Pageviews Trends for Wikipedia: absolute pageview counts and trend metrics

Independent tool. Not affiliated with, sponsored or endorsed by the Wikimedia Foundation or Google.

Get absolute daily or monthly pageview counts for any list of Wikipedia articles in any language edition. Every result includes trend metrics: growth %, week-over-week change, peak, average and median, and a 0–100 index like Google Trends uses. You can also pull the top 1–1,000 most-read articles for a day or month.

Data comes from the official Wikimedia Analytics Pageviews API. It is CC0, available from 2015-07-01, and needs no login, proxy or API key.

Compared with Google Trends

	Google Trends	This Actor
Numbers	Relative 0–100 only	Absolute view counts, plus a 0–100 index
Terms per comparison	Up to 5 (Google)	Any number of articles
Reproducibility	Based on a sample of searches (Google)	Published counts (from 2015-07-01)
Languages	Per country	300+ Wikipedia language editions (list), mixable in one run
Bot filtering	Not exposed	Choose user (bot-filtered), spider, automated or all-agents
Licence	Google ToS	CC0 1.0

Wikipedia views measure reading interest, not search volume. The two are related but not the same.

Use cases

• Brand and product interest: ChatGPT vs Claude vs Gemini, phones, SaaS tools.
• Alternative-data features: companies, cryptocurrencies, commodities.
• Content and SEO planning: rising topics, seasonal peaks, news spikes.
• Research: nowcasting and attention studies.
• Daily "what's trending" feeds (top mode).

Real run (2026-10-26, about 9:58 PM America/Chicago = 2026-10-27 02:58 UTC)

Input (demo/INPUT.json, the exact input that produced every sample below):

{
  "mode": "articles",
  "articles": [
    "ChatGPT",
    "Claude (AI)",
    "Google Gemini",
    "Grok (chatbot)",
    "Microsoft Copilot",
    "Perplexity AI",
    "DeepSeek",
    "https://de.wikipedia.org/wiki/ChatGPT",
    "This Article Does Not Exist Zq9"
  ],
  "project": "en.wikipedia.org",
  "granularity": "daily",
  "access": "all-access",
  "agent": "user",
  "includeSeries": true
}

No dates were given, so the Actor checked what Wikimedia had published (2026-10-26 UTC was not published yet) and used the 30 days ending 2026-10-25. Result: 8 ok rows and 1 free not_found row, from 11 API requests in 3.6 s of Actor time (per demo/articles-run.log).

Article	Project	Total views	Avg/day	Trend % (2nd half vs 1st)	Week-over-week %	Peak date	Index
ChatGPT	en.wikipedia.org	1,883,871	62,795.7	68.8	176.7	2026-10-21	100
Claude (AI)	en.wikipedia.org	171,393	5,713.1	22.3	50.9	2026-10-22	9.1
Google Gemini	en.wikipedia.org	82,990	2,766.3	35.9	102.5	2026-10-20	4.4
Grok (chatbot)	en.wikipedia.org	51,834	1,727.8	4.7	7.6	2026-10-22	2.8
Microsoft Copilot	en.wikipedia.org	36,832	1,227.7	5.2	12.8	2026-10-21	2
Perplexity AI	en.wikipedia.org	65,099	2,170	-1.2	17.4	2026-10-21	3.5
DeepSeek	en.wikipedia.org	91,656	3,055.2	32.3	27.3	2026-10-20	4.9
ChatGPT	de.wikipedia.org	229,724	7,657.5	0.3	18.8	2026-10-21	12.2

Output

One item per article. The SUMMARY record in the key-value store ranks them. This is the first item of the run above, verbatim except that series is trimmed to 2 entries:

