Reddit GrowthMonitorBlog
Reddit GrowthMonitorBlog

How to Get a Wikipedia Page for Your Company: Rules, Process and What AI Engines Do With It

Oct 16, 2026 · 16 min read

Wikipedia's own rules for company articles, read from its policy pages, plus what OpenAI, Meta, and the Wikimedia Foundation say about Wikipedia's role in AI. Written for brand, comms, and SEO leads who have been told to "get on Wikipedia" and need to know whether they can.

How to Get a Wikipedia Page for Your Company: Rules, Process and What AI Engines Do With It
byIsaac Tarrab

Table of contents

  1. Key Takeaways
  2. Why a Wikipedia Page Matters for Your Company in AI Search
  3. Does Your Company Qualify? Wikipedia's Notability Guidelines
  4. What Counts as a Reliable Source on Wikipedia
  5. Conflict of Interest: What You Can and Cannot Do
  6. The Process: Draft, Submit and Review
  7. Wikidata: The Lower-Barrier Entity Record
  8. When Not to Try, and What to Do Instead
  9. Keeping the Page Accurate After It Goes Live
  10. FAQ

How to get a Wikipedia page for your company starts with a test you cannot buy your way through: multiple independent, reliable publications must already have covered the company in depth. If that coverage exists, a connected writer must disclose the relationship and submit the draft through Wikipedia’s Articles for Creation review. If it does not, no draft, agency, or budget will make the page stick.

A Wikipedia page for business is not a marketing channel, and Wikipedia says so itself. Its FAQ for article subjects answers "Should I pay someone to write an article about me or my organization?" with one word: "No."

The question reaches marketing teams now because of AI search. OpenAI's GPT-3 paper and Meta's LLaMA paper both list Wikipedia in their training data. The Wikimedia Foundation says search engines and social media platforms "prioritize its information to respond to questions from their users." Every rule below was read on Wikipedia's and Wikidata's own policy pages on 15 October 2026.

Key Takeaways

  • Notability is earned elsewhere: Wikipedia presumes a company notable only after "significant coverage in multiple reliable secondary sources that are independent of the subject." Funding news, launches, and "top 100" lists do not count.
  • Your own team counts as paid: anyone compensated for publicity work on the company is a paid editor under Wikipedia's disclosure policy, whether or not they were hired to edit Wikipedia.
  • You propose, volunteers decide: connected editors are "strongly discouraged" from editing their company's article directly. New pages go through Articles for Creation, and changes go to the talk page.
  • The practical point: the page is the last step, not the first. If your coverage fails the four-part source test, a draft is wasted effort. Build the coverage, and the article becomes possible.
  • Wikidata has a lower bar, not no bar: an item without a Wikipedia article usually needs "serious and publicly available references," and statements you add about yourself need a verifiable source.

Why a Wikipedia Page Matters for Your Company in AI Search

A Wikipedia page for business matters in AI search for two documented reasons: model builders have trained on Wikipedia, and AI companies access its content at scale. Neither makes Wikipedia a ranking factor. No provider read for this guide says a Wikipedia article moves a brand up an answer; the record shows only that Wikipedia is frequently used as source material.

How an engine uses those facts is a separate question, covered in how models learn what your brand is.

What the Training Papers Disclose

OpenAI's GPT-3 paper is specific. Brown et al.'s GPT-3 paper, submitted in May 2020, lists Wikipedia at 3 billion tokens and 3% of the training mix. The authors say datasets "we view as higher-quality are sampled more frequently," and Wikipedia was seen 3.4 times over the full training run.

Meta's LLaMA paper went a step further. In Meta's LLaMA pretraining paper, submitted in February 2023, Wikipedia dumps in 20 languages made up 4.5% of the pretraining data, read 2.45 times. Meta also trained a model to "classify pages used as references in Wikipedia" and discarded web pages it did not classify that way.

That second detail is the useful one. In that pipeline, resembling a page Wikipedia cites was the quality filter for the CommonCrawl web data, which made up 67% of the mix.

