Wikipedia article problems — what can actually change.

A Wikipedia article about you or your company is different from every other reputation problem, for one structural reason: there is no company to appeal to. Wikipedia is written and governed by volunteer editors operating under written policies they collectively enforce. The Wikimedia Foundation hosts the encyclopedia; it does not edit it on request, and it will not overrule the community because a subject is unhappy. There is no support queue, no service level, and no commercial relationship to lean on.

What there is instead is a policy framework that is unusually explicit about what belongs in an article and what does not. Businesses that learn the framework get results. Businesses that treat Wikipedia as a customer-service problem get reverted, and frequently make their situation permanently worse, because everything they do is logged in public forever.

The policies that decide everything

Four policies do almost all the work. Every viable request maps onto at least one of them.

Verifiability. Wikipedia content must be attributable to reliable published sources. The threshold is not whether something is true — it is whether a reliable source says it. This cuts both ways: a claim you know to be wrong can be removed if its own citation does not support it, and a claim you know to be misleading stays if a reliable source states it plainly.

Notability. Subjects merit a standalone article when they have received significant coverage in reliable sources that are independent of the subject. Press releases, your own website, sponsored content, and routine business announcements do not establish it. The guideline for companies and organisations is applied more strictly than the general one, which matters if the question is whether an article should exist at all.

Biographies of living persons. This is the strongest lever available to an individual subject. Contentious material about a living person that is unsourced or poorly sourced may be removed promptly and without waiting for consensus, and the policy applies not just to articles but to talk pages, drafts, and project pages. It exists precisely because the encyclopedia takes real-world harm to living people seriously.

Conflict of interest. Editors with a personal or financial stake in a subject are strongly discouraged from editing that subject's article directly. Separately, the Wikimedia Terms of Use require anyone receiving compensation for contributions to disclose their employer, client, and affiliation. These two rules together define the only route that works for a subject or their representatives.

The correct route: declared conflict of interest, talk page request

If you are the subject of an article, or you work for the subject, or you are being paid by the subject, you should not edit the article. The route that works instead is this:

1. Create a named account and state your connection openly on your user page. If the work is paid, disclose the client and employer as the Terms of Use require.
2. Go to the article's talk page and post an edit request using the conflict-of-interest request template, so it enters the queue that uninvolved editors monitor.
3. State the change precisely. Quote the sentence as it currently reads, propose the exact replacement wording, and cite the reliable source that supports it. Requests written as complaints get ignored; requests written as ready-to-apply edits get applied.
4. Wait. An uninvolved editor will accept, amend, or decline it. That editor's judgment is the decision.

This feels slow and powerless to executives used to escalation paths. It is also the only approach that reliably holds, and the reason is structural: an edit made by an uninvolved editor after review survives, while an edit made by an interested party is reverted on sight and leaves a permanent record that makes every subsequent request harder to get accepted.

Why undisclosed editing is the worst available option

Editing the article about yourself while pretending to be a disinterested member of the public fails in a specific and expensive way. Editing patterns are public. Account histories are public. Wikipedia has a dedicated noticeboard where suspected conflicts are investigated openly, and it has a long institutional memory for this behaviour. Undisclosed paid editing breaches the Terms of Use and is treated as a serious matter, with consequences that include blocked accounts and, in documented cases, the deletion of the articles produced.

The reputational side is worse than the procedural side. A discovered attempt to manipulate an article frequently generates press coverage of its own, and that coverage is a reliable source — which means it can end up cited in the very article you were trying to improve. It is difficult to think of a reputation tactic with a worse risk profile.

The same applies to sockpuppetry: operating multiple accounts to manufacture the appearance of independent support for a change. It is prohibited, it is routinely detected through technical and behavioural analysis, and discovery invalidates every argument the accounts made.

There is also a scam ecosystem here worth naming. Operators approach subjects offering to create, "protect", or clean up articles for a fee, sometimes returning later to demand payment to prevent a page being damaged. Wikimedia has publicly warned about these schemes. Anyone guaranteeing an outcome on Wikipedia — an article created, a paragraph removed, a page protected — is describing something no participant in that community can promise.

What genuinely gets fixed

The successful requests are unglamorous and narrow. In rough order of reliability:

• Claims their own citations do not support. A date that contradicts the cited source, a job title lifted from a press release the source never carried, a figure attributed to a publication that says something different. These are correctable because the evidence settles them and no judgment is required.
• Poorly sourced contentious material about a living person. Under the living-persons policy this can be removed quickly, and it is the fastest-moving category on the encyclopedia when it genuinely applies.
• Outdated facts with a better source. Leadership changes, corporate structure, product status — straightforward where you bring the reliable citation with you.
• Undue weight. Where a minor episode occupies a disproportionate share of an article relative to its coverage in sources, proportionality is a legitimate argument under the neutral-point-of-view policy. It is harder than a factual correction because it requires editors to agree on balance, but it is a real route.
• Genuine privacy issues. Non-public personal information, and in narrow categories, material can be removed from the visible page history altogether through the projects' revision-deletion and suppression processes.

