Is Google really cropping your recipe images if you fail to provide the right formats?

Why does Google require multiple image formats for recipes?

Recipe rich results appear on various platforms — mobile, desktop, Google Discover, carousel. Each context demands a specific width/height ratio to optimize the available visual space.

When you provide only a single image, Google has to adjust it for different formats. Autocropping then becomes the only technical option for ensuring your visual integrates well into each interface. The problem? The algorithm does not know what matters in your photo — it cuts mechanically.

What are the required image formats for recipes?

Google's documentation specifies three main ratios: 16:9 (landscape), 4:3 (standard), and 1:1 (square). Each ratio corresponds to a specific display context — 16:9 for wide banners, 1:1 for rectangular grids in Discover.

Without these variants, Google applies an automatic crop that can ruin your composition. A vertical photo cropped to 16:9 often loses the essence of the dish. A horizontal image cropped to 1:1 cuts off the sides — goodbye context or peripheral ingredients.

How does Google decide which area to crop?

Müller remains vague about the cropping algorithm. It is assumed that it uses object detection and visual focus to identify the "central" area — but nothing has been officially confirmed.

This lack of transparency is problematic. You cannot anticipate where Google will cut. A deliberately off-center dish for aesthetic effect? Risky. A setup with multiple elements? Unpredictable. Hence the imperative to provide your own versions.

• Three minimum required ratios: 16:9, 4:3, 1:1 to cover all displays
• Automatic cropping is unpredictable and can degrade the visual composition
• No quality guarantee on what Google will keep from the original image
• The official documentation precisely lists the expected formats for each type of rich display
• Providing the right variants = total control over what the user sees

Is this statement consistent with observed practices on the ground?

Yes, and it is confirmed by tests. Sites that provide only one image format indeed see their visuals randomly cropped in rich results. Some lose up to 40% of the useful surface area depending on the display context.

What is missing in Müller's statement: no quality metrics. Google does not say whether automatic cropping penalizes CTR or ranking. However, negative correlations are observed — results with poorly cropped images generate fewer clicks. [To be verified] with large-scale controlled A/B tests.

What nuances should be added to this rule?

Müller talks about “required formats,” but the reality is more flexible. Google still displays your rich results even if you provide only one format — it crops, that’s all. This is not a strict eligibility criterion, unlike other properties of Recipe markup.

Moreover, not all ratios are equal. The 16:9 format dominates on mobile and in Discover. If you must prioritize, start with this one. The 1:1 ratio remains crucial for grids. The 4:3 ratio? Less critical — [To be verified] based on your enriched traffic source analytics.

When does this optimization become truly critical?

For high-volume recipe sites, every CTR point counts. If your images are already professionally done, automatic cropping degrades a significant visual investment. Here, providing the variants is non-negotiable.

On the other hand, for an amateur blog with basic smartphone photos, the impact remains marginal. Google crops an average image into another average image. The effort of producing multi-format might not necessarily be justified — it’s better to first improve the inherent quality of your visuals.

What concrete steps should be taken to provide the right formats?

Produce three versions of each recipe image: one in 16:9 (minimum 1200x675px), one in 4:3 (1200x900px), one in 1:1 (1200x1200px). Declare them in the Schema.org Recipe markup using the image property with an array of URLs.

Two possible approaches. First option: manually crop each photo in an editor to precisely control the composition. Second option: use a script that generates the variants automatically, but by defining priority focus areas so that the crop preserves the essentials.

What mistakes should be avoided when producing multi-format?

Do not simply stretch or compress an image to reach the target ratio. A 1:1 photo stretched to 16:9 appears distorted — Google might even reject it. You need to crop intelligently, not resize thoughtlessly.

Another pitfall: providing variants in too low resolutions. Google recommends a minimum of 1200px on the longest side. Below that, visuals pixelate on desktop and some rich formats. Your multi-format efforts become counterproductive if the image quality doesn’t follow suit.

How can you verify that Google is indeed using your variants?

Use the Rich Results Test tool from the Search Console to validate that your three image URLs are correctly detected. Then inspect your pages under real conditions — mobile, desktop, Discover if you are eligible.

Monitor your click metrics on rich results in the Search Console. A jump in CTR after implementing the variants confirms the impact. If there’s no change, either your initial images were already properly cropped by Google, or cropping was not your performance issue.

• Produce 3 minimum versions of each image: 16:9, 4:3, 1:1
• Respect the minimum resolution of 1200px on the longest side
• Declare the URLs in an array via Schema.org Recipe > image
• Test with Google’s Rich Results tool
• Check the actual rendering on mobile, desktop, and Discover
• Monitor the CTR of rich results before/after implementation