Roughly 78% of Wikipedia article deletions cite a problem with sources or notability — meaning the draft either didn't qualify in the first place, or the citations couldn't carry the claims. Drafts that get declined at Articles for Creation (AfC) before they ever reach mainspace are an even larger pool. Most of them fail for the same six reasons, in roughly the same order, and most of those failures are predictable from the draft before submission.
This piece is a reference for anyone whose Wikipedia draft just came back declined — or anyone considering submitting one. The six reasons below cover well over 90% of declines in our experience. Each is a real Wikipedia policy with a shortcut you can quote on the talk page, and each has a fix. Some fixes are mechanical and can be done in a day. Others — the notability ones — require a fundamentally different approach, because no amount of rewriting will turn a thin source base into a publishable article.
We work on Wikipedia drafts professionally, so we have a stake in this. The honest framing: most of the drafts we don't take on, we decline because the notability case isn't yet there — and we say so in writing rather than running an expensive failure. Most of the drafts we do take on don't fail on the process problems below, because we've engineered those out. Where they fail — and a small fraction do — it's at WP:N or WP:RS, the same place every honest draft can fail when the underlying coverage runs short.
1. The subject isn't notable under WP:N or WP:NCORP
By a wide margin, the most common reason. Wikipedia notability is not a measure of importance, fame, or success. It's a specific technical standard: significant coverage in reliable independent secondary sources. The general guideline is WP:N; for companies and organisations the stricter test is WP:NCORP. To meet either, you need multiple independent, reliable, secondary sources that cover the subject in depth — typically three or more substantive pieces from outlets the community treats as reliable, written by independent journalists, not derived from press releases.
Founders are routinely surprised by what doesn't count. A LinkedIn following of half a million, a podcast appearance on a major show, a Forbes contributor column written by the founder themselves, three hundred press-release pickups about a funding round, a TEDx talk — none of these establish Wikipedia notability. The same goes for impressive metrics: revenue, users, headcount, market share. Wikipedia doesn't measure success; it measures coverage. The two correlate, but imperfectly, and a draft built around success metrics rather than independent coverage will be declined.
The fix is not a rewrite. The fix is either (a) building the source base — an earned-media programme that produces the independent coverage that doesn't yet exist — or (b) accepting that the subject isn't yet notable and not pursuing the page right now. Our Wikipedia notability audit runs every prospective source against WP:N and WP:NCORP and gives back a written verdict per source — so before any drafting starts, you know whether the case is real or whether the honest answer is "not yet, here's the gap."
2. Promotional tone — WP:NPOV violation
Wikipedia's neutral-point-of-view policy (WP:NPOV) is non-negotiable. Drafts written in marketing language are declined on sight, regardless of how strong the sources are. The specific tells reviewers spot in the first sentence: words like revolutionary, leading, innovative, world-class, premier, pioneering, cutting-edge, game-changing, disruptive, award-winning, market-leading. None of these belong in an encyclopedia. Reviewers treat them as evidence the article was written by someone too close to the subject, and treat the rest of the draft with heightened scrutiny.
The fix is mechanical. Replace promotional adjectives with attributed facts. Don't say "leading marketplace for X"; say "marketplace for X, which according to [reliable independent source] had Y users as of [date]." Don't say "award-winning team"; if the team won an award worth mentioning, name it, cite the awarding body, and let the fact speak for itself. Don't say "innovative approach"; describe what the approach actually is, in plain terms, and let the reader judge.
A useful internal exercise: read the draft aloud, and every time you encounter a positive adjective without a citation, flag it. If the adjective remains because the citation says it — "described by The Financial Times as 'one of the most influential firms in European fintech'" — keep it. If the adjective is unsourced editorialising, cut it. We routinely halve a client-supplied draft on this pass alone, and the cut version reads as more credible, not less, precisely because it sounds like an encyclopedia rather than a press release.
3. Weak sourcing — WP:V violation
The verifiability policy (WP:V) requires that every contested claim be supported by a reliable published source — and "reliable" has a specific meaning under WP:RS that doesn't match common usage. The community-maintained Perennial Sources list (WP:RSP) rates which outlets are reliable for what kinds of claims. Drafts cited largely to blog posts, LinkedIn, press releases, Crunchbase, the company's own website, Medium articles by the founder, or Forbes contributor pieces will be declined because none of these clear WP:V for substantive claims about the subject.
What does clear WP:V: reported pieces in established outlets with editorial oversight — Reuters, AP, BBC, Financial Times, The New York Times, The Guardian, Le Monde, major national broadsheets in your market — plus reputable trade press and peer-reviewed sources. What sits in the middle: Forbes staff (not contributor) work, Business Insider, Wired, TechCrunch (situational depending on the section and the year), most trade outlets. What fails: press-release wires (PR Newswire, BusinessWire), contributor networks, social media, the company's own website, LinkedIn, Crunchbase, founder-authored op-eds.
The fix depends on whether reliable sources exist and weren't cited (easy — replace the citations), or whether reliable sources don't exist for the claims being made (hard — either remove the claims, or build the coverage first). A draft heavy on company-website and press-release citations isn't a sourcing-format problem; it's a notability problem in disguise. See reason 1.
4. Conflict of interest — WP:COI
Wikipedia's conflict-of-interest policy (WP:COI) says editors should not edit articles about subjects with which they have a financial, personal, or close relationship. Editing on behalf of an employer, client, or family member triggers automatic scrutiny — and the Wikimedia Terms of Use require paid editors to disclose the relationship on their editor user page, on the article talk page, and in edit summaries. Failure to disclose is a Terms of Use violation, and the consequences run from page deletion to permanent editor bans to public Sockpuppet Investigation cases that attach the client's name to the violation in perpetuity.
