Merchant Center Image Requirements 2026: 500×500 by January 31, 2027

From January 31, 2027, Google stops approving product images smaller than 500 × 500 pixels. Products with undersized images get disapproved and drop out of Shopping ads and free listings — globally, with no category exemptions. The requirement was announced on August 21, 2026, and Merchant Center has been flagging affected items in the Needs attention section since April 2026.

It sounds like an hour of developer work. For a catalog with 30,000 SKUs it is a multi-week project, and resizing is the easy part. Two things make it hard: some products have no high-resolution source anywhere, and when you replace a file at an existing URL, Google can take up to six weeks to notice. The real deadline for a bulk migration is not the end of January — it is mid-December 2026.

This guide covers the exact Merchant Center image requirements, how to find affected products, a six-step migration plan, the full image policy beyond pixel count, and what to do with products that will never have a decent photo.

Why the stakes are higher than they look

Your image_link stopped serving only Shopping ads a long time ago. The same URL now feeds the Shopping Graph, and through it:

• Shopping ads and Performance Max,
• free product listings,
• AI Mode results and shopping journeys in Gemini,
• visual search and virtual try-on in Google Lens.

An image disapproval pulls the product from all of those at once. That is not “fewer Shopping impressions” — that is the product leaving Google’s commerce ecosystem entirely. And the more of the purchase the algorithm handles on its own — as with AI Mode native checkout, where the product card effectively replaces your storefront — the more every weak image costs.

The exact requirements

Parameter	Value
Minimum from Jan 31, 2027	500 × 500 pixels, all categories, no exemptions
Recommended size	1500 × 1500 pixels or larger
Pixel ceiling	64 megapixels
File size cap	16 MB
Formats	JPEG, WebP, PNG, GIF, BMP, TIFF
Product coverage (best practice)	75–90% of the frame
Background	Solid white or transparent
Where to spot problems	Needs attention section in Merchant Center (since April 2026)

Google notes that some smaller images will be automatically optimized to meet the new requirement. The scope of that automation is unpublished, so it cannot be the basis of a migration plan. Assume it will not save you; treat it as a bonus if it does.

The crawl trap that moves your deadline forward

A new image URL is recrawled within 24–72 hours. A changed file at an existing URL can take up to six weeks to be detected.

This is the most underestimated line in the whole announcement. The classic failure: in mid-January the team swaps images on the server, keeps the same paths, reports the project complete — and gets a wave of disapprovals at month end because Google is still reading the old files.

The fix is simple: on a bulk replacement, change the URL itself — version it in the path or a parameter (/img/12345-v2.jpg) so the crawler treats it as a new resource and processes it within days. And do it early. For a large catalog, the last comfortable window is mid-December 2026, leaving a full six-week buffer before enforcement.

How to find affected products

1. Merchant Center → Needs attention. Undersized image warnings have been live since April 2026. Fastest way to size the problem.
2. Feed diagnostics. The full method for reading disapprovals and warnings is in product disapprovals and feed diagnostics.
3. Your own catalog sweep. More reliable than waiting for a report: script every image_link and build a table of SKU → width × height → file size. For 30,000 products that is a few hours of runtime and a complete picture instead of a sample.
4. Join it to revenue. Attach trailing twelve-month revenue and margin to the affected SKU list. Everything after this runs in order of money, not alphabet.

Six-step migration plan

1. Inventory

Export the full product list with actual image dimensions. Split into three buckets: comfortably compliant (≥1500 px), borderline (500–1499 px), failing (<500 px).

2. Prioritize by revenue

Failing high-revenue products first, then the failing tail, then borderline items that look weak in large formats. Custom labels in Merchant Center are the practical tool here: tag image status as a label and track progress inside your normal reports.

3. Source better images

In descending order of quality: studio originals or PIM assets, supplier or vendor imagery, reshoot, AI upscaling. Upscaling is a last resort — it satisfies the pixel count but can introduce artifacts and plasticky textures that hurt trust at large display sizes. For top SKUs, reshoot.

4. Process to one standard

Square 1500 × 1500, white or transparent background, product filling 75–90% of the frame, nothing cropped at the edges. A single template across the catalog saves weeks of debate and looks cleaner than competitors in the same results row.

5. Publish at new URLs

New file paths, not overwrites. Update image_link and, where relevant, additional_image_link. If editing the primary feed is painful, apply the changes through feed rules and supplemental feeds instead of rewriting the export from your commerce platform.

