can change how crawlers behave, and the first days may put more strain on your origin server, not less. If that server slows down or fails, real visitors suffer too.
The bigger risk is a rushed reaction. A strict bot rule meant to stop scrapers can also shut out Googlebot, and prolonged blocking can remove pages from the index. Planning the crawler rules and cache setup before a migration is cheaper than diagnosing lost traffic afterwards.
Source: Search Engine Journal. This brief summarises their reporting; read the original for the full detail.