Inside Wikipedia's Plan to Survive the Answer Engine Apocalypse
This story was originally published in On Background with Mark Stenberg, a free, weekly newsletter that explores the key themes shaping the media industry. You can sign up for it here.
Wikipedia has spent 25 years building a reputation as one of the most trusted sources on the web.
Now, as answer engines rewrite how people find information online, that reputation faces a threat familiar to many publishers: ubiquity without credit.
According to the search engine measurement Ahrefs, ChatGPT cites Wikipedia more than almost any other single source, second only to Reddit. And yet, fewer humans are visiting the site itself to read, and more importantly, to edit.
As a nonprofit with a minimal reliance on advertising, Wikipedia is somewhat insulated from the traffic erosion crushing other publishers. That makes it a revealing case study in how institutions are rethinking relevance in the AI era.
At Brandweek, I asked Zack McCune, the global head of brand marketing at the Wikimedia Foundation, how the organization is responding.
Mark: Wikipedia just launched a brand campaign, created by the agency Kin, to celebrate its 25th anniversary. What prompted the campaign?
Zack: Wikipedia is 25 years old, which is quite a milestone. But we hit 25 at the exact time our relationship to knowledge is changing, and Wikipedia is at risk of becoming invisible—invisible to us and invisible to our awareness. When you search for something today, your voice assistant is giving you the answer, or your AI agent is giving you the answer, hiding the fact that, more often than not, the answer is coming from Wikipedia. For an organization that relies on hundreds of thousands of volunteers and hundreds of millions of donors, it was crucial to generate a whole new wave of support and create not only awareness but connection and engagement.
Mark: Why did you feel this was necessary, and what sort of goals were you hoping to accomplish?
Zack: Wikipedia is the backbone of knowledge on the internet, the backbone of its most trusted elements. And yet what we’re noticing on a brand health level is decline: decline in stability, a decline even in awareness or likability. It’s strange, because a lot of brands are thinking, “How do I get my content into the AI platforms?” I don’t have that problem. My content is everywhere, and yet the brand problem remains. We’re more visible than ever, and yet less known. Part of why we came to Kin, wanting to make our 25th birthday a big, splashy, emotionally driven project, was trying to get people to feel something for the knowledge they rely on. We do brand work globally all the time, but this was actually one of the first, one of the biggest campaigns done domestically in the U.S.
Mark: That was a bit of a canary in the coal mine, because brand awareness in the U.S. is usually not a problem for an institution like Wikipedia.
Zack: I’ve been with the Foundation leading brand for almost 11 years, and in that time I mostly worked on projects in Nigeria, India, Iraq, South Africa—places where we wanted to grow awareness. For us, brand awareness isn’t a metric by itself. Brand awareness becomes usage, and usage becomes editing. So the more people who know our platform and use it, the more complete our knowledge record is. It’s a public library where sooner or later we say, “We’d like you to write a book—could you do that on something you love?” We’ve been doing regional marketing for a decade. But as we moved into an underperforming market here, that’s exactly where we said: We need to do something bigger in the U.S. and Europe.
Mark: This is something I cover a lot at Adweek: websites disappearing as search engines change. Normally the concern for publishers is commercial: People aren’t coming to the website, they’re losing ad revenue. Wikipedia doesn’t have that same acute commercial exposure. But you were describing the real conversion problem as visitors to contributors, which I hadn’t thought about before. Is that the real “disappearing” issue for you, losing the people who contribute to the site?
Zack: That’s right. Year over year we see about an 8% decline in human traffic. We separate that from bot traffic, which is way up, crushing our servers. But when fewer humans make it to the site, that’s where the knowledge itself is composed. If you see a typo, if you realize something’s missing or wrong in an article, that’s where it gets fixed. All the composition happens on our site. So we’re thrilled about the distribution of knowledge through AI platforms, but we also need to constantly refresh and fact-check that content, and that means people have to come back to the homepage.
Mark: I’d love to return to the core thesis here. The aim of this campaign was to raise awareness of Wikipedia because it’s not as visible in search engines anymore, and that’s increasingly where people get their information. That’s a problem publishers across the internet are facing. One solution a lot of sites are taking is what one CEO called the “woo and sue” approach: You either strike a content licensing deal or you take AI companies to court. I know Wikipedia has some licensing arrangements. Can you talk about how Wikipedia is responding to AI as an organization, and how it plans to show up on the internet going forward?
