A deleted Wikipedia page feels like a closed door. It isn't — but the door you walk back through depends entirely on how it closed, and most people start pushing on the wrong one. They fire off an angry message to the last editor who touched the page, or quietly recreate it under the same title, and both moves usually make things worse. Some get the title locked against re-creation for their trouble.
This article is the recovery playbook we run when a company comes to us with a page that's gone. We'll use the real process names throughout — speedy deletion, PROD, AfD, REFUND, DRV, AfC — because the right move is different for each, and because anyone you hire should be able to tell you which one you're in within ten minutes of reading the logs. If they can't, that's your answer about them.
One thing to hold onto before we start: in almost every case, the path back doesn't run through arguing the deletion was unfair. It runs through fixing the reason the page was deleted in the first place. Recovery and re-qualification are the same project.
First, diagnose: how was it actually deleted?
There are four broad ways a Wikipedia article disappears, and they sit on a spectrum from "fast and unilateral" to "slow and deliberated." The fix for each is genuinely different, so this is the first thing to establish — before any action at all.
Speedy deletion (CSD). The fast lane. A single administrator can delete a page immediately under one of the criteria for speedy deletion, without any discussion, when it falls into an obvious bucket. The two you'll meet most often as a company are G11 (unambiguous advertising or promotion — a page so promotional it would need a fundamental rewrite to be neutral) and A7 (an article about an organization, person, or web content that doesn't even credibly indicate why the subject might be significant). There's also G12 for copyright violations, often triggered when a page is pasted from the company's own "About us" text. CSD is the bluntest instrument on Wikipedia and, paradoxically, often the easiest to recover from — because a speedy deletion is a judgment about the article as written, not a community ruling that the subject can never have one.
Proposed deletion (PROD). The quiet middle path. Any editor can tag an article as an uncontroversial deletion candidate; if nobody objects within seven days, an administrator removes it. The defining feature of PROD is that it's meant only for cases where deletion is not expected to be contested — which is also its weakness from the deleter's side and your opportunity: a PROD deletion can be undone almost on request. More on that under REFUND below.
Articles for Deletion (AfD). The deliberated path, and the one with the most weight. An editor formally nominates the article for discussion, and over (usually) seven days the community debates whether the subject meets notability — for a company, that's WP:NCORP, not just the general guideline. An administrator then closes it: keep, delete, redirect, merge, or no consensus (which defaults to keep). An AfD "delete" is a community judgment, which makes it the hardest of the four to simply reverse — you generally can't just ask for it back. It usually requires either new evidence or a flawed-process argument, which we'll get to.
Office actions. The rarest and the heaviest. These are deletions or locks imposed directly by the Wikimedia Foundation — the legal entity behind Wikipedia — usually for legal reasons (a court order, a credible legal threat, child-protection issues). They are not community decisions and cannot be appealed through the normal community channels at all. If your page was removed by an office action, none of the ladder below applies; that's a legal-and-communications matter, not an editorial one, and you need counsel rather than an editor. It's uncommon, but you have to rule it out first — spending three weeks preparing a Deletion Review for an office action is wasted effort.
The single most important early task, then, is correctly classifying which of these four happened. Everything downstream depends on it.
How to read the deletion log and the closing rationale
Wikipedia keeps a permanent, public record of every deletion, and learning to read it is the difference between a guess and a diagnosis.
Start at the page title itself. Even after deletion, visiting the old URL usually shows a notice that the page was deleted, when, by which administrator, and — crucially — a short reason, often citing the exact criterion (for example, "G11: Unambiguous advertising or promotion"). That single code tells you which of the four routes you're in. A "G11" or "A7" means speedy. A "PROD" or "Expired PROD" means proposed deletion. A reason like "per AfD" or a link to a discussion means it went through Articles for Deletion.
From there, dig one level deeper:
- The deletion log for the page (via the page's logs, or by searching the title in Wikipedia's log interface) lists every deletion and restoration event, with timestamps, administrator names, and the reason entered for each. If a page has been deleted and recreated multiple times, you'll see the whole history here — and a long history is itself a warning sign that the title may be protected.
