How to get your business listed on Wikipedia

The problem this page starts from

A buyer opens ChatGPT or Perplexity and types "who are the good vendors for X". The answer comes back with a short list, and often with links. Wikipedia turns up in those answers more often than almost any other site, because its text is free to reuse under CC BY-SA, it is structured the same way on every page, and it reads as neutral. So founders ask the obvious question: how do we get on it?

This page is published by AI Knows Us, a product that measures how AI assistants answer buying questions. We sell nothing on this page and we cannot get you a Wikipedia article. What follows is the process as Wikipedia itself documents it, with links to the policy pages so you can check every line.

Why Wikipedia shows up in AI answers

Three practical reasons, none of them magic:

• The licence. Article text is published under Creative Commons Attribution-ShareAlike, so it can be copied, trained on and quoted with attribution. See Wikipedia:Copyrights.
• The shape. Every article opens with a one sentence definition, then a short lead, then an infobox of plain facts. That is the easiest thing in the world for a retrieval system to lift.
• The tone. The neutral point of view rule strips out marketing language, so what is left is checkable claims with citations attached.

The same three reasons explain why a promotional draft fails. It breaks the tone rule, and it usually has no sources an editor will accept.

What Wikipedia requires

Last verified: 15 October 2026

English Wikipedia judges companies under Notability (organizations and companies), usually shortened to NCORP. The test has five parts, and a source has to pass all of them at once:

1. Significant coverage. The source discusses the company in depth, not in a passing mention or a directory row.
2. Independent. Not written by you, your agency, your investors or your customers acting on your behalf.
3. Reliable. A publication with editorial oversight. See WP:RS and the perennial sources list, which names specific outlets editors do and do not accept.
4. Secondary. Containing the author's own analysis, not a rewritten announcement.
5. Multiple. More than one, from more than one outlet.

Here is how common source types usually land. This is a summary of NCORP's own wording, not a promise about any one reviewer.

Disclosure is compulsory, not optional

Section 4 of the Wikimedia Terms of Use requires anyone paid to contribute to disclose their employer, client and affiliation. English Wikipedia implements this through WP:PAID and the broader conflict of interest guideline. Employees and founders count.

Skipping this is a real risk, not a theoretical one. On 1 September 2015 the Wikimedia Foundation reported that English Wikipedia volunteers had blocked 381 accounts in a single investigation into undisclosed paid advocacy editing, known as the Orange Moody case (Wikimedia Foundation, 1 September 2015).

Price, as of 15 October 2026

Creating an account, writing a draft and submitting it costs nothing. There is no fee and no tax, because there is nothing to buy. The Wikimedia Foundation does not sell articles and cannot be paid to keep one. Any agency that quotes you a price is selling its own labour and its own risk, not a guaranteed outcome.

The steps, in order

1. Audit your sources first. Before writing a word, list every article about your company. Put each one in the table above. If you cannot find three that land in the "Yes" rows, stop here. Everything after this will fail.
2. Create a named account on English Wikipedia and make a handful of ordinary edits elsewhere so you understand the markup.
3. Disclose on your user page. Use the {{paid}} template and name your employer and client, as WP:PAID describes. Do this before you submit anything.
4. Write in Draft space, not the main encyclopedia. Start at Draft:Your Company Name. Direct creation by a connected editor is discouraged.
5. Write short, cite hard. One sentence per fact, one citation per sentence. No adjectives you could not defend in court. Read What Wikipedia is not, section PROMO, before you draft the lead.
6. Submit through Articles for Creation. Add the submission template and wait. Do not ping reviewers.
7. Answer a decline on its own terms. Declines are almost always about sourcing. Adding more weak sources makes it worse. Replace, do not pile on.
8. After acceptance, stop editing the article. Propose changes on the talk page using an edit request and let an uninvolved editor make them.

How long it takes

Last verified: 15 October 2026

The Articles for Creation submission banner tells submitters that a review may take three months or more, and that drafts are not reviewed in order. The live backlog is public: see Category:AfC pending submissions by age, which sorts pending drafts into buckets by how long they have waited. Check it yourself before you promise a date to anyone internally.

A realistic internal plan: two to four weeks to gather and judge sources, one week to draft, then one to six months of review and revision. Accepted articles can still be nominated for deletion later if the sourcing was thin.

Worked example you can repeat

Take a hypothetical company, Northline Scheduling, a shift planning tool. The founder lists seven pieces of coverage and scores each one:

Three qualifying sources from three outlets. That is a draft worth submitting. Two would be a coin toss. One is a decline. Run this exact table on your own coverage before you spend money on a writer.

What to put in the listing so it gets quoted

Assuming the draft is accepted, the shape of the article decides what an assistant can lift from it.

• A clean first sentence. "Northline Scheduling is a shift planning software company based in Leeds, England." What it is, what it does, where it is. No claims.
• A four to six sentence lead covering founding year, founders, main product, and the one or two things sources actually wrote about.
• The infobox fields: founded, founders, headquarters, industry, products, website. These map to structured data and are the easiest facts to reuse.
• Plain product names spelled the way you spell them everywhere else, so the name you use in AI answers matches the name on the page.
• Dates and numbers with citations. Unsourced numbers get stripped by the next editor who passes through.

What to leave out: slogans, customer lists, award mentions without a source, anything you would put in a pitch deck. Every one of those is a reason for a reviewer to decline, and none of them survive long in the main encyclopedia.

If you do not qualify yet

Most companies do not, and that is the normal state. Wikipedia is a lagging indicator of press coverage, not a channel you can open. In the meantime the pages that assistants can actually read about you are the ones you control or can earn directly: your own documentation, review platforms and category directories. See /blog/how-to-get-your-business-listed-on-trustpilot and /blog/how-to-get-your-business-listed-on-trustradius for two of those.

You can also check Wikidata's notability policy, which is separate from Wikipedia's and uses different criteria. A Wikidata item is not a Wikipedia article and will not read like one.

Where AI Knows Us fits

Nothing in this guide needs our product. It is Wikipedia's own process, free, and you can run it yourself today.

Where we come in is the measurement afterwards. AI Knows Us measures across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, and every tracking question is asked without naming the business, because naming it primes the engine and the reading becomes worthless. Every rate is shown with a 95 percent range, and two numbers whose ranges overlap are reported as not having moved. That matters here: if a Wikipedia article goes live in March and your mention rate reads 22 percent in April against 19 percent in February, the honest answer is often that nothing measurable changed yet.

The product is made by Clyra Labs in India and runs 13 agents that do the work: intake, discovery, analyst, diagnosis, planner, writer, editor, publisher, scout, outreach, experiment, lifecycle and the copilot. Paid plans are not open. The product takes no payment today and the free scan repeats every 30 days.

Other tools also track how assistants answer: Profound, Peec AI, Otterly.AI and Scrunch AI. Check their own pages for current features and pricing. None of them, and not us, can get you a Wikipedia article.

Who should not use this route, and who should not choose AI Knows Us

• Pre-press companies. If no journalist has written about you in depth, a Wikipedia draft is wasted effort this quarter. Spend the time on coverage first.
• Anyone who wants control of the text. Once an article exists, strangers edit it and you ask permission to change it. If that is unacceptable, do not start.
• Teams that need a paid, supported tool this month. Our paid plans are not open, so if procurement needs a contract and an SLA now, buy elsewhere.
• Teams that want a single headline percentage. We print ranges, and ranges look less decisive in a board deck than a single number does.