The Wikipedia-services market has a sorting problem. A handful of operators do this work seriously — with editor accounts that disclose paid editing, source assessments that match WP:NCORP, refund clauses that actually pay out, and live URLs you can read. The rest, and it's most of the market, range from sloppy to outright fraudulent: ghost agencies that disappear after payment, sock-puppet rings that get clients banned by association, Trustpilot pages stuffed with fabricated reviews, and brand-new domains buying Google Ads on "Wikipedia page creation" the week they registered.
The reason this is hard for buyers is that the surface looks similar. Both kinds of agency promise the same outcome ("a Wikipedia page about your company") at roughly overlapping price points. The honest ones quote €4–8k for a page that will probably survive; the predatory ones quote €490 and disappear, or €25,000 and never deliver. From a procurement deck, they read like vendors competing on price. They aren't. They're different categories of risk.
This piece collects the nine warning signs we've seen recur across hundreds of intake calls from clients who'd already been burned somewhere else. None of them is conclusive on its own. Three or more, stacked, almost always means the agency in front of you isn't safe to hire — and the last section gives you five questions any legitimate operator can answer in a discovery call.
1. No real social-media or website footprint
Wikipedia work is editorial. Agencies that take it seriously tend to have a visible track record — a LinkedIn company page with named employees, a website older than the current sales cycle, blog posts that demonstrate they understand WP:RS and WP:N, conference appearances, podcast interviews, archived press coverage. None of this is decorative. It's how you tell whether the people you're hiring exist beyond the proposal in front of you.
When an "agency" has a single landing page, a Gmail contact, no LinkedIn company entity, and no named editorial team, that's a ghost. Ghosts can take payment and vanish; they can also use junior staff or outsourced freelancers with no Wikipedia experience and absorb the consequences invisibly. There's no recourse because there's no entity to recur against.
What to look for instead: an agency with a multi-year website history (check the Wayback Machine), founders and editors whose names appear on their LinkedIn profiles and have been in this field longer than 12 months, and at least some independent press coverage. Our own press archive is on the site for exactly this reason — buyers should be able to read what third parties have said about us, not just what we say about ourselves.
2. Trustpilot clusters that read like the same person wrote them
Fabricated reviews follow patterns. The reviewer profile is one to three weeks old; the profile reviews two or three companies (the agency, plus one or two unrelated businesses to look plausible); the review text uses formulaic praise ("amazing service, highly recommend, very professional"), avoids any specific detail about what was delivered, and clusters with twenty similar reviews posted across a two-week window.
Real client reviews are messier. They name the deliverable ("our company's Wikipedia page in the German edition"), reference timeline and pricing, sometimes mention the editor or account manager by name, and often disclose minor friction ("took longer than the original quote because we had to track down extra sources"). They're posted irregularly, by reviewers with mixed review histories, and the language varies from one to the next.
If a Trustpilot page shows a steep curve of identical-sounding 5-star reviews from new profiles inside a short window — and zero substantive 3-star reviews — assume the reviews were bought. Cross-check by Googling specific phrases from the reviews; if the same sentence appears verbatim on multiple profiles, it's a template. And remember: Trustpilot itself does not verify clients. The presence of a Trustpilot score is not evidence of legitimacy.
3. Vague language and "guaranteed approval"
Wikipedia does not "approve" pages, and no agency can guarantee that one will survive. Drafts pass through Articles for Creation (AfC) review by independent community editors who apply WP:N, WP:NCORP, WP:RS, and WP:NPOV — and even a well-built draft can be declined, sent for additional work, or ultimately deleted if the underlying notability case is weaker than it looked. Anyone who promises "100% approval" or "guaranteed Wikipedia page" is either lying or about to use sock-puppet accounts to push a draft through against policy (which gets the page deleted later, usually with the client's name attached to a public Sockpuppet Investigation case).
Honest operators talk about probabilities and prerequisites. They explain WP:NCORP's three-part test (significant coverage, independence of source, reliability of source). They describe what their draft will look like and which clauses of WP:NPOV it will respect. They quote ranges, not certainties. Our own guarantees page sets out exactly what we cover and what we don't — explicitly, in writing, with refund triggers stated.
If the proposal in front of you uses the word "guaranteed" and doesn't define the failure case in policy terms, the agency either doesn't understand the policies or is willing to violate them. Either way, the page won't last.
4. No verifiable case studies — only anonymous testimonials
The single most useful question on a discovery call: "Show me three live Wikipedia URLs you've published in the last six months." A real agency can. They'll send you links to live English Wikipedia (or German, or French) articles, point you at the editor account that created them, and walk you through the source list at the bottom of each page. You can independently verify the editor's user page declares paid editing, check the edit history, and read the discussions on the article's talk page.
A fake or struggling agency will give you anonymised "case studies" — "a SaaS company we worked with last year saw their Wikipedia page rank in Google" — without naming the company, linking the page, or identifying the editor. There's no way to verify any of it. In our experience, when the case studies can't be linked, they don't exist.
Pages we've created are visible at the bottom of our Wikipedia page creation service. If a competitor refuses to do the same, ask why.
5. Pricing anomalies — sub-€500 or unjustified five-figure quotes
There is a real economic floor under Wikipedia work. A defensible draft requires a source audit against WP:RS (3–6 hours), drafting in encyclopedic tone (8–15 hours), reference formatting (2–4 hours), AfC submission and response to reviewer comments (multiple rounds, often weeks), and editor time at €40–80/hour. Plus paid-editing disclosure infrastructure, project management, and the cost of an editor's reputation if the work is shoddy. Total realistic cost: €4,000–€8,000 for a standard B2B company page.
