Tag-Gen

AI-generated product tags for every category

Turn product images, product feed fields and API-provided product information into rich, structured metadata that powers search, filters, merchandising, SEO and recommendations.

Generate tags from product images, feed fields and API data
Structured attributes across colour, material, style and more
Consistent metadata across thousands of SKUs at scale
Supports fashion, furniture, homewares, craft, outdoor and more
Export or connect via API to your existing catalogue workflow
Okkular · Tag-Gen
Grey Scandinavian sofa
Analysing image...
Men's Linen Shirt
Missing Fit Missing Fabric Missing Colour Missing Occasion Missing Sleeve
Men's Slim-Fit Linen Shirt
Slim Fit Linen Navy Formal Full Sleeve Button-Down Collar Solid Regular Length Office Wear Party Ready

Why metadata matters

Great product discovery starts with great product data

Customers search in detailed, natural ways. They look for style, colour, material, fit, function, occasion, sleeve length, collar type and many other attributes. If those details are missing from your product data, your products are invisible to shoppers actively looking for them.

Tag-Gen generates rich, structured metadata at scale from product images, existing feed fields, titles, descriptions, specifications and API-provided data — so every product can be found.

Filter-ready attributes — automatically
Tag-Gen extracts the exact attributes your filters need — fit, fabric, colour, occasion, length and more — directly from product images and feed data.
From #12 to #2 — same product, richer data
Adding structured metadata to an existing listing improves how your search engine scores and surfaces it — without changing a single word of your product description.
Consistent across every product, every category
Manual tagging produces inconsistent results across teams and suppliers. Tag-Gen applies the same structured taxonomy to every SKU, at any scale.
How it works

From product data to structured tags

Tag-Gen analyses product images and feed data together — extracting visual attributes, normalising text-based fields and mapping everything to your taxonomy through a purpose-built retail AI workflow.

  • 1

    Product inputs

    Images, feed fields & API-provided product data

  • 2

    Visual & text analysis

    AI extracts attributes from images and normalises text fields from your feed

  • 3

    Tag generation & mapping

    Structured tags generated and mapped to your taxonomy and attribute configuration

  • 4

    Review & delivery

    Tags reviewed and approved, then delivered to your platform or export format

Not just tagged. Structured, normalised and mapped to your taxonomy.

What Tag-Gen creates

Generate attributes that make products discoverable

Tag-Gen extracts thousands of structured data points across the attributes shoppers actually search for — building a comprehensive taxonomy of visual features, functional details and category-specific terms.

Product type
Maxi dress Dining chair +450 more
Colour & colour family
Navy blue Warm neutrals +120 more
Pattern & print
Floral Geometric +80 more
Material & texture
Linen Velvet +200 more
Style & design details
Minimalist Embellished +300 more
Shape & silhouette
A-line Round back +150 more
Usage & occasion
Evening wear Party +50 more
Search synonyms
Joggers/sweatpants +1000s more

Attributes generated from images. Enriched further from your product feed and API data.

Product image Visual
Crochet stuffed bunny toy
stuffed-animal animals soft medium
Product feed Optional
Title Handmade Crochet Bunny
Category Toys > Stuffed Animals
Material 100% cotton yarn
Dimensions 28 cm tall
Safety CE certified, non-toxic
Enriched tag set Ready to publish
product-type: stuffed-animal theme: animals character: animal features: soft size: medium material: cotton yarn age: 0-2 years height: 28cm
5 attributes from image 3 fields from feed
Generated (visual)
Extracted (feed / API)

Input sources

Use the product data you already have

Okkular does not rely on images alone. Retailers can provide product information through existing feeds or APIs — including titles, descriptions, supplier copy, specifications, category fields, materials, dimensions, care instructions and other available attributes.

Okkular analyses text fields alongside images to extract, normalise and enrich product metadata. This is especially useful when product images do not show every important feature, or when technical specifications are already available in the product feed.

