The Authority Trap – Why Being Worth Quoting Requires an Enterprise-Wide Shift
The Authority Trap – Why Being Worth Quoting Requires an Enterprise-Wide Shift
The Authority Trap
Why Being Worth Quoting Requires an Enterprise-Wide Shift
Issue 291, November 19, 2026
Most of us have looked up a simple recipe and had to scroll past a long story about someone’s summer in Tuscany, a dozen photos of a cutting board and several ads before finally getting to the ingredients. Or we searched for the answer to a straightforward question and landed on a page that repeated the question five times in the first paragraph without ever answering it, then asked for an email address to get to the real information. A lot of what we read online over the last twenty years was written that way, and it wasn’t really written for us. It was written for a search engine that rewarded length, repetition and the right keywords in the right places.
For more than twenty years, organizations treated their websites and much of their outside communication as an exercise in pleasing that gatekeeper. We stuffed keywords into web pages, chased rankings, tuned meta descriptions and other elements to match the signals search engines score, and cut long, careful explanations down to whatever would capture the click. Even when the terms didn’t fit a complex subject, the direction from the top was usually clear: optimize for search, get the traffic and feed the algorithm. And with a wave of the wand, the organization, its products and its services would be found by interested parties. Credibility, real expertise and primary sources were often the first things sacrificed in the scramble to be seen. A marketing department put on its armor to fight the good fight and played the game to score points.
Last week, in The Visit That Never Comes, I wrote about how AI answers are replacing the search, the click and the visit. When researchers look at what these AI systems actually pull into their answers, they find something worth paying attention to, because it may signal a much-needed shift that is already underway. The systems favor the very things the search era taught organizations to strip out: credible sources, real expertise and specific evidence. The machines, it turns out, are hungry for exactly what we spent two decades taking out of our websites, our communications and our content.
That is the authority trap. Organizations spent years building a voice designed to be found, and now they need one worth quoting, and the practices and habits that got them found are the ones now standing in the way.
It isn’t only a marketing problem. The same habit of stripping things down to whatever is quick and easy to consume runs through how information moves inside most organizations, from the people closest to the work up to the people making the decisions. And an organization that can’t hold onto depth inside its own walls is going to have a very hard time showing any depth to the outside world.
How We Learned to Write for the Crawler
It didn’t start out as a battle or a game. In the early years, learning how search engines indexed pages, how links from other sites signaled credibility and how the words on a page helped people find it felt like sensible marketing. If people were searching for what you did, you wanted to be found. I share with my students that algorithms were first created to help users sift through the vastness of the web efficiently. What we didn’t see coming was how writing for the algorithms would change us in so many ways.
Over time, the rules started shaping the work. As competition grew, the reward shifted from substance to a complete focus on optimization. Search engines rewarded volume and frequency, so organizations produced more of both. Articles were stretched or trimmed to hit an ideal word count. Headings were filled with phrases no real person would say out loud. Original reporting was copied, reworded and republished across thousands of sites until much of the web started to read exactly the same. And much of it still does. Think about how hard it is to find what you are really looking for in a search engine, and how long it takes to sift through the same short summary or set of bullets repeated on countless sites.
To fight the good fight and score points, organizations put real money into the algorithmic machinery. They hired agencies, built internal content teams whose main job was volume, and judged success by traffic dashboards full of clicks, impressions and bounce rates. I still routinely see RFPs from organizations seeking exactly the services built around those practices, even as they become less relevant and less effective.
As I wrote in September in Who Owns That Number?, measures like those go onto the dashboard and almost never come off. The trouble was what those numbers were actually measuring. They told you how many people showed up, not whether anyone trusted what they found when they got there. They measured how well an organization played the search game, not how well it understood the problems its customers were trying to solve. And because the system rewarded whoever had the loudest, most far-reaching and most repetitive megaphone, organizations kept turning up the volume, even as their audiences grew numb to the noise.
What the New Machines Reward
Now the ground seems to have moved underneath all of it. When someone asks an AI assistant about running shoes, accounting software or how to choose an association management system, the assistant doesn’t hand back ten blue links ranked by who did the best job of optimizing. It reads across what is available, weighs it and delivers an answer.