What Wikimedia Says About AI Reuse

StatementSourceDate
"Almost all large language models (LLMs) train on Wikipedia datasets"Wikimedia's user trends post on its Diff blog17 Oct 2025
Human pageviews fell "roughly 8%" against the same months of 2024, which Wikimedia attributes to "generative AI and social media"Same post17 Oct 2025
Generative AI should credit the human contributions it uses, and "most AI developers" should access Wikipedia through Wikimedia EnterpriseWikimedia's AI era statement10 Nov 2025
Amazon, Meta, Microsoft, Mistral AI, and Perplexity joined Wikimedia Enterprise, alongside Google and othersWikimedia Enterprise's partner announcement15 Jan 2026

What This Does Not Prove

The two papers describe 2020 and 2023 models, not the systems behind today's assistants. A training-mix share is not a citation rate. An Enterprise partnership shows that a company pulls Wikipedia content at scale, and says nothing about how its engine weighs that content against other sources.

Wikipedia is one source among many in an answer. The case for why off-site sources beat on-site content covers the reviews, threads, and publications that sit beside it.

Does Your Company Qualify? Wikipedia's Notability Guidelines

The Wikipedia notability guidelines for businesses live on one page, the organizations and companies guideline, known as WP:NCORP. Its core rule: a company "is presumed notable if it has been the subject of significant coverage in multiple reliable secondary sources that are independent of the subject."

Size buys nothing. The guideline states that "No company or organization is considered inherently notable." It lists revenue, raised capital, headcount, and company age among the "arbitrary statistics" that do not make coverage significant.

The Four Tests Every Source Must Pass

NCORP judges each source on its own, and "An individual source must meet all of these criteria":

  • Significant: the source addresses the company "directly and in depth." The guideline says a 400-word article in The Village Voice "is a lot more significant than a single-sentence mention in The New York Times."
  • Independent: the author has no stake in the company, and the content carries "original and independent opinion, analysis, investigation, and fact checking."
  • Reliable: the publisher has editorial oversight and a reputation for checking facts.
  • Secondary: an author's own analysis, a step removed from events. Press releases, executive interviews, annual reports, and customer testimonials are primary sources, and primary sources "cannot be used to establish notability."

Then there must be several of them. NCORP says "multiple" "is not a set number." A 1960s Bangladeshi women's rights group "might establish notability with just one or two quality sources." A tech start-up in a major U.S. metropolitan area gets no such allowance, in the guideline's own comparison.

Coverage That Does Not Count

NCORP lists routine coverage that never meets the bar, including stories about:

  • "a capital transaction, such as raised capital"
  • "a product or a product line launch, sale, change, or discontinuance"
  • "the hiring, promotion, or departure of personnel"

It also excludes quotes from your staff inside other people's stories, and directory or database listings. Inclusion in "best of", "top 100", "fastest growing" or similar lists does not count either. Audience counts too: "At least one regional, statewide, provincial, national, or international source is necessary."

A Five-Question Self-Check

This checklist is this guide's own shorthand for NCORP, not a Wikipedia tool. Run it against every source you plan to cite.

  1. Is the company the subject? The article discusses the company in depth, rather than mentioning it inside a story about something else.
  2. Was it written at arm's length? No financial tie between the author and the company, and no press release underneath it.
  3. Does it go past the news hook? Analysis of what the company does, beyond funding, launches, hires, or rankings.
  4. Did an editor stand behind it? A publication with editorial oversight, with at least one source carrying regional or national reach.
  5. Is it a different publisher? NCORP treats a series by one journalist, or several stories in one paper, as a single source.

If most sources fail any one of these questions, the draft is premature. Build the coverage first.

What Counts as a Reliable Source on Wikipedia

Wikipedia reliable sources are defined in the reliable sources guideline: "Articles should be based on reliable, independent, published sources with a reputation for fact-checking and accuracy." Its summary adds the test that applies in any dispute: editors "need to be able to explain why they relied on that source."

For companies, two tests meet in one list: reliability and independence. NCORP classifies most marketing-produced material as dependent coverage. What counts is earned media coverage from independent outlets.

Source typeCounts toward notability?What the guideline says
In-depth feature by a staff reporter at a regional or national outletYes, if independentSignificant coverage goes "well beyond brief mentions and routine announcements" (NCORP)
Independent analysis or study that discusses the companyYes, if significant, reliable and independentA secondary source holds "an author's own thinking based on primary sources" (NCORP)
Press release, or a story rewritten from oneNoExcludes "any material that is substantially based on such press releases even if published by independent sources" (NCORP)
Funding or launch announcementNoListed as routine coverage (NCORP)
Contributor post on Forbes, Inc.com, Entrepreneur.com, or TechCrunchNoExcludes pieces by non-staff "contributors" where there is no "meaningful editorial oversight" (NCORP)
Sponsored or paid articleNoExcludes "any paid or sponsored articles, posts, and other publications" (NCORP)
Trade magazine coverageRarely"there is a presumption against the use of coverage in trade magazines" (NCORP)
Your website, blog, or social postsNoSelf-published sources "are generally not accepted as evidence of notability" (NCORP)
Review platforms, forums, and community threadsNo"Websites whose content is largely user-generated are generally unacceptable as sources" (RS)

Your own pages still have a narrow job. Once notability is established, NCORP lets editors cite primary and self-published sources "with appropriate care" to verify some facts. Your site can confirm a founding date, but it cannot make the case that the article should exist.

