Compare Wikipedia topic interest, explore autocomplete ideas and open a matching Google Trends comparison. Use the results to build a research shortlist, with source observations clearly separated from heuristic scores. This tool does not report Google keyword search volume.
Wikipedia charts use complete UTC months only. Suggestion relevance is not search volume; competition and opportunity are heuristic scores.
Keyword Demand Explorer
LiveOptional: choose the matching encyclopedia article before analysis.
Since your last check
Save a snapshot and this tool remembers it. Come back next week or next quarter and see exactly what moved — attention, device mix, market spread. No account, stored on your device only.
Research summary
A rule-based summary of the available observations. Treat its recommendations as prompts for further research.
Wikipedia language comparison
Compare available linked Wikipedia editions by article views. Language is not geography, and differences in article scope or coverage affect the comparison.
Related topics worth checking
Keyword suggestions
Explore autocomplete suggestions across the available search verticals. Start with Quick expansion, then filter the results. Missing relevance stays unavailable.
Content brief for an AI assistant
Create an optional research brief from the available observations and model outputs. Review its assumptions and missing values before using it to draft content.
Google Trends
Google Trends. The 0–100 scale is relative to this term's own peak, not a count of searches. Interest by region shows where demand concentrates geographically — a different signal from the language-market map above, and worth reading together.
How to research a topic
- Choose comparable topics. Enter up to five topics with the same level of specificity. Inspect the Wikipedia article matched to each phrase before interpreting the figures.
- Set the language and history window. The language controls Wikipedia data. The region controls Google Trends and suggestion settings; it does not turn Wikipedia views into country-level data.
- Review the reading history. Look at monthly averages, peaks and the displayed comparison window. A news spike is different from sustained interest.
- Explore phrasings. Start with Quick suggestions. Increase expansion depth when useful, then apply include, exclude and word-count filters.
- Validate before publishing. Review the actual search results and audience intent. Use CTR & Traffic Forecast to model explicit impression assumptions, rather than treating article views as search impressions.
Match, verify and shortlist
Use Choose matching articles to inspect up to five Wikipedia matches per topic. Your choice applies to that topic and language for this page session; it is not included in snapshots or shareable links. After choosing, press Analyse demand again. Without a choice, the first automatic match is used.
Each result now shows returned versus requested months and a separate year-over-year comparison of the last three complete months with the same months one year earlier. Missing comparison months or a zero prior-year total produce an unavailable percentage rather than an estimate.
Tick suggestions to build a shortlist. Copy selected and CSV selected export only those keywords. Selection is temporary and separate from the existing Copy all and full CSV controls.
What each result means
| Result | What it represents | Important limit |
|---|---|---|
| Monthly article views | Pageviews for the matched Wikipedia article in the selected language and window. | Visits, not unique readers or Google search volume. |
| Trend percentage | The average of the last three available months compared with the first three, within the selected history. | Not automatically year-over-year growth or a seasonally adjusted trend. |
| Device split | The selected article’s desktop, mobile-web and mobile-app views over three complete months. | Describes that Wikipedia audience, not your website visitors. |
| Language comparison | Views of linked article editions in the supported Wikipedia languages. | Language editions are not countries; coverage and article scope can differ. |
| Suggestion relevance | A normalised provider value, when a numeric relevance field is supplied. | Missing values stay unavailable. Suggestions do not provide search volume. |
| Competition / opportunity | Custom heuristic scores for organising a suggestion list. | Not measured keyword difficulty or proof that a topic is easy to rank for. |
Complete months and trend calculations
Wikipedia requests end on the last day of the previous UTC month. A 12-month run in October 2026 requests October 2025 through September 2026, avoiding an incomplete October. Device and language comparisons request the previous three complete months. Publication delays or missing records can still reduce available coverage.
Trend % = (last-three-month average / first-three-month average − 1) × 100
With fewer than six available months, the trend uses equally sized end groups of up to three months. A zero starting average does not support a percentage growth calculation. Check the chart dates before comparing separate exports.
An article receives 120,000 views over 12 returned months: its monthly average is 10,000. If the first three months average 8,000 and the last three average 10,000, the change is (10,000 ÷ 8,000 − 1) × 100 = 25%. This describes article views in that window; it does not mean Google searches increased by 25%.
How suggestion scores work
Autocomplete responses can contain predictions without numeric relevance. The tool keeps those suggestions but leaves relevance and opportunity unavailable. It does not substitute search volume or invent a score from a suggestion’s position.
Opportunity heuristic = round(Relevance × (100 − Competition heuristic) / 100)
The competition heuristic starts with a word-count weight: 95 for one word, 82 for two, 66 for three, 50 for four, 37 for five, 27 for six, 20 for seven and 14 for eight or more. It adds 11 for a recognised commercial modifier, subtracts 9 for a recognised question opener and subtracts 12 for a recognised local modifier. The result is restricted to 5–100. These are design choices, not a trained or validated difficulty model; modifier rules are primarily English-oriented.
A three-word question such as “how mortgages work” starts at 66 and receives the question adjustment of −9, producing 57. If its supplied relevance normalises to 80, its opportunity estimate is round(80 × 43 ÷ 100) = 34. Without supplied relevance, opportunity is unavailable.
Coverage and language-priority indicators
The article-view-to-coverage ratio divides monthly article views by Wikipedia’s matching-article count. Its bands use thresholds of 3, 12 and 50. That denominator concerns Wikipedia coverage, not the number or quality of competing web pages.