6. Verify

Check after 3–5 days that new images are indexed and warnings cleared. Re-check at two weeks, and again two weeks before the deadline. For large catalogs, automate the check through the Merchant API.

Image policy: more than pixel count

Size is the visible requirement, but content is what gets most images rejected. Prohibited in the main image:

• watermarks and overlaid captions;
• barcodes;
• borders;
• promotional text, calls to action, prices and discount badges;
• brand or retailer logos, unless they are part of the product itself;
• placeholders, generic illustrations or logos in place of a product photo (narrow exceptions exist for Hardware, Vehicles and Software).

The main image must show the complete product with minimal or no staging. Lifestyle compositions with models, props and captions belong in additional images, not the primary one.

One recurring mistake deserves a callout: pricing burned into the image. It usually happens when a store pushes the same creative it made for social into the product feed. Promotions in Google are expressed structurally — through Merchant Center promotions and annotations and promotion mode — not as text on a photo.

Products that will never have a good photo

Every real catalog has them: discontinued models, one-off clearance units, supplier items whose only asset is a 2014 thumbnail. Make the call deliberately and early:

• Exclude them from the feed on purpose — a clean feed of 24,000 products beats 30,000 with a February disapproval wave.
• Retire or clear them out if they were tail items anyway.
• Batch a shoot for products that actually earn. Per SKU, studio work is usually cheaper than teams assume, especially for small items.

“Leave it and hope the auto-optimization covers us” is not a plan. Google has not published how far that automation reaches, and product visibility is too expensive to gamble on an undocumented fallback.

Technical details teams trip over

• Format. JPEG stays the safe default for photography; WebP delivers smaller files at equal quality and is supported. PNG is for genuine transparency. GIF, BMP and TIFF are allowed but rarely make sense in a catalog.
• Aspect ratio. Square is not formally required, but it renders most predictably across card formats. For long-form products (curtains, fishing rods, guitars), pad into a square rather than cropping the product.
• Weight and speed. 16 MB is the ceiling; the practical target is 200–500 KB at 1500 px. Heavy assets slow more than crawling — they hit your own landing pages, as covered in landing page speed and Core Web Vitals.
• CDN transforms in the URL. If images are served through a CDN with on-the-fly resizing (?w=300), verify the feed carries the full-size URL and not the thumbnail used on category pages. This is the single most common cause of “but our photos are good, why the disapproval”.
• Crawler access. robots.txt rules, authentication, User-Agent blocks and aggressive rate limits can prevent the image from being fetched at all. Same logic as landing page access — see landing page crawlability.
• URL stability. Do not generate image URLs with a fresh hash on every site rebuild; each deploy then looks like a brand new asset and burns processing cycles for nothing.

Scoping the work and the budget

To have a concrete conversation with leadership, size it in three lines:

1. How many SKUs fail — the exact count from the inventory step, not “roughly a third”.
2. How many can be fixed without a camera — originals exist in the archive, the PIM or with the supplier. Typically 40–70% of failing items, and the cheapest part of the job: batch processing against a template.
3. How many need a shoot — this is where money appears. Decide by revenue: if a SKU did not earn the cost of its own photography in twelve months, it does not get photographed.

Planning rule of thumb: batch processing is minutes per SKU and scales linearly with a script; photography is days and is the bottleneck. Start the project by splitting the two streams so processing never waits on the studio.

The only prioritization metric that survives contact with a real catalog is trailing twelve-month revenue per SKU. Everything else turns the project into endless tail work.

Migrating without pausing sales

• Move in batches, not the whole catalog overnight. One category, verify statuses, next. A misconfiguration costs you one category for a few days instead of the entire feed.
• Do not change other attributes at the same time. Swapping titles, categories and images on the same day makes diagnosis impossible — title work has its own playbook in feed rules.
• Watch hero products individually. The two dozen SKUs carrying your revenue deserve manual verification after the swap, not a line in an aggregate report.
• Keep a rollback. Do not delete the old files immediately — if a new batch gets rejected on image content, being able to restore yesterday’s assets for a day or two protects revenue.

Common mistakes

• Upscaling without a source. A 200 × 200 stretched to 500 × 500 passes the pixel check and loses the sale in every large placement.
• Overwriting files at the old URLs. Six weeks of detection lag, discovered in February.
• Working alphabetically instead of by revenue. Week two and the team is still on accessories while core categories are untouched.
• Inconsistent standards across batches. Half the catalog on white, half on grey, all of it displayed side by side with competitors.
• Ignoring additional images. Product image requirements apply to secondary images too.
• Scheduling the swap into peak season. Changing imagery during a promotional peak is strictly worse than doing it before — see preparing for promotional peaks.

