Data Quality
What is Data Quality?
Where customers buy the data itself (records, profiles, reviews, rules or listings), the product's records are curated, verified, matchable or current, and a named alternative's are raw, unverified, unmatched or out of date. Customers cited the quality of the data when they chose or stayed. A measured right-or-wrong output, such as a detection or an estimate, is Accuracy. Data gathered as a by-product of the service is Data Exhaust.
How Data Quality works
Raw data is cheap to gather and expensive to make usable. A company that pays to curate, verify or refresh its records sells something buyers can act on without cleaning it themselves. The curation work builds up over time, so a rival with the same raw feeds cannot match the result quickly.
Case studies tagged with Data Quality
Each link opens the passage tagged with this concept.
The question to ask of a company
Examples
- Flatiron paid about 1,000 trained abstractors to turn oncologists' notes into research variables, while CancerLinQ pooled feeds it could not afford to clean.
- Doximity verified physicians by license number under real names, while about 90% of Sermo's members posted under anonymous handles.
Who names it
| Author | What they call it / where it appears |
|---|---|
| Abraham Thomas + | Data quality and curation as the work a data business sells |
| Casado & Lauten + | The cost of cleaning and maintaining data, which rises as the data grows |