Conflict of Interest: What You Can and Cannot Do

The Wikipedia conflict of interest guideline, the WP:COI guideline, opens with a two-sentence summary: "Do not edit Wikipedia in your own interests, or in the interests of your external relationships. Paid editing must be disclosed." Its working rule is plain: COI editors "are strongly discouraged from editing affected articles directly, and can propose changes on article talk pages instead."

Who Counts as a Paid Editor

Paid editing is regulated by disclosure. Wikipedia's paid-contribution policy summarizes itself in one line: "If you were paid or expected to be paid in any way for contributing to Wikipedia, you must disclose it." The Wikimedia Terms of Use, in effect since 7 June 2023, require you to "disclose each and any employer, client, intended beneficiary and affiliation" for any compensated contribution.

The scope is wider than most teams assume. Anyone "compensated for any publicity efforts related to the subject" of their edits is deemed a paid editor, whether or not the job included Wikipedia. Interns count as employees, and "Non-disclosure agreements do not invalidate this requirement."

What Connected Editors Can and Cannot Do

You canYou should not
Disclose the connection on your user page and the article's talk pageEdit the company article directly, beyond narrow fixes
Propose changes on the talk page with the {{edit COI}} templateRemove negative material because it is unflattering
Submit a new article through Articles for CreationAct as a reviewer of your own draft at Articles for Creation
Fix minor "hard fact" errors such as a wrong date, citing a reliable sourcePaste copy from your own website into the article

A request is only a request. The guideline states that "Requested edits are subject to the same standards as any other, and editors may decline to act on them."

The Process: Draft, Submit and Review

Articles for Creation lets you "write an article as a draft and submit it for review by an experienced volunteer before publication." Connected editors have no shortcut. If you have a conflict of interest or are paid, the page says you "should submit your draft through Articles for creation." The steps below are how to get a Wikipedia page for your company inside those rules.

Build the Source File First

  1. Search for an existing article or draft, including former company names. Improving an existing page beats starting a second one.
  2. List every qualifying source in a sheet with four columns: significant, independent, reliable, secondary. NCORP advises that when "the suitability of a source is in doubt," you exclude it.
  3. Disclose before you write. Name your employer and client on your user page or the draft's talk page.
  4. Create the draft. Use Wikipedia’s Article Wizard, or write it in your sandbox. A sandbox draft can be submitted directly to Articles for Creation; it does not have to be moved to a Draft: page first.

Write and Submit the Draft

  1. Write in your own words, from the sources. Wikipedia "prohibits using large language models (LLMs), such as ChatGPT, to generate or rewrite article content," with limited exceptions for basic corrections and translations.
  2. Cite inline, so each claim points at the source that supports it. The draft does not need images, an infobox, or perfect formatting before review.
  3. Submit the draft. It then displays the notice "Review waiting, please be patient," and you can keep improving it while it waits.

Read the Review Outcome

OutcomeWhat Articles for Creation says it meansNext step
Accepted"accepted and published as a Wikipedia article"Keep to the conflict-of-interest rules from here on
Declined"not ready to become an article yet, but the problems may be fixable"Fix what the reviewer named, then resubmit
Rejected"fundamentally unsuitable for Wikipedia, with little prospect that further work will make the draft acceptable"Ask the reviewer to reconsider if you have "significant new evidence" or believe the decision was mistaken

Set expectations from the page itself. It says "There is no fixed waiting time," and drafts "are not necessarily reviewed in the order they were submitted." The notice on a submitted draft gives a rolling estimate, not a promise. Acceptance is not permanent either, because "An unsuitable article may be returned to draftspace or nominated for deletion."

Wikidata: The Lower-Barrier Entity Record

Wikidata has two stated goals: "to centralize interlanguage links across Wikimedia projects and to serve as a general knowledge base for the world at large." A Wikidata entry for company facts is a structured item, with a label, a description, and statements such as official website or headquarters, each backed by references.

