What Is Alternative Data? – Definition and How It Is Used
Learn what alternative data means, which categories exist, how web data fits, and what makes a dataset usable for decisions.
Alternative data is information outside traditional reported sources — used to form a view earlier or more precisely than official figures allow. The term originated in investment research and now describes a broader practice.
1. Traditional Versus Alternative
Traditional sources are periodic, audited, and lagging: financial statements, official statistics, industry reports.
Alternative sources are continuous, unaudited, and current: web pages, satellite imagery, transaction panels, app usage, job postings, and reviews.
The trade is timeliness against reliability, and the whole discipline is managing that trade honestly.
2. The Categories
| Category | Examples |
|---|---|
| Web-sourced | Pricing, product ranges, job postings, reviews, hiring pages |
| Transactional | Aggregated card and receipt panels |
| Geospatial | Satellite imagery, footfall, shipping movements |
| Usage | App downloads, engagement estimates |
| Sentiment | Public social and forum content |
| Environmental | Weather, agricultural conditions |
3. Why Web Data Is the Largest Category
It is public, broad, and continuous. A company's own website reveals its pricing, product range, positioning, hiring, and expansion — updated by the company itself, in public, without any intermediary.
Job postings are the most consistently informative: they describe intentions months before products exist.
4. What Makes a Dataset Usable
- History. A single snapshot has no signal; change over time does.
- Consistency. The same method applied identically across periods.
- Coverage. Known and stable, so gaps are not mistaken for changes.
- Provenance. Where each observation came from and when.
- Documented method, so a consumer can judge what it can support.
A dataset that changes its collection method mid-series produces an apparent shift that never happened — the most common failure in this work.
5. The Standard Pitfalls
- Survivorship — sources that disappear leave a biased remainder.
- Coverage change mistaken for a real change.
- Backfill impossibility — web sources cannot be re-read, so gaps are permanent.
- Spurious correlation, abundant when many series are tested.
- Decay — a signal that works becomes crowded and stops working.
6. Compliance Is Part of the Product
Personal data carries privacy obligations regardless of public availability. Material non-public information carries securities-law obligations. Terms of service and copyright apply to the collection and to any redistribution.
Serious users document provenance and permissions per dataset, because the question arrives eventually and cannot be reconstructed later.
7. What Makes It Credible
Method documentation, stated coverage, honest gaps, stable definitions, and a clear account of what the data cannot show.
A dataset presented without those invites conclusions it cannot support — which is a failure of the publisher rather than the consumer.
8. Summary
Alternative data is non-traditional information used to see current conditions ahead of reported figures, with web-sourced data the largest category.
Usability comes from history, consistent method, known coverage, and provenance. Its pitfalls are coverage change, unrepeatable collection, spurious correlation, and signal decay — and its compliance obligations are part of the dataset rather than an afterthought.