Skip to main content

Custom properties

Overview​

Custom properties let you capture attributes specific to your business that don't fit into the standard product fields (price, brand, category). A custom property has a data type — number, text, checkbox, date, or multiple choice — and is assigned to each product with its own value. This is what lets the AI agent filter precisely on questions like "products under 10 kg", "color blue", "waterproof", or "launched after 2024".

How to Access​

  • Navigate to Sales Catalog > Settings > Custom properties (/sales/settings#properties).
  • Also available at the direct route /sales/properties.
  • Required role: Administrator.

Interface Overview​

The screen shows:

  • List of defined properties — Table with name, data type, number of products using it, and actions (edit / delete).
  • Add property button — Opens the form to define a new one.
  • If the property is Multiple choice, you'll also see the list of available options.

Once the property is defined, you assign values to each product from the product detail page (not from this screen).

Data types​

Aurora supports five types. Pick the type based on how you want the agent to filter.

Number​

For magnitudes with implicit units: weight, volume, voltage, count. Enables "greater than", "less than", or "between X and Y" filters.

Examples: Weight (kg), Voltage (V), Capacity (L), Page count.

Text​

For free-form labels. Enables exact-match and "contains" filters. Useful when the values are too varied to list as options.

Examples: Main material, Technical description, Origin, Manufacturer model.

Checkbox (boolean)​

A yes/no property. Enables "yes" or "no" filters.

Examples: Waterproof, Vegetarian, Gluten-free, Extended warranty included.

Date​

For dates relevant to the product. Enables "before", "after", or "between two dates" filters.

Examples: Launch date, Expiration date, Promotion start.

Multiple choice​

A property with a closed list of predefined values. Enables exact-match and "any of" filters. The best option when values repeat across many products and you want the agent to recognize them unambiguously.

Examples: Color (Red, Blue, Green, Black…), Size (XS, S, M, L, XL), Style, Fabric type.

Features & Actions​

Create a custom property​

What it does: Defines a new attribute you can assign to products.

Steps:

  1. Click Add property.
  2. Enter a clear Name (e.g., "Weight", "Color", "Waterproof").
  3. Choose the Data type (Number, Text, Checkbox, Date, Multiple choice).
  4. If you chose Multiple choice, add the list of available options (one per line or in the editor).
  5. If you chose Number or Date, you can define a suggested range or a unit of measure.
  6. Click Save.

Important notes:

  • The name is what the AI agent uses internally to map customer questions. Pick clear, consistent names ("Weight" is better than "Wgt" or "Product weight").
  • Once created, we do not recommend changing the data type — the values already assigned would become inconsistent. If you need to change, create a new property.

Assign a value to a product​

What it does: Sets the property's value for a specific product.

Steps:

  1. Open the product detail page (/sales/products/:id).
  2. Go to the Custom properties section.
  3. For each property defined in the catalog you'll see an appropriate field or selector.
  4. Set the value and save.

Edit a property​

What it does: Changes the name or (if applicable) the available options.

Steps:

  1. In the property list, click Edit.
  2. Adjust the name or (for multiple choice) add/remove options.
  3. Save.

Important notes:

  • Renaming a property does not break the values already assigned.
  • Removing a Multiple choice option that's in use requires that no product has it assigned.

Delete a property​

What it does: Removes the property and the values assigned to products.

Steps:

  1. In the property list, click Delete.
  2. Confirm.

Important notes:

  • This action deletes the values assigned across every product. It cannot be undone.

How the AI agent uses them​

When a customer asks a question, the AI agent looks at the names and types of your custom properties and builds structured filters as appropriate:

  • "Products under 10 kg" → filter Weight < 10 (if Weight is of type Number).
  • "In color blue" → filter Color = Blue (if Color is of type Multiple choice with Blue as an option).
  • "Waterproof ones" → filter Waterproof = Yes (if the property is of type Checkbox).
  • "Launched after 2024" → filter Launch date > 2024-01-01 (if it's of type Date).
  • "With cotton material" → filter Material contains "cotton" (if it's of type Text).

If the customer doesn't mention an attribute, the agent does not automatically add it as a filter — it doesn't assume context the customer didn't provide. This preserves the quality of results.

Important Notes​

  • Few and well-named works better than many and vague. A dozen clear properties is preferable to 50 fuzzy ones.
  • Multiple choice is the most reliable type for colors, sizes, and styles: the agent recognizes them unambiguously when the customer mentions them.
  • Numeric values should be in the same unit per property. Don't mix "10 kg" with "10000 g" in the same property — pick one unit and stick with it.

FAQ​

Q: What type do I pick for "color"? A: Multiple choice with your closed palette. That way the agent distinguishes "red" from "dark red" if you list them as separate options.

Q: Can I assign several values of a multiple choice to the same product? A: Yes, for Multiple choice properties you can assign more than one option to the same product where it applies (for example, several available colors).

Q: Are properties imported from CSV? A: Yes. The importer supports columns mapped to custom properties. See Catalog import.

Q: Does the AI agent know about all the properties automatically? A: Yes, when the catalog is activated for the agent, it receives the list of properties, their types, and (for multiple choice) the available values. You don't have to configure this per agent.