End-to-end examples comparing approaches to identity resolution with real-world datasets.
| Example | Records | Sources | What's Inside |
|---|---|---|---|
| Customer Identity Resolution | 6,539 | 5 | Same problem solved 3 ways: pure dbt/SQL, Splink, and Kanoniv |
Resolve 6,500 customer records across CRM, Billing, Support, App, and Partner systems. Three approaches compared side by side:
| Approach | What | Lines of Code | Runtime |
|---|---|---|---|
| dbt-sql/ | 5 hand-written SQL models in dbt | 350 | <1s (DuckDB) |
| splink/ | Splink + DuckDB (Python) | 440 | 2.6s |
| kanoniv/ | Declarative YAML spec + Rust engine | 170 | 0.4s |
All three use the same shared dataset of 10 CSV files.