First-party copy intelligence
VSL copywriting datasets built from real transcript research
Study how direct-response copy is structured without digging through raw scripts. These datasets turn processed VSL and ad transcripts into de-identified hooks, mechanisms, promises, proof moves, CTAs and position-aware patterns.
Corpus data last changed . Page views never alter this date.
30,373
high-confidence derived patterns
100
fully processed source transcripts
6
niches above the publication gate
Browse datasets by niche
A niche becomes public only after at least five transcripts, five distinct products and 1,000 high-confidence derived patterns pass the processing gate.
Weight Loss VSL Copywriting Dataset
13,577 patterns from 40 processed transcripts and 30 distinct products.
Memory VSL Copywriting Dataset
6,045 patterns from 22 processed transcripts and 17 distinct products.
Nerve VSL Copywriting Dataset
3,358 patterns from 11 processed transcripts and 9 distinct products.
Diabetes VSL Copywriting Dataset
3,350 patterns from 11 processed transcripts and 11 distinct products.
Joint Pain VSL Copywriting Dataset
2,118 patterns from 8 processed transcripts and 6 distinct products.
Erectile Dysfunction VSL Copywriting Dataset
1,925 patterns from 8 processed transcripts and 8 distinct products.
Commercial master edition
Use the full derived corpus in your research workflow
The self-serve edition includes CSV.GZ, JSONL.GZ and a data dictionary. It excludes raw transcripts, verbatim excerpts, product names and internal source identifiers. The current immutable release contains 28,194 records.
De-identified by design
No raw transcript, filename, storage path, product name or external source reference appears in the public sample or self-serve export.
One consistent schema
Compare the same functional fields across niches and source formats instead of reconciling separate spreadsheets.
Stable research artifacts
Releases are immutable and checksummed. New processing creates a new version; it never silently rewrites a file you bought.