A draft submitted by an editor with an obvious COI but no disclosure will be declined even if every other element is correct. Worse, it may be deleted later under the speedy deletion criterion G11 (unambiguous promotion) or G5 (created by a banned user) once the COI surfaces. We've seen pages that survived for two years get deleted in a week when the editor who created them was identified as undisclosed paid editor — and the deletion carried into other pages the same editor had touched.
The fix is structural. Either: (a) the article is edited by a properly disclosed paid editor who declares the relationship on all required surfaces, complies with the WP:PAID policy, and follows the WP:COI guidance to propose changes on talk pages rather than editing directly; or (b) the subject's own employees do not edit the article and the work is handed off to disclosed third parties. Our Wikipedia page creation work and the underlying conflict-of-interest policy explainer lay out exactly how disclosure is handled on our projects: every editor's user page declares agency, client, and end-beneficiary; talk pages carry the same declaration; edit summaries tag each contribution.
5. Style violations — WP:MOS
The Manual of Style (WP:MOS) is enormous, but five specific style errors account for the bulk of MOS-related declines. Heading hierarchy: Wikipedia articles use H2 for major sections and H3 for sub-sections — H1 is reserved for the article title and must not appear in body content. Excessive bullet lists: encyclopedia prose runs in paragraphs; long bulleted lists of features, products, or executives signal corporate copy rather than reference content. Marketing infographics and decorative images: Wikipedia uses informational images only — product shots, logos in infoboxes, charts with cited data — not lifestyle photos or marketing visuals. Non-standard citations: every citation should use the {{cite web}}, {{cite news}}, or {{cite journal}} template family; bare URLs, footnote shortcuts, or inline links to commercial pages all fail. External links inside body text: links to the company's website, products, or social media inside the article body get stripped automatically.
These are all mechanical fixes. A draft that's strong on notability and sourcing but fails MOS is the easiest category to repair — usually a few hours of cleanup. The danger is that MOS violations combined with promotional tone or weak sourcing compound: the reviewer reads the first paragraph, sees marketing language, scans the rest, notices the wrong heading levels and bullet-heavy structure, and declines the whole draft on the cumulative impression rather than any single issue. Cleanly-formatted prose buys the rest of the draft a fairer reading.
6. New account or low edit count
Wikipedia's community is suspicious of accounts that register, immediately create a draft about a company, and disappear. This is a behavioural signature of undisclosed paid editing and sock-puppetry, and the AfC reviewer pool watches for it. A draft submitted by an account with three edits — all of them on the draft itself — gets harder scrutiny than the same draft from an account with two years of edit history across many topics.
This is not formally a notability standard, but it's a real friction. Drafts from new accounts often get declined on the first pass for borderline issues that would have passed for an established editor. The fix is structural: don't submit drafts from brand-new accounts, especially if there's any COI. Either build edit history first (slowly, organically, on unrelated topics — which takes months and is rarely worth the time), or — much more commonly — hand the work to an established editor with a paid-editing declaration, which is the disclosed, policy-compliant route.
The pattern we've seen most often: a marketing or PR agency, told by a client to "get a Wikipedia page," registers a fresh account and submits the draft. The draft is declined. They register another account. That gets declined too. By the third or fourth round, the pattern is visible to administrators, the accounts get blocked, the agency is identified, and the client's brand is publicly associated with what's now a sock-puppet case. None of this is recoverable cheaply. The right path was disclosure from the start.
What this looks like for the drafts we run
A fair question for any agency: when your drafts fail, what do they fail on? We track this honestly. The vast majority of our drafts pass AfC on the first or second round. The ones that fail — a small percentage — fail almost exclusively at reason 1: the underlying notability case turns out to be thinner than initial source review suggested, usually because a piece we'd counted on as independent turns out to be a churnalised press-release pickup, or because a Forbes citation we'd flagged as staff-written turns out to be a contributor piece, or because a key source has been deprecated since our last review.
What our drafts don't fail on, because we've engineered the failure modes out at the process level: promotional tone (we cut it before submission), weak sourcing format (we use proper citation templates), undisclosed COI (every editor on our team has a public paid-editing declaration), MOS violations (we run a style pass before AfC), or new-account submission (our editors have established histories and reputations).
That distinction matters because it tells you which kinds of failure are addressable by process and which aren't. WP:N and WP:RS failures, when they happen, are honest signals that the underlying coverage isn't yet sufficient — and the right response is either to build the coverage or to acknowledge that the subject isn't yet ready. Any of the other five failure modes is a process problem the agency should have caught. If your last draft failed on tone, sourcing format, COI, MOS, or account history, what you needed wasn't a better source list — you needed a different team running the submission.
What to do now
If your draft has just been declined, the AfC reviewer leaves a comment with the policy shortcut they cited. Read the comment first — it tells you which of the six reasons above applies. If the reason is notability or sourcing (WP:N, WP:NCORP, WP:RS, WP:V), the fix is upstream of the draft; the question isn't how to rewrite, it's whether the case is real. Our notability audit addresses exactly this. If the reason is tone, format, MOS, or account history, the fix is in the draft itself, and a competent rewriter can repair it.
If the page has already been deleted rather than just declined, the recovery path is different — REFUND, DRV, and AfC all play different roles depending on how the deletion happened. We've written a full guide to that ladder under Wikipedia page recovery. The wrong move is to register a new account and resubmit the same draft; the right move depends on which deletion process closed the page.
For the projects we do take on, the disclosure framework and full process are described under Wikipedia page creation, and the failure-case clauses are in our guarantees. If you'd like a second opinion on a declined draft before deciding what to do with it, the next section is the simplest way to get one.
Have a declined draft you'd like read by an experienced editor? Email team@wikibusines.com with the draft and the AfC reviewer's comment, and we'll send back an honest read on what failed and what — if anything — would fix it.