- William Beutler
- Aug 13
- 6 min read
Wikipedia turns 25 next year—ancient by internet standards. Chances are you first stumbled across it while writing a paper in school, then kept coming back because it was the most convenient place to find information. But the old dog has some tricks left in it: Wikipedia is also an influential source of information for modern LLMs, making it an important part of generative engine optimization.
It should come as no surprise. Wikipedia is free to use, highly structured, and staggering in scope, making it ideal for both training and citations. Multiple independent analyses of AI training data verify its prominence. ChatGPT cites it in nearly half of all answers. It’s not a stretch to say today’s LLMs couldn’t function the way they do without it.
Meanwhile, AI is taking center stage: ChatGPT is now the fifth-most visited website, and AI search is growing 721% month-over-month. The era of AI-driven answers has arrived, and Wikipedia is its secret sauce.
I should know. I’ve been helping companies navigate Wikipedia because of its impact on search for almost 20 years. And recently, my inbox has filled up with questions about Wikipedia’s impact on AI like nothing I’ve seen before.
The potential and the pushback
Here’s the rub: the opportunity Wikipedia presents is offset by the challenge of getting it right.
Wikipedia isn’t like your website or LinkedIn page. You can’t just create an account and post whatever you want. In addition to being a great place to find out what else that actor from that one thing was in, it’s also a collaborative knowledge project governed by a tight-knit community with strict rules about neutrality, sourcing, and whether that actor deserves their own page. That’s why it’s taken seriously.
These editors know quite well that marketers see Wikipedia as a high-authority platform, and they’re quick to push back on anything that feels promotional. Many companies encounter this when trying to create a new article and getting rejected—sometimes repeatedly.
Notability ≠ Importance
The problem often starts with a misunderstanding of Wikipedia’s inclusion criteria, which is at least partly Wikipedia’s own fault. The relevant guideline is called Notability, which leads many to assume it’s about how important a person or company is. If you’re important, you should be in Wikipedia, and if Wikipedia says no, they’re saying you’re not important—and a harm has been done.
But that’s not what the Notability guideline says at all. It’s about how well a topic has been covered by reliable, independent sources. What you really need is dedicated coverage by professional journalists in major news outlets explaining to readers why you are interesting. Writing for such an outlet yourself, going on podcasts, or doing TV news hits won’t get you there.
That means achieving recognition on Wikipedia is as much about running a successful earned media campaign as it is your intrinsic value. If major news outlets haven’t written about you at length, Wikipedia probably doesn’t want an article about you, either.
A company featured in multiple national publications describing them as an industry leader would clearly qualify. A fast-growing startup with mostly tech coverage might, but it has a better chance with a couple of profiles in top-tier sources. A mid-sized law firm that doesn’t make headlines? Probably not. A regional hot chicken chain? That one might actually get created on its own.
Even existing articles need work
Even if you already have a page, that doesn’t mean you won’t have trouble with it, especially if it’s missing opportunities to send good vibes to LLMs listening in. It’s unfortunately common for articles to go stale, get bogged down in minor controversies, or just be poorly written.
As a category, Wikipedia has largely neglected business topics. Its guidelines allow for coverage of a pretty generous range of businesses and businesspeople—Britannica would never—but Wikipedia editors tend to focus on their personal interests. Editors are hobbyists, not brand historians. They’re not thinking about the poor, confused chatbot trying to summarize your company.
That’s when the temptation to “help” kicks in. After all, who knows your brand better than you?
But editing your own page is risky. If you jump in and start adding language that reads like marketing copy, or uses the wrong sources, it may get removed, flagged, or worse. Warning banners are notoriously difficult to resolve and remove.
So, what's a conscientious marketer to do?
A Wikipedia marketing strategy for AI search
Learn how Wikipedia works
Start with the content policies. After getting your head around Notability, the famous Neutral Point of View policy explains how to write balanced content, Reliable Sources explains what kinds of sources you need, and the Manual of Style explains formatting. By the way, I actually recommend asking AI to summarize these policies—all of which are well-considered, but on the longer side.
Next, Wikipedia has its behavioral expectations, most of which you should have learned in kindergarten (play nice with others, learn how to compromise). But the one to bone up on is Conflict of Interest (COI), which establishes the rules of the road for company representatives. Ultimately, it boils down to this: if you’re paid to work on a topic, you must disclose that connection. And most of the time, you shouldn’t edit directly. Make your case on the article’s Talk page instead.
Follow the edit request process
To suggest a change to an article, post your message on the article’s Talk page, and attach the Edit COI template. That signals to independent editors reviewing COI requests that you need help. If your suggestion is well-founded, they may make the change for you.
Don’t take it personally if it takes a while, or the first response is a tough question. Wikipedia isn’t a customer service desk, and you may have to go back to the drawing board at least once. But also, know when to take a hint. I’ve often seen company directives disintegrate on impact with Wikipedia’s editor community. Know when to cut your losses. It’s their platform.
Familiarize yourself with articles for creation
If you’re creating a new article, use the Articles for Creation process, which the COI guideline instructs paid editors to use. It’s slow, strict, and opaque—but it’s your best bet. You’ll need more sources than you think, getting the wikitext just right is a pain, and sometimes you need to wait months for review—but if you did it right, the result is your article live on Wikipedia. If it’s rejected, Wikipedia editors don’t have the time to write a detailed review, let alone an action plan. It’s best to assume that you need better sources.
Assess your brand’s position
If you have a page, audit it: What’s missing? What’s outdated? Where is it linked from other pages? If you don’t have a page, determine whether you qualify based on your press coverage. In either case, gather the best independent sources you can find in respected outlets. You’ll need them at some point.
Also consider trying one of the new breed of AI monitoring tools like Anvil, Profound, or xFunnel. They can help track where and how your brand is mentioned across popular LLMs, and help see how changes on Wikipedia affect AI answers over time.
Assign a point person
If you think you have the goods, it’s time to operationalize. You’ll probably want to find a seasoned consultant—yes, I have a conflict of interest here—who knows the rules, the culture, and how to avoid common pitfalls, of which there are many. Ask them to show you their track record, including the diffs. If they can’t show you their actual work, it’s better to look elsewhere.
Even if you keep your Wikipedia marketing efforts in-house, assign one person to represent your brand on the platform. They should use at least a real first name—not something like “AcmeCorpTeam.” For existing articles, start with a simple, well-sourced request. If no article exists yet, the first step is to learn how to write one.
Not ready for Wikipedia? Try Wikidata.
If your company doesn’t qualify for Wikipedia, there’s a useful alternative: Wikidata. It’s a structured database of machine-readable facts about real-world entities. It has broader inclusion standards and is more welcoming of commercial editing. Google’s Knowledge Graph has long run on it, and LLMs almost certainly draw from it too. Making sure your Wikidata entries are robust and accurate is a good way to boost AI’s confidence in how it understands your brand, and to limit hallucinations. Even if you have a Wikipedia page already, paying attention to Wikidata is low-risk, high-reward, and just good information hygiene.
The human element still matters
Getting Wikipedia right takes discipline, humility, and a long-term view. It requires seeing your company the way a cautious Wikipedia editor might—and knowing how to meet them halfway. Turning that skeptical editor into a fair-minded ally could have a major impact on how LLMs understand and describe you. Because even in this futuristic era with promises of robots that can do almost anything for us, there’s still one thing they can’t do: persuade a Wikipedia editor.