Brands and resellers have different jobs

The same deadline translates into very different work depending on whose product it is.

If you are the manufacturer or brand owner, imagery is an asset that drives your sales and every partner’s sales at once. Publishing an updated, compliant photo pack for dealers ahead of the deadline is far cheaper than untangling why your product vanished from listings at half your retail partners in February. It also means you control how the brand looks in commerce results instead of hoping a reseller re-saves an image out of a PDF catalog.

If you are a reseller, you inherit whatever the supplier provides, and there is a second problem beyond pixel count: dozens of stores use the identical image. In a results row your card is visually indistinguishable from competitors, so the decision collapses to price and seller rating. Shooting your own photography for fast movers is one of the few ways to create visual differentiation — worth doing only for products that actually earn. For the rest, the job is narrower: bring supplier images up to spec without violating policy (no added badges, no store logo overlay).

One detail specific to resellers: a brand logo in the image is acceptable when it is genuinely part of the product or packaging. Overlaid badges and watermarks are not, and they are a recurring disapproval cause in categories where the feed is generated automatically from a supplier price list.

Checklist

1. Export the catalog with real image dimensions for every SKU.
2. Bucket products: failing / borderline / compliant.
3. Join trailing twelve-month revenue and margin, sort by money.
4. Decide the image source per bucket.
5. Make the call on hopeless SKUs: shoot, exclude or retire.
6. Process to one standard: 1500 × 1500, white or transparent, product at 75–90%.
7. Publish at new URLs and update the feed.
8. Finish the bulk replacement by mid-December 2026.
9. Verify at 3–5 days, at two weeks, and two weeks before January 31, 2027.

Beyond compliance

500 × 500 is the entry bar, not the level that wins. Commerce results in 2026 put your photo next to a competitor’s in an identical frame, where the only differences visible at a glance are image quality and price. That makes this migration a good moment to also address the two levers with the largest effect on CTR and conversion: price competitiveness and seller and product ratings.

There is a second reason not to rush it in January. Adjacent 2026 changes hit the same surface area: new inventory inside Performance Max widens where products appear, and agentic purchases turn the product card into the storefront. A catalog cleaned up once pays off across all of them.

If the catalog is large and the team is small, split the work: outsource image production, keep feed, campaign and account infrastructure close. What we handle is listed on the PPC Rebels services page, including Google Ads agency accounts for teams running several stores and markets.

FAQ: Merchant Center product image requirements

When does the 500 × 500 requirement take effect?

January 31, 2027. Images below 500 × 500 pixels stop being approved from that date; Merchant Center has warned about them since April 2026.

Are apparel or other categories exempt?

No. The 500 × 500 minimum applies uniformly across all product categories.

What happens to non-compliant products?

They are disapproved and removed from the Shopping Graph — Shopping ads, free listings, AI Mode and Gemini shopping experiences, and visual search.

Will Google upscale small images automatically?

Google says some smaller images will be automatically optimized, but the scope is unpublished. Planning around it is a risk.

What size should I upload to be safe long term?

1500 × 1500 or larger, which is Google’s own recommendation for performance across placements. Ceilings are 64 megapixels and 16 MB per file.

Why can’t I just overwrite the file at the same URL?

Content changes at an existing URL can take up to six weeks to be detected, while a new URL is recrawled in 24–72 hours. For bulk work, change the URLs.

Fastest way to list every affected product?

Start with Needs attention in Merchant Center, then script your catalog to read actual dimensions from every image_link for full coverage.

Can I keep a brand logo in the image?

Only when it is genuinely part of the product. Overlaid retailer or brand logos, watermarks, borders, barcodes, prices and CTAs are not allowed.

How do I show a discount if price text is banned?

Through the structured path: Merchant Center promotions and annotations, plus promotion mode in campaigns. It is also measurable, unlike text baked into a JPEG.

Does this apply to additional images?

Yes. Product image requirements cover secondary images that serve alongside the main one.

What is the realistic completion date for a bulk swap?

Mid-December 2026. That preserves a full six-week buffer for detection and for fixing whatever goes wrong.

What if good source images simply do not exist?

Decide by revenue: shoot the items that earn, and deliberately exclude or retire the tail. A clean feed always beats a large feed full of disapprovals.