{
  "mode": "articles",
  "input": "ChatGPT",
  "article": "ChatGPT",
  "articleTitle": "ChatGPT",
  "project": "en.wikipedia.org",
  "articleUrl": "https://en.wikipedia.org/wiki/ChatGPT",
  "granularity": "daily",
  "access": "all-access",
  "agent": "user",
  "startDate": "2026-09-26",
  "endDate": "2026-10-25",
  "dataAvailableThrough": "2026-10-25",
  "status": "ok",
  "dataPoints": 30,
  "totalViews": 1883871,
  "averageViews": 62795.7,
  "medianViews": 50069.5,
  "peakViews": 163328,
  "peakDate": "2026-10-21",
  "minViews": 28298,
  "minDate": "2026-10-11",
  "firstValue": 40694,
  "lastValue": 71140,
  "firstHalfAverage": 46724.9,
  "secondHalfAverage": 78866.5,
  "trendChangePct": 68.8,
  "last7Average": 121163,
  "previous7Average": 43794,
  "weekOverWeekPct": 176.7,
  "zeroDays": 0,
  "series": [
    {
      "date": "2026-09-26",
      "views": 40694,
      "index": 24.9
    },
    {
      "date": "2026-09-27",
      "views": 41296,
      "index": 25.3
    },
    "..."
  ],
  "relativeIndex": 100
}

• relativeIndex is 0–100 relative to the article in this run with the highest average. series[].index is relative to the highest single period in the run.
• dataAvailableThrough is the latest day (or month) Wikimedia had published when the run started. The Actor never goes past it, so unpublished days never show up as fake zeros. Later endDates are clamped, with a warning in the log and endDateClamped: true in SUMMARY.
• zeroDays counts genuine zero-view days inside the published range (the API omits days with no views).

A free row (never charged):

{
  "mode": "articles",
  "input": "This Article Does Not Exist Zq9",
  "article": "This Article Does Not Exist Zq9",
  "articleTitle": "This_Article_Does_Not_Exist_Zq9",
  "project": "en.wikipedia.org",
  "articleUrl": "https://en.wikipedia.org/wiki/This_Article_Does_Not_Exist_Zq9",
  "granularity": "daily",
  "access": "all-access",
  "agent": "user",
  "startDate": "2026-09-26",
  "endDate": "2026-10-25",
  "dataAvailableThrough": "2026-10-25",
  "status": "not_found",
  "error": "No pageviews recorded for this exact title in this date range. Check spelling/capitalisation; if the title is a redirect, open the article and use the title shown in the address bar."
}

Monthly granularity always ends at the last complete published month. A part-month is never mixed in. Example from the same evening (demo/INPUT-monthly.json: Bitcoin and Ethereum from 2025-10-01; September 2026 was the last complete month), Bitcoin row with series trimmed:

{
  "mode": "articles",
  "input": "Bitcoin",
  "article": "Bitcoin",
  "articleTitle": "Bitcoin",
  "project": "en.wikipedia.org",
  "articleUrl": "https://en.wikipedia.org/wiki/Bitcoin",
  "granularity": "monthly",
  "access": "all-access",
  "agent": "user",
  "startDate": "2025-10-01",
  "endDate": "2026-09-30",
  "dataAvailableThrough": "2026-09",
  "status": "ok",
  "dataPoints": 12,
  "totalViews": 1401557,
  "averageViews": 116796.4,
  "medianViews": 107535.5,
  "peakViews": 188739,
  "peakDate": "2026-02-01",
  "minViews": 76252,
  "minDate": "2026-07-01",
  "firstValue": 158676,
  "lastValue": 100154,
  "firstHalfAverage": 140051,
  "secondHalfAverage": 93541.8,
  "trendChangePct": -33.2,
  "last7Average": null,
  "previous7Average": null,
  "weekOverWeekPct": null,
  "zeroDays": 0,
  "series": [
    {
      "date": "2025-10-01",
      "views": 158676,
      "index": 84.1
    },
    "...",
    {
      "date": "2026-09-01",
      "views": 100154,
      "index": 53.1
    }
  ],
  "relativeIndex": 100
}

Top mode (demo/INPUT-top.json: {"mode":"top","topLimit":25}, with the date defaulting to the latest published day). This is the first row, verbatim:

{
  "mode": "top",
  "status": "ok",
  "rank": 1,
  "originalRank": 4,
  "article": "2026 Brazilian general election",
  "articleTitle": "2026_Brazilian_general_election",
  "views": 271783,
  "project": "en.wikipedia.org",
  "period": "2026-10-25",
  "granularity": "daily",
  "access": "all-access",
  "agent": "user",
  "articleUrl": "https://en.wikipedia.org/wiki/2026_Brazilian_general_election",
  "dataAvailableThrough": "2026-10-25"
}

Top-mode rows: rank (after special-page filtering), originalRank (Wikimedia's rank), article, articleTitle, views, project, period, granularity, access, agent (always user), articleUrl, status, dataAvailableThrough. If you ask for a date that isn't published yet, you get a single free no_data row.