Wikidata's Own Inclusion Criteria

Wikidata's notability policy accepts an item that meets at least one of three criteria:

  • A sitelink: the item links to a page on a Wikimedia project such as Wikipedia. Draft and user pages do not count.
  • A described entity: "a clearly identifiable conceptual or material entity that can be described using serious and publicly available references."
  • A structural need: the item is "needed to make statements made in other items more useful."

The bar is lower, and deliberately loose. The policy says its guidelines "are intentionally left a bit vague," that disagreements go to a request for deletion, and that a discussion about reforming the policy is ongoing.

Editing an Item You Represent

Wikidata's autobiography page, written for people and those who represent them, allows editing the item about yourself. It says not to create one unless you are sure it meets the notability criteria, and that statements "must be sourced to a reliable, verifiable source." Your own website may be cited, but only for information about yourself.

Paid work carries its own rule. Wikidata's paid-editing policy states that "A user has to disclose the name of every organization (regardless of legal status) and individual that pays the user money for contributing to Wikidata."

Properties That Carry Company Facts

PropertyWikidata's descriptionPractical note
official website (P856)"URL of the official page of an item (current or former)"After a domain change, Wikidata says to "add an additional statement with preferred rank. Do not remove the former URL."
industry (P452)"specific industry of company or organization"The value is another Wikidata item; the property page's own example gives Red Hat the value "software industry."
inception (P571)"time when an entity begins to exist"Use the date your references support, since a mismatch with other sources is easy to spot.
headquarters location (P159)"city or town where an organization's headquarters is or has been situated"The value is a city; Wikidata says to use a qualifier for the specific building.
Legal Entity Identifier (P1278)"identifier for a legally distinct entity per ISO 17442"Add it only if the company already has an LEI, using its Global LEI Index record as the reference. Other registry identifiers have their own Wikidata properties.

When Not to Try, and What to Do Instead

If the self-check fails, do not draft. A thin draft is the slowest way to learn the answer: Wikipedia's FAQ warns that "If you write an article about yourself, the chances are pretty good that it will be deleted."

Build the Coverage the Test Requires

  • Editorial coverage: in-depth stories by staff reporters are the kind NCORP counts, which is where earned media coverage earns its place.
  • Stories with original reporting: original data or a named expert gives a journalist a reason to write about the company rather than rewrite a release. The mechanics sit with digital PR.
  • Analyst and research mentions: independent analysis counts as secondary coverage when it discusses the company in depth.
  • Review sites and communities: user-generated sites are "generally unacceptable as sources" for Wikipedia. They still shape what AI engines say about you while the editorial record grows.

Check What the Engines Already Say

Before deciding, record which sources the engines use for your category's buyer questions: run an AI visibility audit by hand to see whether Wikipedia appears for those prompts at all, and which publications appear beside it.

Treat Offers to Publish a Page as a Warning Sign

Articles for Creation carries a standing scam warning: "If anyone asks you for money or any kind of payment to help you publish, protect, restore, or undelete a Wikipedia article or draft page, it is a scam." Wikipedia's FAQ adds that such articles tend to "end up deleted within a few weeks." Disclosed paid editing exists under the rules above, but no one outside the volunteer review can promise that a page will be published or stay up.

Keeping the Page Accurate After It Goes Live

Nobody owns a Wikipedia article, including its subject. Wikipedia's FAQ states that editors "will not give you exclusive editorial control over the article." The COI guideline adds that other editors "may also decide to delete the article or decide to keep it should you later request deletion."

Correcting Errors the Right Way

  • Small factual errors: after disclosing your connection, Wikipedia's FAQ allows you to correct "minor 'hard fact' errors, such as an incorrect date," citing a reliable source.
  • Everything else: make an edit request on the talk page, "preferably through the Wikipedia:Edit Request Wizard."
  • Unflattering but sourced material: leave it. The FAQ lists "remove negative material" among the things PR staff must never do, with an exception for material that breaks the policy on living people.
  • Wikidata statements: do not remove sourced statements you think are wrong. Wikidata asks you to raise them at Project chat.

Watching the Page and the Answers

Every article's history page is public. Check it on a schedule and after any news about the company, then rerun the same AI prompts and compare. A wrong founding date on Wikipedia and a wrong date in an answer may share a source, and the fix belongs upstream.

Decide in advance which changes justify a response, then monitor what AI engines say about your brand against those thresholds. An accurate but negative article is an online reputation management problem, and editing the article will not solve it.

