AI Store Scanner

The AI Store Scanner examines your entire product catalog and generates actionable recommendations for filters, synonyms, and search optimization. Instead of manually reviewing thousands of products, let the scanner do the analysis for you.

What the Store Scanner Does

The scanner reads your indexed product data and identifies patterns that inform your search and filter configuration. It looks at:

• Product attributes -- Which fields (vendor, product type, tags) have enough distinct values to make useful filters
• Tags -- Common tag patterns and groups that map to filter categories
• Metafields -- Custom metafields that could enhance search or filtering
• Variants -- Option names and values across your catalog (Color, Size, Material, etc.)
• Text patterns -- Common terms in titles and descriptions that suggest synonym opportunities

Running a Scan

1. Navigate to AI > Store Scanner in the app dashboard (or go directly to /app/store-scan)
2. Click Start Scan
3. The scanner processes your catalog -- progress is shown in real time

Scan time depends on your catalog size:

You can navigate away from the page while the scan runs. A notification appears in the dashboard when it completes.

Scan Results

After the scan completes, you see a results page with three sections:

Recommended Filters

The scanner suggests filters based on your product data, ranked by usefulness:

• Filter name -- The attribute to filter on (e.g., "Color", "Brand", "Material")
• Filter type -- The recommended UI type (checkboxes, swatches, range slider)
• Coverage -- What percentage of your products have this attribute populated
• Distinct values -- How many unique values exist (e.g., 12 colors, 45 brands)

Filters with higher coverage and a reasonable number of distinct values (5-50) are ranked higher.

Suggested Synonyms

The scanner identifies terms in your catalog that shoppers might search for differently:

• Product type variations (e.g., "tee" and "t-shirt")
• Brand abbreviations (e.g., "N&F" and "Naked & Famous")
• Common alternate spellings or regional terms (e.g., "grey" and "gray")

Each suggestion shows the synonym pair and the number of products it would affect.

Data Quality Flags

The scanner also identifies potential data issues:

• Products with missing images
• Products with empty descriptions
• Inconsistent tag formatting (e.g., "Blue", "blue", "BLUE")
• Variant options with very few values (which may not make good filters)

Applying Scan Suggestions

You do not have to apply suggestions one at a time. The results page provides bulk actions:

1. Review the suggestions -- Each recommendation has a checkbox
2. Select the ones you want -- Or click Select All Recommended
3. Click Apply Selected

For filters, applying a suggestion creates the filter with the recommended settings. For synonyms, it adds the synonym pair to your synonym list. You can always modify or remove applied suggestions later.

When to Re-run the Scanner

Re-run the scanner when:

• You have added a significant number of new products (100+ or a new product category)
• You have restructured your tags or metafields
• You notice shoppers searching for terms that return poor results
• You want to refresh your filter and synonym strategy after seasonal catalog changes

There is no limit on how often you can run the scanner, but each scan consumes AI credits from your plan allowance.

Connection to AI Setup Wizard

When you first install FilterIQ, the AI Setup Wizard runs a version of the Store Scanner as part of the onboarding flow. The wizard uses scan results to pre-configure your initial filters and synonyms.

The standalone Store Scanner (described on this page) is the full-featured version that you can run at any time. It provides more detailed results and more granular control over which suggestions to apply.

The AI Store Scanner is available on the Pro and Plus plans.

AI providers and data flow

The Store Scanner sends catalog metadata to one of our approved AI sub-processors — currently Anthropic (Claude API) or OpenAI (GPT API). Both are disclosed in our Sub-processor List on the Privacy Policy page, which is the authoritative live list generated from docs/governance/SUBPROCESSORS.yaml. Catalog metadata is minimized and redacted before submission; neither provider is authorized to use submitted merchant data to train public models.