Wikipedia Pageviews
Daily pageview counts for tracked Wikipedia articles. A crude but genuinely useful proxy for public attention to a company, product or topic, and one of the few attention series that is free, stable and openly licensed.
Collection
- Expected frequency
- 1 day
- Stale after
- 3 days
- Last successful run
- 2026-09-14 21:53:58 UTC
- Latest source period
- 2026-09-12 00:00:00 UTC
- Earliest coverage
- 2026-06-14 00:00:00 UTC
- Latest coverage
- 2026-09-12 00:00:00 UTC
- Rows
- 182
- Partitions
- 2
- Owner
- —
Point in time
What a historical query can reconstruct
- Level
- published
- On revision
- supersede
- Effective field
- view_date
- Published field
- — (source does not date its releases)
- Primary key
- project, article, access, agent, view_date
Quality
| Check | Result | Severity | Detail |
|---|---|---|---|
| row count anomaly | pass | warn | only 2 prior run(s); need 5 to judge |
| schema drift | pass | warn | observed columns match the manifest |
| freshness | pass | warn | last update 69.9h ago, within the 72.0h SLA |
| required non null | pass | fail | all 5 required column(s) populated |
| numeric range | pass | warn | 1 numeric column(s) within range |
| enum domain | pass | fail | 2 enum column(s) within domain |
| primary key unique | pass | fail | 91 row(s), all keys distinct |
Known limitations
What this dataset does not cover
- Only the articles in the tracked list are collected; it is not a full Wikipedia crawl.
- Counts for the last day or two are provisional and are commonly restated.
- An article rename breaks continuity: the new title is a different key, by design, because silently merging them would fabricate a history the source never reported.
- Attention is not sentiment. A spike says people looked, not why.