Generated from product images
Visual attributes are extracted directly from product images — product type, theme, character, features, size, colour and more. Works across fashion, furniture, toys, craft and any visually rich category.
Extracted from product feed & API data
Non-visual attributes — materials, dimensions, age suitability, safety certifications, care instructions, category fields — are extracted automatically from your feed or API data and merged into the final tag set.
Tag-Gen across segments

The right tags for every product, in every category.

Every retail segment has its own taxonomy. Tag-Gen is configured per segment to extract exactly what matters — so your metadata always matches how shoppers in that category actually think and search.

Fashion & Apparel
Fashion & Apparel
Fit, fabric, occasion and styling cues

Extract fit, fabric, silhouette, colour, occasion and styling cues — exactly how fashion shoppers search.

Shape Fit Occasion Neckline Collar-Style
Footwear
Footwear
Construction, comfort and sole technology

Extract construction, comfort, sole technology and style versatility — for everything from sneakers to dress shoes.

Closure-Detail Heel-Height Sole Footwear-Print
Accessories
Accessories
Material, style, functional features and functionality

Extract material, features, functionality, dimensions and styling cues — helping shoppers visualise how it completes a look.

Material Style Features Dimensions
Furniture
Furniture
Material, finish, dimensions and style

Extract material, finish, size, room fit, visual style and functional benefits — in the language furniture shoppers use.

Material Finish Style Shape Product-Type
Home Decor
Home Decor
Aesthetic, material, versatility and how it fits a space

Extract aesthetic, material, functionality, durability, dimensions and how a piece complements a space.

Aesthetic Material Durability Dimensions
Outdoor, Sports & Leisure
Outdoor, Sports & Leisure
Application, comfort and features

Extract application, comfort, portability, features, suitability, materials and activity relevance — for every sport and outdoor category.

Fit Top-Length Neckline Print
Toys & Kids' Products
Toys & Kids' Products
Age suitability, safety and play value

Extract age suitability, developmental benefits and play value — creating safe, engaging data sets for parents.

Age Character Theme Size Features
Electronics & Lifestyle
Electronics & Lifestyle Products
Specs, compatibility and key features

Extract spec-accurate features, compatibility and key technical benefits — clear enough for both browsers and buyers.

Specs Compatibility Features
Segments Fashion Womenswear
Segment
Fashion Home Decor Furniture Electronics Books + more
Categories — Womenswear
Blouse Alias Original: Womens-Blouse · 1 Synonym: Tops
Womens-Dresses 16 Attributes
Womens-Jeans 09 Attributes
Womens-Jumpers 10 Attributes
Blouse · Attributes & tags 14 attributes
Fit
Slim-FitAlias Original: Relaxed
Fitted
Oversized
Sleeves
Long-SleeveAlias Original: Full
Half
+3
Print PatternAlias Original: Pattern
GinghamAlias Original: Checkered · Synonyms: Print-Gingham, Print-Checkered
Polkadot
Camo
+14
Enable default synonyms from Okkular
Alias marks a renamed attribute or tag — the original name and any synonyms show beneath it
How synonyms improve onsite search
A product tagged "Checkered" is also found when a shopper searches for "Print-Gingham" or "Print-Checkered" — because Tag-Gen enriches each tag with its configured synonyms.
Shopper searches: "Print-Gingham" Product found: Checkered

Scale & consistency

Configure once. Tag every product to your standard.

Tag-Gen uses your attribute configuration to generate consistent, structured tags across every product in your catalogue. Define which attributes matter for each category, set your preferred names and values, add synonyms — and Tag-Gen applies that taxonomy to every SKU.

Configuration works at every level: segments, product ranges, categories and individual attributes. Changes apply automatically to all products in that scope — no manual rework required.

Your naming, applied everywhere
Rename any attribute or tag value to match your brand language. Use aliases so Tag-Gen generates tags in the exact terms your team uses — without changing the underlying taxonomy.
Synonyms that close the search gap
Shoppers don't always use your brand's terminology. Configure synonyms for each tag so products are found however a shopper describes them — "Gingham" finds "Checkered", "Long-Sleeve" finds "Full". Better onsite search, without changing your taxonomy.
Consistent across every segment and category
Every product gets tagged to the same standard, regardless of which segment it belongs to. The same configuration logic applies across your entire catalogue — no inconsistencies between teams, suppliers or product lines.
Set once, run continuously
Configure your taxonomy once and Tag-Gen runs it on every new product automatically. No per-product setup, no manual reformatting — just consistent metadata at catalogue scale.
Okkular · Tag-Gen · Image Alt Text — Wall Décor
Botanical floral painting in gold frame
Generated alt text
Alt text Generated

A framed botanical print featuring pink garden roses, purple iris flowers and yellow wildflowers with green foliage, displayed in a gold ornate frame against a white brick wall.

