How does Gmail's image proxy work?

Before 2013, when a Gmail user opened an email and images loaded, those image requests came straight from their device. Senders could see approximate location, email client, even device type from the request logs. Then Google changed everything with the Gmail image proxy.

But Here's how it works today: when Gmail receives an email with image URLs, Google's servers fetch those images themselves and cache them. When the recipient opens the email, they receive the cached versions served from Google's infrastructure, not from the original image host. The recipient's device never directly contacts your tracking server.

The practical effects: IP-based location data for Gmail opens is no longer meaningful (you'll see Google datacenter IPs instead). Device and client detection through image requests becomes unreliable. Open timing gets murkier because Google may fetch images before the recipient actually reads the email. If you're doing any geo-targeting or personalization based on opens, Gmail recipients are effectively a black box beyond "opened" or "didn't open."

For open rate tracking, the proxy does mean that an open logged from Gmail is generally a real open (or at least Google thought it likely enough to cache the images). This is different from Apple's Mail Privacy Protection, which pre-fetches for everyone regardless. Gmail's proxy is more conservative and doesn't inflate opens the same way MPP does. Image caching through the proxy also means that personalized images, countdown timers, and "live" content that changes based on when someone opens, need special handling. The image gets fetched once and cached. Later opens may show stale content if you're not using a CDN or dynamic image service designed to work around proxy caching.