Wikipedia Pageview Spikes

Find the Wikipedia articles whose daily views suddenly jumped. For each day you choose, the Actor compares every article among the wiki's 1,000 most viewed pages with its own normal level (the median of its daily views over the previous days) and returns the spikes: article, link, date, views, normal level, spike ratio, rank and the day-by-day views behind the baseline. It works for any language edition of Wikipedia and for the other Wikimedia wikis (Wiktionary, Commons, Wikidata and more), using Wikimedia's official pageview statistics.

Use cases

• Trend and news monitoring: see which people, events and topics suddenly drew attention yesterday, in any language.
• Content and SEO planning: find rising topics early and plan articles, videos or posts around them.
• Research datasets: collect daily attention spikes over weeks for media, finance or social-science research.
• Alerting: schedule a daily run and send the new spikes to a spreadsheet, Slack or your own pipeline.

Quick Start

Open the Actor and click Start: the prefilled form returns yesterday's 10 biggest spikes on English Wikipedia. The same as input:

{"maxResults": 10}

German and French Wikipedia for the last 3 days, articles at least 5 times their normal views:

{
  "projects": ["de", "fr"],
  "lastNDays": 3,
  "minSpikeRatio": 5,
  "maxResults": 100
}

An empty input ({}) scans English Wikipedia for the latest published day and returns up to 20 spikes.

Input

Field	Type	Default	What it does
projects	array	["en.wikipedia"]	Wikis to scan: language codes (en, de, ja), projects (en.wikipedia, fr.wiktionary, commons.wikimedia, www.wikidata) or wiki addresses. Up to 10.
date	string	latest	Day to report (UTC), YYYY-MM-DD. Empty = the latest day with published statistics, usually yesterday.
lastNDays	integer	1	Report this many days, ending on date. Up to 31.
baselineDays	integer	7	Days before the reported day used for the normal level (3-30).
minSpikeRatio	number	3	Only articles with at least this many times their normal views.
minViews	integer	1000	Ignore articles with fewer views on the reported day.
access	string	all-access	all-access, desktop, mobile-web or mobile-app.
excludeSpecialPages	boolean	true	Leave out the main page, search, and talk, user, project, file, category and portal pages (in each wiki's own language).
exactBaseline	boolean	true	Read the exact daily views of the strongest spikes whose normal level is otherwise only known as an upper limit.
maxLookups	integer	10	How many articles get exact daily views per run (about 10 a minute).
maxResults	integer	20	Maximum spikes returned, strongest first. Up to 5,000.

What it returns

Every row has spikeKey, project, language, site, article, articleKey, url, date, views, rank, previousRank, baselineViews, baselineDays, baselineFrom, baselineTo, baselineSource, baselineIsUpperBound, baselineDailyViews (one {date, views} per baseline day), daysInTopList, spikeRatio, viewsChange, access, sourceUrl, source and scrapedAt.

Rows are sorted by spikeRatio, strongest first. Views count people, not bots. Dates are UTC days (YYYY-MM-DD), scrapedAt is ISO 8601. spikeKey (project:date:articleKey) is unique per article and day, so you can merge runs on it.

How the normal level is measured: Wikimedia publishes each wiki's 1,000 most viewed pages for every day. An article that was on all of the previous days' lists has an exact normal level from those lists (baselineSource: "topLists"). An article that was missing on some days had fewer views than that day's last listed page, so its normal level is at most that. For the strongest of these spikes the Actor reads the article's own daily views and makes the normal level exact (baselineSource: "dailyViews"). The rest keep the upper limit, marked with baselineIsUpperBound: true, and their real spike is at least the spikeRatio shown.

Exact and estimated numbers: views, rank and previousRank are always exact (Wikimedia's published figures). baselineViews, spikeRatio and viewsChange are exact when baselineIsUpperBound is false. When it is true, baselineViews is an upper limit, so spikeRatio and viewsChange are lower limits: the real spike is at least that big. A null in baselineDailyViews means the article was not in that day's top 1,000, or that day's list was unavailable.

Output example

{"spikeKey":"en.wikipedia:2026-10-26:Eric_Bauza","project":"en.wikipedia","language":"en","site":"wikipedia","article":"Eric Bauza","articleKey":"Eric_Bauza","url":"https://en.wikipedia.org/wiki/Eric_Bauza","date":"2026-10-26","views":511715,"rank":4,"previousRank":null,"baselineViews":1309,"baselineDays":7,"baselineFrom":"2026-10-19","baselineTo":"2026-10-25","baselineSource":"dailyViews","baselineIsUpperBound":false,"baselineDailyViews":[{"date":"2026-10-19","views":1366},{"date":"2026-10-20","views":1160},{"date":"2026-10-21","views":1309},{"date":"2026-10-22","views":1206},{"date":"2026-10-23","views":1164},{"date":"2026-10-24","views":1547},{"date":"2026-10-25","views":1406}],"daysInTopList":0,"spikeRatio":390.92,"viewsChange":510406,"access":"all-access","sourceUrl":"https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/en.wikipedia/all-access/user/Eric_Bauza/daily/20261019/20261026","source":"wikimedia.org/pageviews","scrapedAt":"2026-10-27T14:59:59+00:00"}

Limits and caveats

• Only pages that reached the wiki's daily top 1,000 can be found. On English Wikipedia that means roughly 8,000 views or more that day; on smaller wikis the bar is much lower.
• Wikimedia lets each user read only about 10 article histories a minute, so beyond the first few, exact normal levels take about 6 seconds per article. The default run reads up to 10 and usually finishes in well under a minute; set exactBaseline to false for instant results with upper-limit baselines, or raise maxLookups for more exact rows in a longer run.
• If the run's time is about to run out, it stops reading and delivers the spikes found so far.
• Statistics for a day appear a few hours after it ends (UTC). Now and then a day's top list is unavailable for a wiki. That day then counts as unknown: the normal level stays an upper limit (baselineIsUpperBound: true) while most of the window is known (up to 3 of 7 days missing); with more missing days only articles whose exact daily views were read are returned.
• A few pages get large automated traffic that Wikimedia counts as people (sudden spikes on obscure pages, often only on one platform). Compare access: desktop with mobile-web runs if a spike looks odd.

FAQ

How fresh is the data? Each run reads the latest published statistics, normally yesterday's (UTC).

How do I get only new spikes each day? Schedule a daily run with the default empty date; each run reports the newest day. Merge on spikeKey.

Can I use other wikis? Yes: any Wikipedia language (ja, pt, zh-yue...), or projects such as fr.wiktionary, commons.wikimedia and www.wikidata.

What happens when nothing spikes? The run succeeds with an empty dataset. Lower minSpikeRatio or minViews to see more.

Export formats? JSON, CSV, Excel, XML or RSS from the dataset, or read it through the Apify API in your own pipeline.

Price

$0.001 per result row, plus $0.005 per run start. A row is one article on one day: the same article spiking on two days, or on two wikis, is two rows. Rows with exact and with upper-limit normal levels cost the same. The run-start event is charged per GB of memory; the default 512 MB run is one start event. The start is charged even if the run saves no rows. Platform usage is included in these prices.

Worked cost: the default 20-spike run is $0.005 + 20 × $0.001 = $0.025. 1,000 spikes (for example lastNDays: 14, exactBaseline: false, maxResults: 1000) cost about $1.005.

Support

For questions or issues, contact us from the Actor page.