SEO benefit
Search engines index alt text. Descriptive, accurate alt text makes product images discoverable in image search — contributing to organic traffic without extra effort.
Accessibility benefit
Screen readers use alt text to describe product images to visually impaired shoppers. Natural language alt text means every customer gets an accurate, useful description — not a filename.

Image alt text

Descriptive alt text — generated automatically for every product image

Okkular generates natural language alt text for every product using generated metadata — describing what is visually in the image in natural language phrasing. No manual writing. No filenames. No missing descriptors/ keywords.

Generated from the product image — describes what is actually shown
Natural language phrasing — not a list of tags or a filename
Covers colour, shape, material, style and other visible details — whatever the category
Improves image indexing for SEO and organic search visibility
Exported alongside tags — no additional workflow required
Accessibility for every shopper
Visually impaired shoppers rely on screen readers to understand product images. Okkular's generated alt text provides a natural language description of each image — so screen readers can accurately describe what a product looks like, making your catalogue accessible for everyone.
Guns n Roses Fitted Longline Tee
womens-tees · SKU 8171028611119
Partially verified
collar-style: round fit: fitted neckline: v-neck hem: curved print: graphic top-length: regular sleeve: short
hem: curved
Back
Tag is correct
Mark as wrong
Belongs to multiple products
Add tag band-tee × longline × graphic-print ×

Human-in-the-loop

AI speed with human control

Generated tags can be reviewed, corrected and approved before they are exported or published. Your team stays in control — Okkular handles the volume, your editors handle the standards.

Review predicted tags for every product before export
Flag a tag as wrong — it is excluded from the download automatically
Add custom tags to any product directly in the review interface
Re-analyse individual products or regenerate specific tags without re-running the full batch
Verify and approve — individually or export directly once satisfied
How retailers use Tag-Gen

Generated, structured and delivered into your workflow

Rich product attributes power every surface where customers discover and evaluate products — and Tag-Gen delivers them directly into the platforms and workflows your team already uses.

Search & filters
Richer product attributes improve onsite search and category filter quality — so shoppers can narrow to exactly what they want.
Merchandising
Create collections and campaigns based on consistent metadata — by colour family, style, occasion or any structured attribute.
SEO & long-tail discovery
Capture how customers actually describe products — unlocking long-tail search traffic that generic metadata misses.
Recommendations
Power visual search and similar-product recommendations using structured attributes that capture visual and functional similarity.
Catalogue operations
Reduce manual tagging effort and improve consistency across teams, suppliers and product lines — without per-product setup.
Platform integration
Connect to ecommerce platforms, PIM systems and marketplace workflows — via API, export or direct integration.
Platform & delivery

Designed to fit your catalogue workflow

Tag-Gen integrates with ecommerce platforms, PIM systems and marketplace workflows. Product data can be exported or connected via API depending on your operating model.

Ready-made integrations
Connect directly to Shopify, SAP and other major ecommerce and PIM platforms with no custom development required.
Structured exports
Export enriched tag sets as CSV or JSON — formatted for your PIM, marketplace or internal tools.
Custom API
Send product data in, receive structured tag sets out — at catalogue scale, integrated directly into your existing tech stack.
Review & approve workflow
Human review is built into every workflow. Tags are reviewed and approved before they publish — individually or in bulk.
Shopify Integration
Shopify
Apparel21
Talkoot

Generated, structured and delivered into your workflow. No manual reformatting required.

Want richer product tags without manual effort?

See how Tag-Gen can enrich your product catalogue with structured, consistent metadata — generated from images, feed data and your existing product information.