What does not get fixed

Equally important, and cheaper to learn here than through a failed engagement:

• A well-sourced fact you dislike. If reliable sources reported it, it belongs in the article, and the strength of your objection is not a factor.
• Coverage of a controversy, litigation, or regulatory action that received substantial independent reporting.
• The tone of an article that accurately reflects the balance of its sources.
• Removal of an article because the subject would prefer not to have one, where notability is clearly established.
• Anything framed as a legal threat. Making legal threats on Wikipedia leads to the account being blocked from editing while the threat stands, and it ends the conversation you were trying to have.

Wikipedia is explicitly not censored, and the community reads requests through that lens. The productive framing is always accuracy and sourcing, never harm to the subject's interests.

Deletion — how it actually works

Deletion is decided by community discussion, not by request. Any editor may nominate an article on the grounds that the subject does not meet the notability guideline; other editors then argue for or against, referring to sources rather than to preferences; and an administrator closes the discussion by assessing the consensus that emerged, not by counting votes.

Two nuances matter for subjects. First, where a living person is only marginally notable and has asked for the article to be deleted, that preference can legitimately be weighed in the discussion — the community has an explicit practice of leaning toward deletion in genuinely borderline biographies. Second, "marginally notable" is doing heavy lifting in that sentence. Where independent coverage is substantial, a subject's preference carries no weight at all, and raising it can make the discussion go worse.

Outcomes are genuinely uncertain, which is why nobody can promise one. A discussion can also produce results other than keep or delete — merging the content into a broader article, or redirecting it — and for a subject with a thin article, a merge is sometimes a better practical outcome than a contested deletion attempt that fails publicly.

The history problem, and the mirror problem

Two structural facts change what "removal" means here.

The first is that Wikipedia keeps a complete public revision history. Correcting a sentence does not erase the earlier version; it remains visible to anyone who opens the history tab. Only the narrow revision-deletion and suppression processes remove material from view, and they are reserved for specific categories such as non-public personal information, serious defamation, and copyright breaches. For ordinary corrections, assume the previous wording remains findable.

The second is that Wikipedia content is freely licensed and mirrored widely. Sites republish articles wholesale, and many of those mirrors update irregularly or never. An article that changes today may persist in its old form across dozens of copies for years. This is why a Wikipedia engagement that stops at the article itself is incomplete, and why the search-layer work is usually the longer half.

"The only argument that moves a Wikipedia article is an argument about sources. Every other argument — fairness, proportionality to your business, how it makes you feel — is addressed to a party that does not exist."

Realistic timelines

There is no service level because there is no service. A talk page request is answered when a volunteer chooses to answer it. What we observe: a narrow, well-sourced correction on an article with active watchers moves considerably faster than a broad rewrite proposed on an article nobody follows. Requests under the living-persons policy attract attention fastest, because experienced editors treat them as a priority. Deletion discussions run for a set period before closing, and can be relisted for further discussion when consensus is unclear. Everything else is genuinely open-ended, and a request sitting unanswered for weeks is normal rather than a signal that something went wrong.

The one thing that reliably slows everything down is volume. A talk page filled with a long list of demands is a task nobody wants to pick up. Three precise, individually sourced requests will be handled faster than twenty vague ones.

When the article does not change

Plan for the article staying substantially as it is, because for well-sourced subjects that is the likeliest outcome. What remains is real work, and it is mostly around the article rather than inside it.

Work the sources, not the article. Wikipedia reflects what reliable sources say. Where the underlying coverage is inaccurate, the corrections belong upstream — a published correction from the outlet changes the source that the article depends on. That is a separate discipline, and it is the one we describe under news coverage and editorial corrections. It is slow, but it is the only route that changes the article's foundations rather than its wording.

Address the search layer. For many subjects the practical harm is not the article's existence but its dominance of a name search, along with its role in feeding knowledge panels and AI summaries. That is a ranking problem addressed through de-indexing and search-result work where a URL qualifies, and through displacement work where it does not. Wikipedia is among the hardest results to displace, and an honest assessment says so before any budget is committed.

Check the adjacent record. Wikipedia articles about people frequently cite court filings and public records, and the underlying records have their own routes — court record removal where a case status has changed, and data broker opt-outs for the personal detail that accumulates around a name independently of any encyclopedia entry.

Consider the European route where it applies. For individuals in jurisdictions covered by the GDPR, search de-listing is a separate mechanism from anything Wikipedia does, with its own qualifying conditions and its own escalation path. We walk through it in our piece on Article 17 requests. It does not touch the article; it can affect whether the article is what people find.

The strategic summary

Wikipedia rewards a very specific discipline: declare the conflict, stay off the article, propose exact wording on the talk page with the source attached, and accept that uninvolved editors decide. Corrections grounded in sourcing succeed. Objections grounded in preference do not. Undisclosed editing, sockpuppets, and legal threats each convert a manageable problem into a permanent public record of an attempt to manipulate the encyclopedia.

The step that determines everything is the assessment. Our Wikipedia practice begins with a clause-level read precisely because it so often changes the plan — subjects arrive convinced the whole article should go and leave with three well-founded corrections, or arrive resigned to a sentence that turns out to fail its own citation. The full sequence, from that read through to disclosed talk-page filings and close-out, is set out on the Wikipedia page removal service. And where the honest answer is that the article is accurate and staying, we say so before the scope is signed rather than billing for requests that were never going to be accepted.