Zack: We have an amazing group of people working on an enterprise offering, a real-time API for LLM companies that need tons of data very quickly, delivered in the way they’re used to programming against. That’s designed to do two things: get our content out on these platforms, and create a commercial relationship where the platforms themselves pay us for that usage. It’s a way to expand beyond the 8 million donors giving an average of $11 to sustain the Foundation, because we want to make sure these companies are paying their fair share too. We’re working with every major AI/LLM company you can think of, trying to establish a virtuous relationship. I hear the word “sustainable” a lot, but that feels too low-stakes. We have to anticipate 2028, 2029, and beyond. How do we keep a human content community in a position to keep all of this fact-checked for these platforms? That’s what we’re building right now. We’ve got deals underway, some public, and I hope more to come.
Mark: Your mission is to keep the world’s information accurate and free, and however people find that, it’s a win for you. So the north star is a little different from a standard website. Is that fair to say?
Zack: It is. I think the advantage we have, which is unique, is that we’re a nonprofit. We’re sustaining a mission: knowledge for the world across 300 languages and counting. What we used to worry about was losing volunteers, the community of fact-checkers and writers. It’s not about converting you to a paid subscriber, and we have no ads. The knowledge itself should go everywhere. But we need the cycle to be virtuous, so people come back and edit.
Mark: I want to talk a bit about how you think about balancing Wikipedia’s dual identity. It’s an institution, a brand, and yet it also gets the benefit of being seen as made up of thousands of individual contributors, so you get both institutional trust and creator-style trust. Do you think of Wikipedia as a brand, or as an assembly of creators?
Zack: That’s a hard question to answer. Wikipedia often gets referred to as “according to Wikipedia,” which makes it sound like “according to the New York Times,” like a single publisher’s voice, and that’s not accurate. It’s really “according to the 480 people who edited this article and reached this consensus view in the past 24 hours.” So we’re actually trying to emphasize the people more—get them from behind the page to the front of the page. The 25th anniversary campaign is entirely composed of screenshots from like 700 Wikipedia pages. That’s the platform—we’re trying to break it back into its atomic units. These are the edits. These are the statements. These are the images. And these are the people who make it happen.
Talking Heds
Politico Decoded (EXCLUSIVE): After 15 years of Politico Pro, the global politics publisher is finally launching a second subscription product. Called Decoded (not to be confused with The Verge’s Decoder), the product bills itself as “Playbook meets Dealbook” for the technology policy sector, according to Politico CEO Goli Sheikholeslami. Whereas Pro, an enterprise subscription, costs north of five figures, Decoded will stay below $1,000 annually. The demand is certainly there—the same strategists advising politicians in D.C. are now counseling AI executives in Silicon Valley—but it will be interesting to see whether Decoded acts as a conversion funnel to Pro or attracts and retains an entirely separate audience.
Heatmap Heat (EXCLUSIVE): The climate news publisher Heatmap, launched three years ago, has stayed largely beneath the radar despite setting an admirable precedent for sustainable growth among media startups. The 20-person organization is on track to bring in $5 million in revenue this year, profitably, thanks in part to the growth of its Heatmap Pro enterprise product, whose 40 or so clients are now responsible for roughly 50% of its overall revenue. In particular, the publisher has seen outsized demand for its coverage of data centers, and it has used original polling and permitting tracking to cover the patchwork of nationwide regulation governing the controversial industry.
Realtor Repast: Next month, I’ll be hosting a private lunch during Advertising Week for telecommunications leaders on Oct. 7. The event will be hosted in partnership with Realtor.com, and the conversation will center around the unique opportunity that phases of disruption—moving, specifically—present to brands. I have partnered with the folks at Realtor a few times now, and in each instance have found the subject matter—the housing market, generational wealth, the psychological heft of relocation—immensely fascinating. If you’re interested in joining me, please reach out to liz@adweek.com.
Howl Hurting (EXCLUSIVE): Last week, I scooped that the affiliate platform Howl had sold its creator business to the Taboola-owned firm Connexity. After looking further into the matter, it appears Howl had once again fallen behind on its payments, both to creators and publishers, just as it had when I unearthed these issues originally. Connexity has since paid off any outstanding debts to the creators whose business it acquired, part of a broader effort to win over their relationships with brands, but several publishers remain unpaid, according to two sources. Howl, for its part, has no plans to let these issues stop it. According to founder Li Haslett Chen, it now has its sights set on the AI space.
