Would you rather open a generic email trying to tell you about a product that might or might not suit your needs, or an email that takes into account you, as a consumer, linking your needs and taste to an aspect of a product? Put more simply, if you’re a runner, would you rather look at swim suits or shoes?
Personalization plays an important role in modern email marketing. Most subscribers receive dozens of emails each day packed with easy-to-ignore images and messaging. When marketing emails feel relevant both in messaging and visuals, recipients are more likely to engage with your awesome product!
Industry studies consistently show that personalization can improve engagement metrics such as open rates, click-through rates, and conversions, though results vary by audience, industry, and execution. Visuals are a key part of this equation: Images that align with a recipient’s interests, intent, or context can reinforce relevance and support stronger engagement. TikTok’s powerful algorithm that creates a profound visual experience is a great example of how personalization can work.
In this post, we’ll explore how Cloudinary AI Vision can be used as an image analysis and metadata signal layer within an email personalization workflow, and how those insights can inform the creation of tailored image assets for different audience segments.
Images in email campaigns serve as fast signals that are designed to trigger an emotional response. A recipient scanning their inbox or reading an email body may not parse every word, but they’ll notice whether an image feels relevant to them.
Examples include:
- Showing tennis rackets or water sports imagery to customers already browsing summer gear.
- Highlighting lifestyle imagery versus product close-ups depending on audience preference as designated by email segmentation.
- Aligning color palettes or visual themes with campaign intent or seasonality.
Cloudinary AI Vision helps by analyzing images at scale and extracting structured information, such as objects, colors, scenes, and text, that can be used by marketers and developers to organize, filter, and select visuals more intentionally.
Importantly, AI Vision does not personalize emails on its own. Instead, it provides image-level insights that can be combined with customer data and campaign logic in external systems.
Cloudinary AI Vision uses multimodal large language models (LLMs), combined with Cloudinary’s image AI capabilities, to analyze visual content and return descriptive and classificatory information, driving automation of key processes, including content moderation, image classification, and custom tagging.
Core capabilities include:
- Image description and analysis. Generate detailed, context-aware descriptions of images, including objects, scenes, activities, and in-image text.
- Classification and tagging. Automatically extract attributes such as dominant colors, themes, environments, and subjects to improve searchability and organization in your media library.
- Content moderation signals. Identify potentially sensitive or inappropriate content, helping teams flag assets that may require review before use.
What AI Vision does not do:
- It does not generate new images or image variations.
- It does not decide which images will perform best.
- It does not understand customer segments or marketing performance.
- It does not run A/B tests or optimize campaigns.
Instead, AI Vision acts as a signal provider, supplying structured metadata that can support downstream decision-making by humans or other systems. Let’s test it out.
Here are some steps you might need to take to launch an email campaign:
- Ensure that your customer and behavioral data lives in your CRM, CDP, or email service provider (ESP).
- Load up your images and media assets in Cloudinary.
- Use AI Vision to analyze those images and add descriptive metadata.
- Use the metadata to select or group those images.
- Create any image variations manually or programmatically.
- Assemble your email campaigns and measure them in your ESP.
Let’s dig into Step 3.
Lucky you! You own a pet store and want to send one campaign to dog owners and another, separate one, to cat owners. Let’s assume that you’re not targeting those households managing multiple pets.
To use Cloudinary AI Vision, you need to install an “add-on” from the marketplace, so click the marketplace button in the left navigation of the console and add it to your account:
Upload your campaigns’ images to your Cloudinary Media Library as you normally would.
Note, you don’t necessarily have to have this image in your own account; you’re going to use an API call to return information about your image by using an LLM to ask questions about it.
Using the Analyze API, analyze images to extract descriptive metadata. The API can give accurate information about images so that you can triage them into your campaigns. The image of the cat above, for example, provides the following information when sending in its URL and a prompt:
{
"source": {
"uri": "https://res.cloudinary.com/beanpot-studio/image/upload/v1766525276/photo-1573865526739-10659fec78a5_tdfxin.avif"
},
"prompts": [
"Describe this image in detail",
"Does this image contain a dog?"
]
}
{
"request_id": "08e513d66447408d38f80002d75b934b",
"data": {
"entity": "https://res.cloudinary.com/demo/image/upload/sample.jpg",
"analysis": {
"responses": [
{
"value": "This is a close-up, eye-level shot of a fluffy, orange tabby cat with striking features, captured in a vertical orientation. The cat is lying on a rustic wooden surface, possibly a table or floor, with its front paws extended forward and its body slightly angled.\n\nThe cat's fur is a vibrant ginger or orange color, with subtle tabby markings visible, particularly on its head and legs. It has long, wispy white whiskers that fan out from its muzzle. Its eyes are a bright, inquisitive amber or light brown, and it's looking upwards and slightly to the right, suggesting it's focused on something out of frame. Its ears are large and pointed, characteristic of breeds like a Maine Coon, with tufts of fur inside.\n\nThe lighting appears soft and natural, highlighting the texture of the cat's fur and creating a gentle glow around its head. The background is softly blurred, indicating a shallow depth of field, which helps to keep the focus entirely on the cat. Hints of a domestic interior can be seen in the background, with muted tones of light blue or grey and some dark, indistinct shapes that might be furniture or objects.\n\nThe overall mood of the image is one of curiosity and warmth, with the cat's engaging gaze being the central point of interest. The rich colors and sharp focus on the cat make it a captivating portrait of a feline companion."
},
{
"value": "No, the image does not contain a dog. It features a cat."
}
],
"model_version": 1
}
}
}Code language: JSON / JSON with Comments (json)Based on AI Vision’s output, you might:
- Group images classified as “cat”, “orange”, or “feline” for cat toy campaigns.
- Identify images with bold, high-contrast colors and match email font colors.
- Filter out images with elements that don’t align with a campaign’s intent.
At this stage, human judgment or custom logic determines which images are appropriate for which audience segments.
Once you’ve selected images for a campaign, you can create variations using Cloudinary Studio, which provides a visual interface for applying transformations at scale.
For example:
- Removing and replacing backgrounds.
- Applying overlays or brand colors.
- Adjusting crops or aspect ratios for email layouts.
This dog’s orange collar, for example, can be converted to green with just a quick prompt:
These tools work in tandem: AI Vision insights can help inform which images to modify, and Studio is where the actual asset creation and editing can happen.
Pairing your email management tools with Cloudinary’s AI-powered image analysis and editing can create great targeted campaigns. Once set up and segmented, with appropriate images added, you can:
- Compare performance across campaigns that used different image styles.
- Correlate engagement metrics with image attributes identified by AI Vision such as color dominance or lifestyle vs. product imagery.
And of course, you can use those learnings to guide future creative decisions. Over time, this creates a feedback loop in which performance data will grow to inform your creative strategy.
Paired with email tooling, Cloudinary AI Vision can help you build and evolve your personalized campaigns by:
- Making large image libraries more understandable.
- Providing structured, searchable visual metadata.
- Enabling teams to scale image analysis without manual tagging.
When used as a signal layer within a broader personalization pipeline, AI Vision can help teams make more informed creative decisions. If you’re already using Cloudinary to manage and deliver images, AI Vision can be a valuable addition to your workflow for organizing and understanding visual content at scale. And if not, give it a try for your next personalized online campaign!