Absolute = clamp((log10(monthly edition views) − 2.5) × 20, 0, 60)
Relative = clamp(edition views / selected-language views × 20, 0, 40)
Language-priority bands at 20, 35 and 55 are planning heuristics. They do not establish translation profitability or the difficulty of reaching that audience. The research summary and optional content brief inherit these limits.
Turn research into a content plan
- Develop angles: use the Keyword Idea Generator to extend a seed topic, then review whether similar phrasings belong on one useful page.
- Check link-strength context: the Domain Rating Checker compares a different metric. DR does not convert to keyword difficulty or a ranking guarantee.
- Draft presentation: use the Meta Tag Generator, Word Counter and Reading Time Calculator to review page presentation. There is no required SEO word count.
- Review implementation: the JSON-LD Validator and Robots.txt Generator address separate technical tasks, not demand measurement.
- Model outcomes explicitly: use the Traffic Growth Calculator for growth scenarios and Ad Revenue Calculator for revenue assumptions. Keep these forecasts separate from the observations here.
Exports, snapshots and sharing
Copy data exports the topic comparison. Suggestion results have separate copy and CSV controls, with source information. If clipboard access fails, a visible text box provides a manual-copy route.
Snapshots and session settings are stored in your browser. A saved snapshot is a past observation, not an automatically refreshed database. Re-run the same article, language and window before interpreting a change. The shareable URL includes topic settings; share it only when you are comfortable revealing those topics. The generated embed is a static representation rather than a live feed.
Lookups send queries or article requests to the external services listed below. Local storage is separate from those requests and from general site analytics. See our Privacy Policy for the wider site practices.
Common mistakes to avoid
- Reading 10,000 article views as 10,000 searches for your exact phrase.
- Assuming an English-language edition represents the United States, or an Arabic edition represents Saudi Arabia.
- Treating a short news spike as a durable content opportunity.
- Reading unavailable relevance as a zero-demand keyword.
- Using a heuristic score as a substitute for reviewing search intent and competing pages.
- Comparing snapshots with different matched articles, languages or history windows.
Frequently asked questions
Does this show Google monthly search volume?
No. It combines Wikipedia article views, autocomplete suggestions and a Google Trends comparison route. Each measures a different signal. Use appropriate search-volume data separately.
Why did my phrase match a different Wikipedia article?
Topic lookup selects an article from Wikipedia’s search response. Inspect its title and link. If it represents a different concept, use Choose matching articles, select an appropriate title and run the analysis again. Refine your phrase if none matches.
Why are relevance and opportunity unavailable?
The autocomplete response did not provide usable numeric relevance. The suggestions remain visible; the dependent opportunity estimate is not calculated. Missing data is not converted to zero.
Is the competition score real keyword difficulty?
No. It is a query-shape heuristic based on word count and recognised modifiers. It does not inspect who ranks or measure the backlink strength of the search results.
Why is the current month excluded?
A partial calendar month is not comparable with a complete month. Wikipedia windows therefore stop at the end of the previous UTC month, subject to available data.
Does the language comparison show country demand?
No. It compares Wikipedia language editions. A language can be read in many countries, and each edition may cover a topic differently.
Why does Google Trends differ from Wikipedia views?
Google Trends concerns relative search interest, while Wikipedia data concerns visits to encyclopedia articles. Different audiences, topics, periods and normalisation can produce different patterns.
What should I do if an embedded Trends chart fails?
Use Open comparison in Google Trends. The direct link carries the chosen topics, region and period; an embedded chart is optional and can be refused by Google.
Can I research Arabic topics?
Yes, select the Arabic Wikipedia edition and enter a suitable topic. Be cautious with suggestion scoring: the query-modifier and intent rules are primarily English-oriented.
Are topics and snapshots saved?
Session settings and snapshots are saved locally in your browser. Running analysis still sends requests to external providers. Use the session’s start-fresh option to clear the saved session; browser site-data settings can remove all local records.
Can I use the generated content brief without checking it?
No. Review the matched article, source observations, missing values and heuristic assumptions. Fact-check any resulting draft and add evidence relevant to the actual audience.
Does a high score guarantee traffic or rankings?
No. Use the result as a research aid. Search intent, content usefulness and many other factors affect real outcomes.
How is year-over-year change calculated?
The last three complete UTC months are compared with the same three months one year earlier: (current total ÷ prior-year total − 1) × 100. All six monthly records are required and the prior-year total must be positive. This comparison is separate from the first-versus-last trend in the chosen history window.
Do missing months count as zero views?
No. The coverage line lists absent records. Missing months are not inserted as zero, and they prevent a year-over-year percentage when they are needed for that comparison.
Can I export only selected suggestions?
Yes. Tick the keyword rows you want, then use Copy selected or CSV selected. The existing full-list exports remain available. Selection is temporary and is not added to the saved session.
Will my chosen article be included in a shared link?
No. Article overrides apply to the current page session and topic/language combination. Tell the recipient the chosen article or include it in the exported report; the share link alone may use the automatic match.
Sources and methodology
- MediaWiki OpenSearch API — article-title matching and candidate results.
- Wikimedia Analytics API — pageview data and access documentation.
- Wikimedia — page views — what a pageview represents.
- Google Trends data FAQ — normalisation and differences from Autocomplete.
- Google Search Central — helpful content — people-first content guidance, rather than a scoring formula.
Numeros is an independent publisher. These links do not imply endorsement or review by the source organisations. Read our calculator methodology and Disclaimer for general scope.
Related calculators
Continue your research with these tools, or browse the Business calculators category.