Google Will AI-Upscale Your Merchant Center Images in 2027 (Unless You Beat It)

Google announced on April 14, 2026 that it will require a 500x500 pixel minimum for every Merchant Center image, with enforcement starting January 31, 2027. Merchants who do not re-upload will have their images AI-upscaled by Google, which can hallucinate product details and shift color fidelity on surfaces like Shopping and Performance Max. Nine months to audit feeds before enforcement kicks in.

The spec update landed in Google's changelog alongside five new attributes: handling_cutoff_time, minimum_order_value, video_link, loyalty_program_label, and loyalty_tier_label. Useful additions, all of them. The one that matters for anyone running a Shopping feed today is the image resolution bump, because it's the one that reaches back into years of product uploads. Search Engine Roundtable clocked it first. Some merchants will not discover their exposure until they see a wave of warnings between now and 2027, and those warnings look identical to the noise generated by every other product disapproval in Merchant Center.

What actually changed on April 14

Google published the update with five new attributes and one resolution change. The quick scan:

• handling_cutoff_time sets a daily deadline at the product level for order processing. Same concept as the account-level version, just per SKU.
• minimum_order_value sets a minimum spend required to ship an individual product. Useful for merchants with bundled shipping logic or low-margin items.
• video_link accepts product video URLs. You can submit today, but Google will not begin serving or policy-validating videos until June 30, 2026.
• loyalty_program_label and loyalty_tier_label become sub-attributes on shipping, so you can specify that your "Gold Tier" members get free shipping on SKU 4421 but not SKU 4422.
• image_link and additional_image_link minimum resolution rises to 500x500 pixels, enforced January 31, 2027. Warnings start now.

The full product data specification lists dozens of attributes that interact with the new shipping sub-attributes, and the loyalty additions pair directly with the in-AI-mode loyalty surfaces Google started rolling out last quarter. The first four changes are opt-in. The image rule is the one every existing catalog has to answer to.

The resolution rule is where disapprovals actually start

The part most merchants have not fully priced in: the current image minimums are 100x100 pixels for non-apparel and 250x250 for apparel. On January 31, 2027, both become 500x500. That is a 25x pixel-count jump for non-apparel and a 4x jump for apparel.

If your feed has been running cleanly for years, that is exactly the problem. You don't know how many of your SKUs are sitting at 100 to 300 pixels, because Google has been happily accepting them. The warnings started on April 14, but they are sitting in the feed diagnostics panel alongside every other warning class. That's where they'll stay until something breaks in a Q1 shopping peak.

I'd run the audit this week, not next quarter. The number to look at is the percentage of SKUs with images under 500 pixels on either dimension. In most catalogs I've looked at that came out of older Shopify, BigCommerce, or Magento builds, that number lands somewhere between 8% and 25%. Enough to visibly dent Shopping impression share if it hits on a peak week.

The 500x500 floor also matches the existing minimum Google uses to serve Shopping ads on YouTube and connected TV. That is not a coincidence. Google is quietly merging the feed spec for its smaller surfaces with the spec for its premium ones, which means any product that gets upscaled is effectively banned from the higher-CPC placements whether Google says so out loud or not.

The last comparable spec overhaul was in 2021, when Google discontinued the "oversize" size_type designation and forced region-specific shipping sub-attributes into cross-border feeds. Merchants who didn't parse the changelog found out months later when disapproval rates started climbing quietly. This one has a longer fuse and a larger surface area.

Google's AI upscaler is not the safety net it looks like

Google's announcement softens the blow by saying it will "optimize some images that are smaller than 500 x 500 by utilizing high-resolution near-duplicate images or through AI upscaling of the original image." Read that sentence twice.

"Near-duplicate" means Google will try to find a matching high-resolution image elsewhere on the web that appears to be the same product. If your product is generic, this works. If it's a differentiated SKU with your specific branding on the packaging, Google might pick a competitor's version of the same object, or an older version of yours. Either way the shopper sees an image you didn't pick.

The AI upscaling path is worse in a different way. Upscalers hallucinate. On product photography this usually shows up as blurred text on packaging, shifted color saturation, and softened edges on fabric or jewelry. The product doesn't get disapproved, but it does get re-rendered on a surface you can't preview until it's already serving.

Google's incentive is to keep products on the surface even when the source image is weak. A merchant's incentive is to keep product images identifiable. Those two objectives diverge the moment the upscaler starts guessing.

There is also a compliance angle that nobody has raised yet. Regulated categories like supplements, alcohol, and medical devices have strict packaging claim requirements where text on the label has to be legible and accurate. An upscaler that softens a dosage number or blurs a warning label pushes the merchant into gray territory with both the platform and the regulator. Treat category compliance teams as another stakeholder for this audit, not a separate workstream.

The video_link slot is a first-mover window

The video_link attribute is live for submission now but will not serve until June 30, 2026. That is roughly 75 days of lead time for anyone paying attention.

This is the rare Merchant Center change where being early is actually free. Google typically launches new attributes and then tunes ranking in favor of early adopters to drive coverage. The attribute supports YouTube links, which means any merchant with existing product demo or unboxing video content can be ready by enforcement day with zero new video production.

The merchants I'd expect to move first are apparel and small electronics, where video_link lets a product tile show motion in Shopping and Performance Max placements where static photography has plateaued. Everyone else will treat it like a nice-to-have until a competitor's tile starts outperforming theirs, and then fight for it at exactly the moment Google already knows what's working.

The three queries that eliminate the exposure

The fastest way to stop reading spec update articles and start fixing exposure is to run three queries on tomorrow's feed export.

1. Count images with either width or height under 500 pixels. That's your 2027 backlog.
2. Count SKUs where handling_cutoff_time or minimum_order_value would differ from your account-level default. Those are the product-level shipping opportunities you can add this month.
3. Upload five product videos via video_link with your three top-traffic SKUs. You'll have real data by July on whether serving them moves impression share.

Teams that quietly run this audit on April 17, 2026 end up with cleaner Q1 2027 numbers than the ones who wait for the warning email. Historically Google's warning email arrives about two weeks after the first wave of disapprovals, which is five to seven days of lost Shopping impressions for a catalog that could have been fixed in an afternoon.

For context on how aggressively Google has been tightening feed-side enforcement, the Merchant API migration hits exactly the same feed pipelines and the same operational teams. Consolidating both audits into one engineering cycle probably saves a week per merchant.

Notice Me Senpai Editorial