Should you really redirect all images during a site migration?

Why does Google emphasize image redirection so much?

Martin Splitt's statement is unequivocal: images are not a detail that can be overlooked during a migration. Visual content generates traffic, sometimes massive, depending on the sectors (e-commerce, tourism, publishing), and Google Images represents a complete acquisition source.

The problem? The image index operates on refresh cycles that aren't synchronized with the textual index. A non-redirected image URL = a 404 error for Googlebot-Image. The result: months spent waiting for Google to recrawl, reindex, and restore visibility. During this time, image organic traffic collapses.

What does “monitoring the server logs of the old domain” mean?

Google advises not to rush the closure of the old domain. The reason: Googlebot continues to crawl the old URLs—including images—even after migration while it updates its indexes.

Monitoring the server logs means identifying residual crawl activity. As long as Googlebot is still regularly hitting the old domain, it means it hasn't finished its update work. Cutting off the server too early = forcing Google to deal with 404 errors instead of clean redirects. In short, you're complicating its life and slowing down the process.

Do image redirects matter as much as HTML page redirects?

Yes. That's precisely the crux of this statement. Visual content is very important for SEO, says Google. Not “may be useful” or “to consider in certain cases.” Very important.

In practice, this means that a migration that perfectly manages HTML redirects but forgets about images is a shaky migration. Images carry semantic context, generate backlinks (Pinterest, social networks), and feed visual featured snippets. Overlooking them is to sacrifice part of the SEO without even realizing it.

• All images must have a 301 redirect, not just those from main pages
• The refresh rate of the image index is slower than the textual index—patience required
• Monitoring the server logs of the old domain reveals when Googlebot has finished its recrawl job
• Do not delete the old domain as long as crawl activity remains significant
• Image organic traffic can represent a significant share of overall traffic depending on the sector

Is this recommendation consistent with field observations?

Yes, and it’s even a welcome confirmation. SEO practitioners who have managed complex migrations know that images are often the overlooked aspect of the process. Focus tends to be on page URLs, internal linking, and XML sitemaps—but images? They get forgotten, or one assumes they will follow.

Except they don’t. Experiences show that a migration without image redirects can lead to traffic drops of 20 to 40% on Google Images, and recovery takes between 3 to 6 months. Martin Splitt confirms what we observe: the image index is slow to update, and 404 errors on visual resources are not trivial.

What nuances should be added to this statement?

The wording “all images” is broad. Specifically, what does it include? Editorial content images, yes. E-commerce product visuals, obviously. But also CSS icons, sprites, technical assets? [To be verified] — Google does not specify whether the recommendation targets only indexable images or all visual resources on the site.

Another point: the waiting period before removing the old domain. Google states “wait for crawl activity to decrease,” but decrease to what extent? A crawl per day—is that negligible or still significant? The statement lacks concrete thresholds. In practice, it is recommended to keep the old domain active for at least 6 months post-migration, sometimes 12 months depending on the site's size. But Google does not provide a figure.

In what cases can this rule be relaxed?

If the migrated site generates little or no traffic through Google Images, priorities may shift. A B2B site focused on technical textual content, with no strong visual dimension, may neglect certain images—even if it’s never ideal.

But be careful: even in these cases, images carry semantic context for textual content. Google uses visual recognition to enhance its understanding of a page. Neglecting image redirects risks weakening the overall signal of the page, even if the image traffic itself is negligible. So yes, it can be relaxed, but one must measure the risk.

What should be done concretely during a migration?

The first step: inventory all image URLs from the current site. Not just those visible in HTML pages but also those present in image sitemaps, in srcset tags, and even former orphaned images that still generate traffic via Google Images. A Screaming Frog or Sitebulb crawl configured to extract images is a good starting point.

Next, map each old image URL to its new URL on the target domain. If the image no longer exists, redirect to an equivalent image or, if there’s none, to the page that best contextualizes the subject. A 301 redirect to the homepage? No. Google wants semantic consistency, not a patchwork.

How to monitor that the image migration process is going smoothly?

Two tools: server logs and the Search Console. For logs, configure an analysis on the old domain to track the activity of Googlebot-Image. As long as the volume of requests remains high (let’s say several hundred per day for an average site), it means Google continues to recrawl the old domain. Don’t cut anything.

On the Search Console side, monitor the coverage index and the 404 error reports for image URLs. If spikes in errors appear after the migration, it’s an alert signal: either the redirects aren’t in place, or Google hasn’t updated its index yet. In all cases, this requires an immediate corrective action.

What errors should be absolutely avoided?

Error number one: assuming that HTML redirects are sufficient. They are not. Images are crawled and indexed independently from the pages that host them. A 301 redirect on the page does not automatically redirect the images it contains.

Error number two: cutting off the old domain too quickly. We’ve said it, but it deserves repeating: as long as Googlebot is regularly hitting the old server, the redirects must remain active. Closing a domain after two weeks just because “visually everything looks good”? That’s a guarantee of losing image traffic for months.

• Inventory all image URLs via a comprehensive crawl (Screaming Frog, Sitebulb)
• Map each old image to its new URL or a semantically close alternative
• Set up 301 redirects server-side (Apache .htaccess, Nginx conf, or CDN if applicable)
• Configure a server log analysis to track Googlebot-Image activity on the old domain
• Monitor 404 image errors in the Search Console post-migration
• Keep the old domain active for at least 6 months, ideally 12 months for large sites