- The AfD discussion, if there was one, is archived permanently and is the most valuable document you can read. Don't skim it for the verdict — read it for the reasoning. Which sources did editors dismiss, and why? Did they collapse a stack of reprints into a single source? Did they call coverage "routine announcements" under WP:ORGIND? Did they say the subject failed NCORP specifically, or was the complaint about tone and promotion? The closing administrator's summary tells you which arguments actually carried weight. That rationale is your repair brief.
Here's the distinction that matters most: was the page deleted because the subject doesn't qualify, or because the article was bad? A page deleted for promotional tone or poor sourcing — when the subject genuinely meets NCORP — is highly recoverable; you fix the writing. A page deleted because the community examined the sources and concluded the subject simply isn't notable yet is a much harder problem, and no rewriting fixes it. Only new, independent coverage does. Misreading which situation you're in is the most expensive mistake in page recovery — it sends you down a months-long path that was never going to work.
The options ladder
Once you've diagnosed the deletion type and read the rationale, there's a sensible order of escalation. Climb it from the bottom — each rung is lighter-weight, faster, and less adversarial than the one above, and skipping straight to the top annoys the very people whose goodwill you need.
Rung one — talk to the deleting administrator. The most underused move in the process. The administrator who deleted the page has a name, a talk page, and — in the overwhelming majority of cases — no agenda against you. A calm, specific message ("I noticed this page was deleted under G11; I'd like to understand what would make a neutral version acceptable, and whether you'd userfy it so I can work on a draft") often gets you a straight answer, sometimes a copy of the deleted text, and occasionally a quiet reversal if the deletion was borderline. Lead with a question, not a grievance. Administrators delete dozens of pages a week; they respond very differently to "help me understand this" than to "you were wrong."
Rung two — REFUND. Wikipedia has a process literally called Requests for Undeletion (shortcut: WP:REFUND), designed for uncontroversial restorations. It's the right tool for two situations. First, PROD deletions: because a PROD is supposed to be uncontested, you can ask for the page back and it's typically restored no questions asked — at which point anyone who still wants it gone has to go through full AfD. Second, getting a copy of deleted content so you can rebuild it properly, often "userfied" into your own drafting space. What REFUND does not do is overturn an AfD: if the community discussed and deleted the page, REFUND declines and points you upward. Knowing that boundary saves a wasted request.
Rung three — Deletion Review (DRV). The formal appeal court, and it exists for one narrow purpose: to review whether the deletion process was followed correctly, or whether significant new information has come to light. DRV is not a re-run of the AfD. You cannot simply re-argue that the subject is notable and hope for a different jury — that gets closed as "endorse" fast. The arguments that win are specific: the closing administrator misread the consensus; a speedy-deletion criterion was applied where it didn't fit; the discussion overlooked sources that genuinely exist; or substantial new independent coverage has appeared since. Bring evidence, frame it as process-or-new-information, and be concise. A DRV that reads like a notability tantrum loses; one that reads like a clean procedural argument has a real chance.
Rung four — redraft through Articles for Creation (AfC). Frequently the smartest route, and the one we recommend most often, because it sidesteps the adversarial frame entirely. Instead of fighting to restore the old (often flawed) page, you build a new, clean draft and submit it through Articles for Creation, where an experienced reviewer checks it against notability and sourcing standards before it goes live. A draft that passes AfC is published with a reviewer's blessing already attached, which makes it dramatically more resilient to a second deletion nomination. For a page deleted on tone or sourcing grounds, AfC is usually faster and more durable than DRV — you're not asking permission to reopen an old fight, you're demonstrating a better article. We treat this as the backbone of most page recovery engagements for exactly that reason.
The ladder isn't rigid — sometimes you talk to the admin and prepare an AfC draft in parallel — but the principle holds: exhaust the light, collaborative options before reaching for the heavy, adversarial ones.
When recreation is allowed — and when the title is salted
A reasonable question: if the article was deleted, can you just write a better one and publish it under the same title?
