Wikipedia Pageviews Scraper - By Language

What does Wikipedia Pageviews Scraper do?

How many people looked something up — and in which language.

Wikipedia pageviews are the broadest open measure of public attention there is: every article, every day, in every language edition. Wikimedia publishes them one article and one language at a time, as raw daily series.

No API key. No account.

---

Where attention actually is

Three car makers, 10 languages, 60 days:

BYD Auto — 44 language versions

Language	Views	Share	Trend	Peak
English	185,323	47.6%	−1.1%	×1.5
German	88,786	22.8%	−40.4%	×4.2
Japanese	26,793	6.9%	−40.7%	×2.8
French	24,781	6.4%	+34.6%	×4.9

English accounts for less than half. German readers are nearly a quarter of all attention on BYD — and that German interest spiked to 4.2× its normal level, then fell back 40%. Something happened in Germany. The English-only view shows none of it.

Rivian — Russian readership +116.8% over the same window. Tesla — 104 language versions, and a 108× spread between the first language and the tenth.

---

The four things this does that the source does not

1. It finds the article in each language.

You supply "Tesla, Inc."; the Japanese article is titled テスラ, the Arabic one تسلا موتورز. Titles are resolved automatically through Wikipedia's own language links. Without this, any international comparison requires knowing the exact title in every language you care about.

2. One row per language, with its share.

shareOfTotalPct is where a brand is known and where it is not. It is the column that turns a list of numbers into a market map.

3. Trend and spike, computed.

The source returns a series and no direction. trendPct compares blocks of seven days — Wikipedia readership drops at weekends, so splitting a period unevenly manufactures a trend that does not exist.

peakVsAverage dates events: a day at 4.9× normal is something that happened; a day at 1.2× is a Tuesday.

4. The number of language versions.

104 for Tesla, 44 for BYD, 25 for Rivian. On its own, that is a measure of global standing, and it comes free with the title resolution.

---

Field coverage

Measured on 30 rows across three subjects and ten languages: 100% on every field — views, share, daily average, peak, peak date, peak ratio, trend, language versions and article URL.

That is unusual, and it is because the source is a clean numeric series rather than a scraped document. What varies is not whether a field is filled, but whether a subject has an article in a given language at all — and when it does not, no row is produced for it.

---

What it is for

• Brand attention by market. Which countries are actually paying attention, ahead of a launch or after a campaign.
• Competitive tracking. Your brand against two rivals, weekly, with the language split.
• Event detection. A spike with a date, per language. Recalls, launches, controversies and earnings all show up here before they show up in structured data.
• Market entry research. A brand with 104 language versions and one with 25 are at different stages, whatever their revenue says.
• Alternative data. Attention on a company, a product or a person as a daily series, free and going back years.

Three dataset views ship with the Actor: Attention by language, Spikes and Measurement detail.

---

Limits

Stated plainly, because they affect what you can conclude.

• Attention is not sentiment. A spike says people looked something up. It does not say whether they liked what they found, and bad news drives more lookups than good news.
• Language is not country. Spanish readers are not only in Spain; English is read everywhere. Treat a language as an audience, not a market.
• Wikipedia readership is declining overall, as answers move into search results and assistants. Compare subjects against each other, not against their own figures from five years ago.
• Title resolution can land on the wrong article. "Stripe" is a payments company, a pattern and a film. Check articleTitle on the rows; supply a more specific name if it is wrong.
• Bots are partly filtered, not entirely. Wikimedia excludes known crawlers, but automated traffic still gets through, especially on small language editions where a few hundred views is the whole signal.
• Data lags by about two days, so the period always ends slightly before today.
• A missing row means no article, not zero interest.

---

How much does it cost?

You pay per language row returned: $0.003 each, that is $3.00 per 1,000. There is no start fee, and subscription plans pay less per language row.

The example input below asks for up to 500 language rows, so it costs $1.50 at most.

If a run reaches the spending limit you set, the output stops at that limit and never goes past it. You are never charged for rows that were not delivered.

Use Wikipedia Pageviews Scraper as an API

Call it from your own code with the Apify client, here in Python:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("DataIO/wikipedia-pageviews-by-language").call(run_input={
    'entities': ['Tesla, Inc.', 'Rivian', 'BYD Auto'],
    'languages': ['en', 'de', 'fr', 'ja', 'es'],
    'days': 60,
    'maxItems': 500,
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item)

It also works from JavaScript, Make, Zapier, n8n, and from AI agents through the Apify MCP server.

Other actors you might like

• Bluesky Followers Scraper: Export Audiences
• OpenAlex Scraper: Research Authors & Funders
• Google Trends Scraper & API: 100 Keywords

FAQ

Is it legal to use this data?

The actor reads public data from its official source, without logging in and without bypassing any access control. What you do with the data, for example contacting people listed in it, is your responsibility under the laws that apply to you, such as GDPR in Europe.

Can I run it on a schedule?

Yes. Create a schedule in Apify Console, daily or weekly for example, and each run delivers a fresh dataset, which you can send by email, webhook or integration.

Can AI agents use it?

Yes. It is available through the Apify MCP server, and every input field is described in its input schema, so an agent can call it directly.

Source

The Wikimedia REST API for pageviews, and the Wikipedia action API for title resolution and language links. Public, no authentication, licensed CC0. This Actor is not affiliated with the Wikimedia Foundation.