Thin, inconsistent product copy costs rankings and conversions
Manually writing copy for thousands of SKUs is impossible, so most stores ship supplier boilerplate — duplicate across the web, thin on detail, and invisible to search. Conversion and organic traffic both suffer.
This solution turns structured product data into unique, persuasive, search-optimised copy at catalog scale, with a human approving before anything goes live.
Ingest feed → draft copy → SEO & QA → sync to store
- ✓ SEO-ready titles & meta
- ✓ Benefit-led descriptions
- ✓ Scannable bullet points
- ✓ Multi-locale translation
- ✓ Tone & length controls
- ✓ Bulk catalog runs
- ✓ Claude for copy & translation
- ✓ PIM / feed connectors
- ✓ Keyword data integration
- ✓ Duplicate-content detection
- ✓ Human approval queue
- ✓ No unsupported claims
- ✓ Spec-grounded generation
- ✓ Brand voice enforcement
- ✓ Compliance term filters
- ✓ Full edit audit trail
Spec-grounded copy generation
Copy is generated from the product's real attributes so it never invents features — and every field is length-checked before it can sync.
from anthropic import Anthropic client = Anthropic() def write_copy(product: dict, voice: str) -> dict: # Ground strictly in real attributes — no invented features prompt = f"""Write ecommerce copy from these attributes ONLY: {product['attributes']} Brand voice: {voice}. Return JSON: {{"title": str(<=60), "description": str, "bullets": [str]}}""" msg = client.messages.create( model="claude-sonnet-4-20250514", max_tokens=800, temperature=0.4, messages=[{"role": "user", "content": prompt}], ) copy = parse_json(msg.content[0].text) assert len(copy["title"]) <= 60 # enforce store limits return copy
Give every SKU copy that ranks and sells
Deploy standalone, or pair it with Catalog Enrichment for clean data in, great copy out.