Sometimes, yes. If a page was deleted for being promotional or poorly sourced, and the subject genuinely meets NCORP, a substantially different, properly-sourced article on the same title is explicitly allowed. Wikipedia even has a speedy-deletion criterion (G4) that applies only to recreations substantially identical to the deleted version — which tells you the inverse plainly: a recreation that is materially better and addresses the deletion reason is not a G4 candidate and is permitted. The keyword is different. Paste back the same promotional text and it'll be gone within the hour, sometimes faster than the first time.
But there's a wall you can hit, and you need to check for it before you spend a day drafting: salting, more formally create-protection. When a title has been repeatedly recreated — usually by someone ignoring earlier deletions, or by undisclosed paid editors spinning up version after version — an administrator can protect it so that new pages can't be created there at all except by editors with elevated permissions. A salted title is Wikipedia's way of saying "stop." You'll spot it in the deletion log (a "protected" entry alongside the deletions) or by the create page itself telling you creation is restricted.
Salting doesn't make a page impossible forever, but it changes the route entirely. With a salted title you cannot simply publish; you have to go through AfC or DRV, demonstrate that a genuinely qualifying article now exists, and ask an administrator to lift the protection or move your approved draft into place. This is precisely why the "just create another one" reflex is so damaging — every careless recreation pushes a title closer to salting, turning a straightforward writing job into a permissions negotiation. If you've already triggered protection through repeated attempts, that's recoverable too, but it's specialist work, and it starts with stopping the recreations cold.
Fix the actual reason before you resubmit
Here is the part most recovery attempts skip, and it decides whether you get the page back or just get deleted again. Before you spend an hour on REFUND, DRV, or an AfC draft, you have to honestly fix whatever actually caused the deletion. The closing rationale you read earlier is your specification, and it almost always points to one of three root causes.
Notability. If the deletion turned on the subject failing NCORP, no rewrite saves you — the problem isn't the words on the page, it's the absence of qualifying sources behind them. The only real fix is new evidence: additional significant, independent, secondary coverage that didn't exist (or wasn't presented) at the time of deletion. That's a coverage-building project, not an editing one. If your source list at the AfD was thin and nothing has changed since, the page isn't coming back yet, and pretending otherwise burns money and credibility with the editors you'll need later.
Sourcing. Sometimes the subject is notable but the article was built on the wrong evidence — press releases, interviews, contributor columns, directory listings, reprints of a single wire story counted as if they were many. The fix is a clean source assessment: identify the handful of citations that genuinely clear the four-word test (significant, independent, reliable, secondary), build strictly from those, and drop the rest. A page sourced to three solid independent features survives where one sourced to thirty announcements dies.
Promotional tone. The most fixable cause of all. If the page was speedily deleted under G11, the issue is voice, not validity. Marketing language ("a leading provider of innovative solutions"), peacock adjectives, mission-statement paragraphs, and over-reliance on the company's own framing all read as advertising to editors. The fix is a genuinely neutral rewrite: encyclopedic structure, claims attributed to independent sources, the promotional register stripped out. A subject that qualifies but was written like a brochure gets deleted for tone and recovered through rewriting — the single most common recovery we handle, and the most reliably successful.
Diagnose honestly which of the three you're facing, because they have different costs and timelines. Tone is days. Sourcing is weeks. Notability is months, or not at all yet.
The hidden risk: AI engines may still be citing the dead version
There's a complication that didn't exist a few years ago, and most people don't think about it until it bites: deleting a Wikipedia page does not delete its influence. Large language models — ChatGPT, Gemini, Claude, Perplexity and the rest — are trained on snapshots of the web, and Wikipedia is one of the most heavily-weighted sources in those training corpora. A page that was live when a snapshot was taken can persist in a model's "memory" long after Wikipedia itself has removed it.
The practical consequences are awkward. An AI assistant may confidently describe your company using facts from a deleted article — including the very promotional claims or errors that got it deleted. Or it may cite an outdated version, repeating information you corrected years ago. Mirrors and scraper sites compound this: deleted content frequently lives on across dozens of low-quality clone sites that AI crawlers also ingest, so the dead page keeps feeding the machines from the shadows. You can win the deletion fight on Wikipedia and still lose the visibility fight everywhere a model is the front door.