The Daily Mail ‘New Media’ (EXCLUSIVE): The Daily Mail officially launched its New Media division last week, an operation that will create upward of a dozen socially native series featuring in-house talent, according to Bob Marshall, the U.S. head of content at DMG Media. The division aims to mirror the success of its U.K. counterpart, which launched last year and has grown to over 100 staff and produces videos that average around 300,000 views apiece. The launch comes freighted with serious commercial ambitions: While social video currently represents roughly 20% of all direct advertising revenue at the Daily Mail, the company expects that figure to top 33% by 2027. The infrastructure is all there, but analyst Sheel Shah is skeptical about whether its copycat formats will prove to be a success in the long run.
Wired to the App: The Condé Nast publisher Wired launched a mobile app on Tuesday, making it the latest editorial brand to launch such a lifeboat as online traffic continues to sink. I have come to see the launching of a mobile app as a litmus test for how fervently leadership believes in a brand (as well as a reflection of the resources it has on hand to make such an investment). In an era of diminishing traffic, where only editorial titles with sticky, durable audiences will survive, media companies are having to make hard choices about which of their holdings have the highest likelihood of success. Congrats, Wired!
Quote/Unquote
Ryan Merkley is the chief operating officer of NPR. He joined the company in January 2025, just months before the public media organization would face dramatic cuts to the federal funding that helps it and its nationwide network of member organizations operate.
Merkley is also intimately involved with the publisher and its artificial intelligence strategy, which involves finding innovative ways to maximize engagement in the face of declining traffic, as well as negotiating with the AI firms over licensing deals. So far, NPR has not struck any content partnerships with any of the major AI companies, although it remains in active negotiations.
This interview has been edited.
Mark Stenberg: NPR is one of the most prominent major news publishers to have neither struck a content licensing partnership with nor sued an AI firm, defying the “woo and sue” approach” I have seen used so often. Why is that?
Ryan Merkey: Blocking them is on the table, but it is more complicated for us because we are a distributed network through our member stations, so we would have to block nationwide. We are also a mission-driven organization with a mandate to serve all Americans, so access is really important to us. That being said, the integrity of answer engine results is paramount, and that has not been a top priority for many of these companies.
Mark: You have said that Google used to be incredibly rigorous about ensuring it surfaced accurate results, but that focus has disappeared in the AI era. Why do you think that is?
Ryan: It is obvious that there is urgency here, but this is the first time where a massive consumer product has arrived half-baked. The public has served as the de facto quality assurance team. It could be because of the enormous [capital expenditures] it laid out to come to market.
Mark: There is an obvious tension between your desire to remain accessible and your ability to continue operating as traffic declines. When does NPR reach an inflection point and have to do something drastic?
Ryan: Every media organization is facing that dilemma. The trend line is going down; the diminishing returns of search are pretty dramatic. We are using Piano to optimize our on-site engagement, while working to more effectively draw in audiences from popular off-platform destinations, like Tiny Desk on YouTube.
Mark: Funny enough, I spoke with Zack McCune, the director of brand at Wikimedia, about this same topic. They feel a similar tension between their mission—providing access to the world’s information—and their continued ability to operate without traffic.
Ryan: I was the chief of staff at the Wikimedia Foundation for almost three years. This is the second time they have been around the block, following Google’s launch of the Knowledge Bar. That time, Google was so careful. They had a team of engineers who scraped Wikipedia in real time but would delay information on health, finance, and living people, because they wanted to ensure that people never made a consequential decision about their lives with inaccurate information. Contrast that with the way they are behaving now, and it is night and day.
Pulled Quotes
“I just haven’t had the time, with the amount of things that I set in motion …”
Tom Cruise, on why he hasn’t done an interview in 14 years
READ MORE
“We are guests of Ralph Lauren, and I am in awe of how they have made a mini Polo Bar at Arthur Ashe Stadium.”
Vogue editor in chief Chloe Malle, on her experience at the U.S. Open
READ MORE
“I haven’t resolved that conflict—if I’d like to be a populist, or highbrow. That’s why I do both.”
The New York Times’ newly appointed media columnist Michael Hirschorn, in a 2007 Observer profile
READ MORE
“By the time I got home, I had concluded, based on a multitude of examples, that the two great pillars of American culture are church and self-storage.”
New Yorker journalist David Owen, on the uniquely American phenomenon of storage facilities
READ MORE