Frequently Asked Questions

sImage is a Shopify app that uses AI (CLIP model) to analyze product images and show customers visually similar items. It converts images into 768-dimensional vectors stored in PostgreSQL with pgvector, enabling fast similarity search. This increases average order value by 15%+ on average.

sImage works in three steps: 1) When you install the app, it analyzes your product images using the CLIP AI model to create 768-dimensional embeddings, 2) These vectors are stored in PostgreSQL with pgvector extension, 3) When a customer views a product, sImage performs a cosine distance search to find visually similar items and displays them as recommendations.

sImage works best for visual-first stores: fashion and apparel (color/style matching), home decor and furniture (aesthetic discovery), art galleries (visual browsing), jewelry stores (pattern matching), and any store where customers shop by appearance rather than category.

sImage uses the CLIP (Contrastive Language-Image Pre-training) model via Replicate API. CLIP generates 768-dimensional image embeddings that capture visual features like colors, patterns, styles, and shapes. This enables semantic understanding of visual similarity.

CLIP (Contrastive Language-Image Pre-training) is a neural network trained by OpenAI on 400 million image-text pairs. It creates a shared embedding space for images and text, enabling visual similarity search. sImage uses CLIP to convert product images into 768-dimensional vectors for fast similarity matching.

Yes, sImage uses PostgreSQL with the pgvector extension to store and query 768-dimensional image embeddings. pgvector enables efficient cosine distance search for finding visually similar products, with sub-millisecond query times even with millions of products.

sImage uses 768-dimensional vectors with a cosine distance threshold of 0.5, which balances precision and recall. This captures visual similarity in colors, patterns, styles, and shapes. Customers who engage with recommendations are 70% more likely to purchase, and visual recommendations outperform text-based ones by 2-3x.

sImage has minimal impact on page speed. Similarity search takes under 100ms with pgvector, and the theme extension loads asynchronously. Recommendation caching further reduces load times. No performance issues have been reported across 500+ active stores.

sImage offers transparent pricing: Free tier (100 products), Small ($6/mo for 1,000 products), Medium ($12/mo for 2,000), Large ($24/mo for 4,000), Pro ($48/mo for 8,000), Champion ($96/mo for 16,000), and Enterprise ($192/mo for 32,000). All tiers include AI features, filter rules, manual overrides, and conversion analytics.

sImage takes 5 minutes to set up. Install from the Shopify App Store, let the AI analyze your products (priority sync shows results in minutes), then use the theme editor to place the recommendation block on product pages. No coding required.

Yes, sImage works with all modern Shopify themes including Dawn, Sense, Craft, Refresh, and hundreds of third-party themes. The theme extension reads your theme's CSS variables for colors, fonts, and border radius, so recommendations look native out of the box.

Yes, sImage offers multiple controls: Filter rules (by collection, tag, or price range), manual overrides (set specific recommendations per product with priority 1-10), and exclusion rules (remove products from recommendations). You're always in control of what appears.

Yes, sImage includes a Quick Add to Cart feature. Customers can add recommended products without leaving the product page. The button supports variants, shows adding state, and confirms successful additions. You can customize button text or disable it entirely.

Yes, sImage provides built-in conversion analytics. Track impressions, clicks, and add-to-cart events with a 30-minute attribution window. View daily aggregated metrics including recommendation impressions, click-through rate, conversion rate, and revenue generated from recommendations.

sImage excels for visual-first products (fashion, home decor, art, jewelry) because it understands colors, patterns, and styles rather than just keywords or categories. Traditional related products work better for clearly categorized items (electronics, tools). Choose sImage when product appearance matters more than taxonomy.

sImage shows visually similar items (same style/color) while Frequently Bought Together shows complementary items (accessories/add-ons). Use sImage when you want customers to discover alternatives or variations, and FBT when you want to increase order value with add-ons.