Image Analysis uses AI to understand photos uploaded or taken by shoppers during a quiz. Instead of returning only a text description, it can generate structured answers such as numbers, multiple-choice values, and yes/no responses.
These answers work like other quiz answers, allowing you to personalize results, recommend products, and create logic based on what the image shows.
What can Image Analysis do?
Image Analysis can return several types of information from a single photo:
Description: A written explanation of what the image shows.
Number: An estimated value, such as a dog's weight.
Choice: A selection from predefined options, such as skin tone or hair color.
Yes/No: A true-or-false answer based on the image.
Each value has its own name and can be used throughout your quiz. You can use image-derived answers to:
Recommend products based on the photo.
Create branching logic and visibility conditions.
Calculate sizes, quantities, portions, or scores using formulas.
Display personalized text and reuse the shopper's photo later in the quiz.
Send information to integrations, webhooks, Shopify customer profiles, and exports.
Add Image Analysis to your quiz
Open your quiz in the editor.
Add an Image Analysis step where you want shoppers to upload or take a photo.
Configure what the photo should show.
Define the information you want AI to return.
Set image requirements and configure what happens if an image can't be analyzed.
Use Preview to test the experience before publishing.
Tip: You can also ask the AI assistant to create an Image Analysis step.
Configure your image analysis
Image Analysis setup includes four main parts.
1. Define what the photo should show
Describe the type of image shoppers should provide. For example, you might ask for a clear photo of a pet, a selfie, a room, or a plant. You can use available templates to help configure your instructions.
2. Choose what AI should return
Define the answers you want AI to extract from the photo. For example, for a skincare quiz, you might configure:
Answer | Type | Possible result |
Skin tone | Choice | Fair, Light, Medium, Tan, Deep |
Warm undertone | Yes/No | Yes or No |
Image description | Description | A short description of the visible features |
Each extracted value can be referenced separately in your quiz.
Tip: Use predefined choices when you want to build predictable recommendation logic. Use numeric values when you need calculations.
3. Add image checks
Set requirements that an image must meet before it can be analyzed.
For example: The image must contain a dog. The subject must be clearly visible. The photo must be sharp and well-lit.
If an image fails a check, the shopper is told which requirement wasn't met and can try again.
4. Configure unsuccessful analyses
Choose what happens when AI can't reliably interpret an image. When AI can't determine a value, it returns unknown rather than guessing. Configure an appropriate fallback experience so shoppers can continue when a photo doesn't provide the information needed.
Use image answers in your quiz
Once an image is analyzed, its extracted values can be used like other quiz answers.
Personalize product recommendations
Use image-derived answers to select products that match what AI identified. For example, a makeup quiz could use a detected skin tone to recommend suitable foundation shades.
Create branching logic
Use extracted values in Jump Rules or Visibility Rules to control which pages, questions, or elements shoppers see. For example, a pet quiz could show different recommendations depending on the dog's estimated size.
Perform calculations
Numeric answers can be used in formulas to calculate personalized values. For example, an estimated weight could be used to calculate a suggested daily food portion.
Display personalized information
Use extracted values in text elements to personalize the shopper's experience.
You can also display the shopper's uploaded photo later in the quiz. For more information about conditional content, see Use branching logic and visibility rules in your quiz.
Understand image checks and privacy
Image Analysis includes safeguards to help protect shoppers and improve the reliability of results.
Image moderation: Photos are moderated before AI analysis. Explicit or violent images are rejected.
Image requirements: Photos that don't meet your configured checks are rejected with an explanation.
Consent: A consent message appears before an image is submitted, with an optional 18+ confirmation.
Private storage: Uploaded photos are stored privately and deleted after 30 days.
Attempt limit: Each shopper can try up to 10 times per Image Analysis block.
Uncertain results: AI returns unknown when it can't determine an answer reliably.
If a shopper skips the image step, their uploaded photo is deleted.
Track Image Analysis performance
Image Analysis includes analytics to help you understand how shoppers interact with the feature.
You can review:
How many shoppers submitted a photo.
How many photos were successfully analyzed.
Which image checks fail most frequently.
Use these insights to improve your instructions and make the upload experience easier. Image-derived answers are also available in People, exports, the API, webhooks, integrations, and Shopify customer sync.
Image Analysis credit usage
Each photo analysis uses 0.3 credits. Image Analysis consumes credits when AI processes a photo. These credits are separate from the credits used when a shopper starts a quiz. You can monitor your credit usage under Billing in your Octane AI dashboard. For more information, see Octane AI plans and pricing: Credits system.
Examples of Image Analysis
Image Analysis can support different types of quizzes:
Skincare and makeup: Analyze a selfie to estimate skin tone and undertone for shade recommendations.
Pet products: Estimate a pet's size or weight to recommend food and calculate portions.
Hair care: Identify visible hair color and characteristics to recommend a routine.
Home decor: Analyze room colors and style to suggest matching products.
Plant care: Analyze a plant or its surroundings to provide relevant care recommendations.
Preview and test your Image Analysis
Before publishing, test your Image Analysis step with different photos.
Check that:
Upload and camera options work as expected.
Instructions clearly explain what shoppers should photograph.
Image checks correctly accept or reject photos.
Extracted values match your configured answer types.
Quiz logic and recommendations use the extracted answers correctly.
Shoppers can continue when an image can't be analyzed.
Tip: Test both successful and unsuccessful photo submissions to make sure the experience works for different shoppers.