50,000 Annotated E-commerce Product Descriptions

A high-quality training dataset for fine-tuning or evaluating product description generation models.

**Contents:**

- 50,000 real e-commerce product descriptions (electronics, clothing, home goods)

- Each description annotated with:

- Quality score (1–5, human-labelled)

- Estimated conversion rate category (low/medium/high)

- Detected persuasion techniques (scarcity, social proof, etc.)

- Readability score (Flesch-Kincaid)

- Category and subcategory labels

**Format:** JSONL + CSV variants included

**Use cases:**

- Fine-tuning product description generators

- Training quality classifiers

- Benchmarking LLM product copy output

- E-commerce conversion rate research

**Licence:** Commercial use permitted. No resale of the raw dataset.