Wikipedia will not take a company's word as evidence of notability, and the independent record it asks for is the part of this process Klarivo works on. Klarivo does not draft, edit, or submit Wikipedia or Wikidata entries, and nothing in this guide suggests otherwise.

Klarivo is a managed AEO and Reddit service that builds the publication, review, and community footprint AI systems draw on when buyers ask who to buy. Only part of that footprint counts toward notability: in-depth, independent publication coverage does, while reviews and community threads shape AI answers without counting for Wikipedia.

Klarivo Monitor tracks five engines independently: ChatGPT, Claude, Perplexity, Gemini, and Grok. It reports top citation sources as the domains the engines drew on. That list shows whether wikipedia.org appears among the cited domains for your category’s buyer prompts. If a page later goes live, the same prompt set gives you a before-and-after record.

To see which domains the engines draw on for your category, start with Klarivo Monitor. To talk through the coverage gap behind those answers, book a Klarivo discovery call.

FAQ

Can I pay someone to write my Wikipedia page?

Wikipedia's FAQ answers "No." Disclosed paid editing is allowed under the Terms of Use, but the editor must name you as the client, and "There is no confidentiality for the employer, client, or affiliations." The draft still goes through volunteer review, so no paid editor can promise acceptance.

How long does review take?

There is no fixed time. The notice on a submitted draft shows a rolling estimate; on 15 October 2026 it read "This may take 5 weeks or more, since drafts are reviewed in no specific order," with 2,562 submissions pending. You can keep improving the draft by editing it normally while it waits. One clock does exist: drafts not edited by a person for six months may be deleted under criterion G13. The page describes that as housekeeping rather than a verdict, and such drafts can normally be restored on request.

Does Wikipedia affect ChatGPT answers?

OpenAI's GPT-3 paper lists Wikipedia in that model's training mix. No OpenAI page read for this guide gives a comparable breakdown for the models behind ChatGPT today, or says how a Wikipedia article is weighted in an answer. Whether ChatGPT cites a Wikipedia page for your prompts is something you can observe by running them and recording the cited domains.

Do I need Wikidata first?

No order is required. An item with a Wikipedia sitelink meets Wikidata's first criterion automatically. Without an article, an item has to stand on "serious and publicly available references," which some companies can meet before they pass NCORP.

Start accelerating your Reddit presence

See how Klarivo can shift your visibility across AI, search and buyer communities.

Book a Demo

Table of contents

  1. Key Takeaways
  2. Why a Wikipedia Page Matters for Your Company in AI Search
  3. Does Your Company Qualify? Wikipedia's Notability Guidelines
  4. What Counts as a Reliable Source on Wikipedia
  5. Conflict of Interest: What You Can and Cannot Do
  6. The Process: Draft, Submit and Review
  7. Wikidata: The Lower-Barrier Entity Record
  8. When Not to Try, and What to Do Instead
  9. Keeping the Page Accurate After It Goes Live
  10. FAQ

Related articles

G2 vs Capterra: Which Review Site Should B2B Software Brands Invest In, and What AI Engines See
AI Search Visibility

G2 vs Capterra: Which Review Site Should B2B Software Brands Invest In, and What AI Engines See

G2 now owns Capterra, Software Advice, and GetApp. Compare vendor pricing, review rules, and the review sites AI engines cite before you set a review budget.

by Valentina Abazi· Oct 14, 2026· 12 min read
Bing SEO: How to Rank on Bing and Feed ChatGPT and Copilot With Your Content
AI Search Visibility

Bing SEO: How to Rank on Bing and Feed ChatGPT and Copilot With Your Content

Bing's index feeds Copilot answers, and Microsoft now reports those citations. How to set up Bing Webmaster Tools, submit a sitemap, use IndexNow, and fix indexing.

by Isaac Tarrab· Oct 6, 2026· 12 min read
Online Reputation Management in the Age of AI Search: A B2B Guide
AI Search Visibility

Online Reputation Management in the Age of AI Search: A B2B Guide

When a buyer asks an AI engine whether a vendor can be trusted, the answer comes from review sites and forums the vendor does not run. What changes in online reputation management when the first reader is a model, and what stays the same.

by Isaac Tarrab· Sep 22, 2026· 11 min read
How it worksReddit GrowthKlarivo MonitorBlogFAQsLinkedIn
Privacy PolicyTerms of ServiceCookie policy

© 2026 Klarivo. All rights reserved.