AI-Powered Setup

FilterIQ can use available catalogue data to prepare a first draft and suggestions for merchant review. The analysis is bounded by plan and service limits, may return no suggestions, and does not establish that a configuration is correct or ready to publish. Timing varies with catalogue coverage and conditions.

How It Works

Step 1: Catalog Analysis

Use the analysis controls available to your preview store. Check which products and fields are included, then review:

1. Product titles, descriptions, tags and existing metafields available to the scan
2. Variant options such as sizes, colours and materials
3. Candidate attributes with enough consistent data to evaluate as filters
4. Suggested categories or groupings against representative products

Step 2: Configuration Generation

Inspect the suggestions returned by the selected workflow. Use this checklist rather than assuming that each analysis configures every setting:

Search configuration:

• Which product fields to search (and their relative weights)
• Typo tolerance against misspellings found in your own catalogue and query data
• Candidate synonym pairs from catalogue evidence; approve only after checking the intended matches

Filter configuration:

• Which attributes become filters (e.g., color → color swatch filter, size → size selector)
• Filter display types matched to the data type (ranges for prices, swatches for colors, checkboxes for tags)
• Filter ordering that reflects store priorities and observed usage
• Mobile layout and keyboard interaction in the theme preview

Merchandising presets:

• Recommended presets based on your inventory patterns
• For example, if you have many out-of-stock items, it suggests "Hide Out-of-Stock"

Step 3: Review and Publish

Review suggestions before accepting them. Confirm which settings are drafts and which changes affect the store immediately:

• Toggle individual filters on/off
• Change filter display types
• Adjust search field weights
• Enable/disable merchandising presets
• Preview how search and filters will look on your store

Use the available preview workflow on a duplicate theme, verify the resulting products and controls, then publish the accepted configuration. Check publish status and the storefront result; delivery timing depends on the sync and theme path.

What the AI Detects

Supported Filter Types

The following are display examples to evaluate where supported by the available configuration. A source attribute alone does not guarantee that analysis will create a filter:

1. Color swatches — Visual color circles with hover preview
2. Size selectors — Button-style size selection (S, M, L, XL...)
3. Range sliders — For supported numeric fields, with units and bounds verified
4. Star ratings — 1-5 star rating filter
5. Checkbox lists — Multi-select for tags, types, vendors
6. Boolean toggles — On/off for attributes like "In Stock", "On Sale"
7. Tag clouds — Visual tag display for browsing
8. Date ranges — For date-based attributes

For collection, variant, metaobject or compound requirements, confirm the supported source mapping and display controls during onboarding before relying on them.

Re-running AI Setup

When catalogue data changes, check the available analysis controls and any applicable usage limits:

1. Open the app and record the current accepted configuration.
2. Check sync status and inspect the analysis scope before starting another scan.
3. Review the returned candidates and any reported errors before adopting changes.

This is useful when:

• You've added many new products
• You've restructured your product metafields
• You want a fresh configuration recommendation

A new suggestion is not an approved change. Confirm the save/apply behaviour of the selected workflow and retain a recovery reference before accepting replacements.

Manual Override

AI Setup is a starting point. Review and adjust the supported settings for your store:

• Search Configuration — Adjust search weights and behavior
• Filter Types Reference — Change or add filter types
• Merchandising Controls — Fine-tune product ranking

Privacy and Data

During AI Setup:

• Catalog context (product titles, descriptions, tags, variants) is sent in minimized and redacted form to approved AI subprocessors, including Anthropic and OpenAI, to generate setup suggestions. See our Privacy Policy for the complete subprocessor list.
• We configure AI subprocessors according to our data-processing agreements and retention controls where available
• Analysis results are stored only for your store's configuration
• You can delete all data at any time by uninstalling the app — product data is removed from our database and search index within 48 hours of uninstallation (see Data Deletion)