Input

Field	Default	Notes
mode	articles	articles (series + metrics per article) or top (ranking for one day or month)
articles	["ChatGPT"]	Current titles (Taylor Swift) or URLs (https://de.wikipedia.org/wiki/ChatGPT). A URL's language edition overrides project.
project	en.wikipedia.org	Any Wikimedia project domain
granularity	daily	monthly = whole, complete months
startDate / endDate	30 days (daily) or 12 complete months (monthly), ending at the latest published day or month	In top mode, endDate is the day or month to rank
access	all-access	desktop, mobile-web, mobile-app
agent	user	user = traffic not classified as bots (self-identified spiders and heuristically detected automated traffic are excluded; some undetected bot traffic may remain)
includeSeries	true	false = summary-only rows
topLimit	100	1–1000 (top mode)
excludeSpecialPages	true	Drops the Main Page and common special/namespace pages (best coverage on English Wikipedia)
contactInfo	–	Optional. Your email or URL, added to the User-Agent next to the Actor's own contact

Pricing (pay per event)

• article-timeseries: one charge per article with status: ok, whatever the date range.
• top-article-row: one charge per ranked row.
• not_found, error, no_data and skipped rows are free.

Live prices are on the Pricing tab. The proposed launch prices are estimates (see PRICING.md). When your Max total charge is reached, the Actor stops fetching, and the remaining inputs come back as free skipped rows.

Tips and limits

• Pageviews count the exact title. Redirects are not followed, so use the current title (the one in the address bar after any redirect). Redirect titles have their own, usually tiny, counts.
• Data is available from 2015-07-01. Each UTC day is usually published within 1–2 days, and each month a few days after it ends. The Actor checks what has been published (on the run's project) and never returns unpublished days as zeros (dataAvailableThrough shows the cut-off).
• Requests are sequential (at most ~150 per minute) and back off on 429/5xx, honouring Retry-After up to 120 s. A run with 1,000 articles takes about 7–9 minutes.
• Memory: every series is held in memory to compute the 0–100 index. When articles × periods goes above 500,000 (e.g. 1,000 articles × 11 years daily), the run logs a warning. Use monthly granularity, a shorter range, fewer articles per run, or 1 GB+ memory.

Legal and data source

• Source: Wikimedia Analytics Pageviews API only. The host wikimedia.org/api/rest_v1/ is not restricted by robots.txt. No Wikipedia pages and no /w/api.php (disallowed in robots.txt) are requested.
• Licence: "All Analytics datasets are available under the Creative Commons Zero (CC0) public domain dedication, unless otherwise specified" (Wikimedia dumps legal page). The Analytics API says the same: "Data provided by the API is available under the CC0 1.0 license" (access policy). Output holds only article titles, URLs and counts. No article text (CC BY-SA 4.0) is copied. Suggested credit: "Pageview data: Wikimedia Foundation, CC0."
• User-Agent, per the Wikimedia User-Agent policy: PageviewsTrendsBot/0.1.0 (+https://apify.com/capable_santolina/pageviews-trends-for-wikipedia; pounderswork@mail.grokbot.com) apify-sdk/3 node/<version>. Pacing stays inside the API rate limits.
• No personal data: pageviews are aggregate counts per article. Nothing behind a login is accessed.
• Wikipedia® is a registered trademark of the Wikimedia Foundation, Inc. "Google Trends" is a trademark of Google LLC. This Actor is an independent tool that uses the public Wikimedia Pageviews API, and it is not affiliated with, sponsored or endorsed by the Wikimedia Foundation or Google. No Wikimedia logos are used.
• Contact / takedown: pounderswork@mail.grokbot.com.