When you see a quote under €1,000, one of three things is happening: (a) the agency is using a non-disclosed sock-puppet account that will be banned, taking your page with it; (b) the work is going to a freelancer in a low-cost market with no Wikipedia experience and the draft will fail at AfC; or (c) it's an outright scam and the agency will disappear after payment.
At the other end — quotes over €20,000 for a single page — what you're often paying for is opacity. Without a line-item breakdown (source audit, drafting hours, editor rate, language editions, monitoring, recovery insurance), you can't tell whether the price reflects genuine complexity or padded margin. Our pricing page breaks costs into components for exactly this reason: you should know what you're paying for at every line.
6. Cannot explain WP:N, WP:COI, or WP:RS on the call
The fastest test of agency competence: ask, on the discovery call, "Walk me through WP:NCORP. What makes coverage independent under WP:ORGIND? How do you handle COI disclosure for our editors?" Anyone qualified to take your money should be able to answer for five minutes without consulting notes.
What you'll hear from competent operators: a clear explanation of the three-part NCORP test, a discussion of independence of content versus independence of publisher (a Forbes contributor piece written by your founder is not independent, even though Forbes is "reliable"), and a description of the user-page declaration their editors maintain under the Wikimedia paid-editing terms.
What you'll hear from incompetent or fraudulent operators: hand-waving, deflection ("our editors handle the policy side, don't worry about it"), invented terminology, or worst, an admission that they use anonymous accounts to avoid the COI policy. The last answer means the page will be deleted within months and the client's name may end up in a public Sockpuppet Investigation case file.
If you want to read the policies yourself first, our notability audit is built around the same framework — and you can read what it covers before you commission anything.
7. Poor website quality, no postal address
This sounds superficial. It isn't — it's a competence signal. Editorial work demands attention to detail. An agency that publishes a website full of grammar errors, broken pages, mismatched fonts, stock-photo "team members," and no physical postal address is signalling either that they don't care about detail or that they aren't a real business.
Wikipedia editors notice. So do Wikipedia administrators. A draft submitted by an editor whose linked agency has a broken website and no registered business gets extra scrutiny, on the reasonable assumption that the operator may be unreliable.
Specifically check: a postal address (not a virtual office or P.O. box), a registered company number you can verify in the relevant business registry, named directors or founders, and a website without obvious typos, broken links, or stock-photo "case studies." Our own legal entity and registration are on the contact page — if a competitor's isn't, ask why.
8. High-pressure sales tactics
"We have one slot left this quarter." "Our prices go up next week." "We're only taking three new clients this month." Real Wikipedia agencies don't talk like this. The work is editorial, capacity scales with editor headcount, and there's no rational reason for artificial scarcity in a multi-week service.
What high-pressure sales tactics usually signal: a sales team operating on commission with no editorial oversight, an agency that knows its quality is weak and needs to close before the client researches further, or a scam optimising for fast payment before the buyer thinks twice.
A serious operator will let you take a week to make a decision, send you written documentation to review, offer to put you in touch with past clients, and respect your timeline rather than pushing theirs. If the proposal says "valid until Friday," ask why.
9. Recently registered domain, no Google search history
Final check, and one of the easiest. Run the agency's domain through a WHOIS lookup. If the domain was registered three to six months ago and the agency claims "ten years of experience," something is wrong. Either they're a brand-new operation pretending to be established, or they're a rebranded predecessor that lost its reputation and started over.
Cross-check by Googling the agency name in quotes and looking for: independent press coverage, mentions in industry publications, conference appearances, archived versions on the Wayback Machine, and LinkedIn profiles of staff that predate the current domain. A real seven-year-old agency leaves a trail. A six-month-old one pretending to be seven years old leaves a thin one — and the gap is detectable in fifteen minutes of due diligence.
What to ask instead — the five-question discovery call
If you take nothing else from this, take these five questions. Ask every Wikipedia agency you talk to. The competent ones will answer in detail; the incompetent or fraudulent ones will deflect.
- "Show me three live Wikipedia URLs you've published in the last six months." Real agencies have them; fake ones don't.
- "Show me the editor's user page where they declare paid editing under the Wikimedia Terms of Use." Disclosure is non-negotiable. An editor without a paid-editing declaration is a sock-puppet waiting to be discovered, and your page goes down with them.
- "What's your refund clause if the draft fails at AfC or is deleted within six months?" A real clause names the trigger, the refund percentage, and the timeline. "We'll work with you" is not a refund clause.
- "Walk me through WP:NCORP. What makes a source pass WP:ORGIND?" Five minutes of substantive answer means they can do the work. Hand-waving means they can't.
- "Who is the editor account that will submit the draft, and is your agency named in their disclosure?" The honest answer names the editor, links to the user page, and confirms agency-client-end-beneficiary are all declared. Anything else means the work will be done under cover, which means it will be reversed.
If three or more of the warning signs at the top of this article apply to an agency, and they can't answer the five questions above in a discovery call, walk away. The cost of doing nothing is far less than the cost of having a Wikipedia page deleted, your editor banned, and your company's name attached to a public Sockpuppet Investigation — which is what predatory agencies leave behind. Our own page creation service, guarantees, and pricing are written explicitly to be checked against these criteria. You should be able to verify everything claimed about an agency, including ours.
Want a second opinion on a Wikipedia proposal you've received? Email team@wikibusines.com with the proposal attached and we'll send back an honest read on the warning signs — even if the verdict is "this looks fine, go ahead."