This reframes what "recovery" even means. Getting the page restored is necessary but not sufficient; the goal is for the current, accurate version to become the source both humans and AI systems converge on. That's a slower correction — it improves as models retrain on newer snapshots and as the live page accumulates authority — but it's the actual objective. It's also why a restored page is worth keeping clean and current through ongoing support rather than treating publication as the finish line. The version that's live and well-maintained today is the version the next generation of models will learn from.
What NOT to do
Some of the most damaging moves are the ones that feel most natural when you're frustrated. Each of these makes recovery harder, and several can get you blocked outright.
Don't use sockpuppets. Creating multiple accounts to recreate the page, pile into a deletion discussion, or simulate community support is called sockpuppetry, and Wikipedia's investigators are extremely good at catching it. CheckUser tools tie accounts to technical fingerprints, and a sockpuppet investigation that goes against you doesn't just lose the page — it can get every involved account blocked and the title salted. It also poisons your credibility permanently; editors have long memories and public records.
Don't edit-war. Repeatedly re-adding the page or reverting an administrator's actions is edit-warring, governed by bright-line rules (the three-revert rule among them). It never works — administrators have more tools than you do — and it converts a content disagreement into a conduct problem, which is far worse for you. The moment a dispute becomes "this account won't stop," the focus shifts from your sources to your behavior.
Don't engage in undisclosed paid editing. If a company (or an agency it hired) edits its own article without disclosing the paid relationship, that violates Wikipedia's paid-editing disclosure policy and its terms of use. Discovery is common, the fallout is public, and it frequently triggers exactly the salting and scrutiny that make a page unrecoverable. The legitimate path is openly disclosed editing within the policy framework — which exists precisely so this work can be done above board. Any vendor who proposes doing it quietly is proposing the thing most likely to destroy your chances, and for recovery it's even more critical, because a deleted page is already under a brighter spotlight than a fresh one.
The throughline: Wikipedia is an open, transparent system with a long memory. Every tactic built on hiding, faking, or forcing works against that grain, and the grain always wins eventually.
When a different route beats fighting the deletion
Finally — and this is the honest part most recovery sellers won't say — sometimes the smartest move is not to fight for the page at all, at least not yet. If the deletion was correct, throwing money at REFUND, DRV, and repeated drafts is just a slower way to waste it. There are three alternatives worth weighing seriously.
Wikidata. If the underlying problem is that you don't yet meet NCORP, a Wikipedia article isn't the right target. Wikidata — Wikipedia's structured-data sibling — has a far lower inclusion bar: it accepts entries for entities that are simply identifiable and referenced. A well-built Wikidata item won't give you a prose page, but it feeds Google's Knowledge Graph and is increasingly read by AI systems, which means you can secure machine-readable visibility now while you build toward eligibility. For many "deleted and not yet recoverable" subjects, this is a better use of the budget than a doomed appeal.
Simple English Wikipedia. A separate edition with its own community and a sometimes more forgiving review culture. It is not a loophole around notability — the standards still apply — but for some borderline subjects it's a more achievable entry point than the full English Wikipedia, and a published Simple English article can be a legitimate foothold while the main-edition case strengthens.
Build the coverage first. When the deletion rationale comes down to "the sources aren't there," the only durable fix is to go make them be there — earned, independent, secondary coverage accumulated over time. This is an earned-media project rather than a Wikipedia one, and it's the rung companies most often want to skip. But it's also the only path that genuinely closes a notability gap, and done properly it doesn't just unlock recovery — it improves how every system, human and AI, understands your company. A new, clean Wikipedia page built on that foundation tends to survive where a forced recreation of the old one never could.
Recovery is real, and a large share of deleted pages can come back — but only after an honest diagnosis of why they left. Fix the reason, climb the ladder in order, stay inside the rules, and pick the route that matches the actual problem rather than the one that feels most like winning. That's the whole method.
If your page was deleted and you want a straight read on whether — and how — it can come back, email team@wikibusines.com with the page title and the deletion log, and we'll tell you which route actually fits.