Last week I mentioned a 2024 study led by researchers at Princeton that tested what actually makes content more likely to show up in those answers. Its findings read almost like a reversal of the last twenty years of digital marketing. The researchers tried different ways of rewriting web content and measured how often AI systems used it. The changes that worked best were citing credible sources, adding quotations from people with real expertise and including specific statistics, which together improved visibility in AI answers by up to 40 percent. Keyword stuffing, the old reliable trick, did little or nothing.
The most telling finding is who gained the most. Smaller and lower-ranked websites benefited far more than the big ones. A site ranked fifth in regular search results more than doubled its visibility in AI answers when it cited its sources well, while the site ranked first actually lost ground. Under the old rules, size and budget bought visibility. Under the new ones, substance may matter more than size. And I would suggest substance has always mattered, but somewhere along the way we forgot that as we acquiesced to what we understood the rules of the game to be.
Pew Research Center found something that fits. When Google showed an AI summary at the top of the results, people clicked on a regular link about half as often as when there was no summary. And the sources cited most often in those summaries were Wikipedia, YouTube and Reddit, places where people explain things, argue about them and share what they know, rather than places built to sell.
The machine, in other words, is looking for a library, one that includes substance, not a billboard made up of five bullet points.
These are early findings, and they should be held fairly loosely for now. The systems are changing constantly, and the companies building them aren’t publishing their rules, so anyone promising a formula for getting an AI to recommend you deserves a few questions. There are some firms, agencies and individual consultants professing AI expertise when, like the rest of us, they are playing catch-up to the technology. That said, the current direction lines up with what people have always looked for in a source they trust. Show your evidence, say who you heard it from and know what you are talking about. What is new is that the gatekeeper finally seems to care about that too. Which is just so refreshing.
The Same Habit Inside the Building
The pull toward the short, the simple and the easy to consume isn’t limited to marketing departments and websites. It is a near-perfect mirror of how many leadership teams take in information inside their own organizations.
Walk into almost any boardroom or leadership meeting and you are likely to hear some version of the same request from an executive under pressure: “Give it to me in three bullet points.”
The impulse is understandable. Leaders are flooded with information. Their calendars are packed, their inboxes overflow, and the pace of decisions leaves very little room to sit with anything of substance. In that environment, the bullet point feels like a lifeline, a way of turning something complicated into something a busy person can understand and act on.
I have heard that request many times over my career, and I have pushed back on it nearly every time. I have always written in long form, and I have always believed that the drive for bullets, sound bites and search-friendly copy did real harm to how people build knowledge and come to understand something. Too often I have watched a decision get made by people who simply didn’t have all the information, because the context that mattered didn’t fit in the bullets they asked for.
But there is a trap inside that efficiency. When every hard strategic question, every source of friction in the organization and every shift in the market gets boiled down to three to five tidy bullets, a lot of what is real about the situation gets left behind. The trade-offs disappear. The ambiguity gets smoothed over. The uncomfortable details that don’t fit neatly on a slide are quietly edited out, often by managers who have learned over the years that their leaders prefer good news, simple answers and sound bites.
What reaches the top is a clean version of reality that may have very little friction left in it, and that leaves leaders more exposed than they realize. Two weeks ago I wrote about the illusion of validity, the confidence we feel when information fits together into a good story whether or not it predicts anything. Three to five bullets are about the tidiest story there is. They feel complete because everything that didn’t fit has already been removed. So leaders end up making big bets with a lot of confidence, not because they lack intelligence or judgment, but because the way information reaches them has shielded them from what they don’t know.
In many organizations, we have built an internal communication culture that looks a lot like the worst of the search era. It is built for quick consumption, stripped of the messy details and designed to avoid the hard conversations that would slow down fast decision-making.
The Mantra That Taught Us Not to Read
Behind a lot of this sits a phrase most of us have heard in a meeting, usually said with complete confidence: “People don’t read.”
It has a history. In October 1997, the usability researcher Jakob Nielsen asked how people read on the web, and his answer was, “They don’t.” In a study he ran with John Morkes, 79 percent of the people tested always scanned a new web page, and only 16 percent read it word by word. His advice followed from that. Highlight keywords, use meaningful subheadings and bulleted lists, keep one idea to a paragraph, start with the conclusion and use about half the words you would in print. For the slow, clunky screens of 1997, that was reasonable advice about how people behaved on web pages.
The problem was how widely we adopted the practice and what we did with it without realizing the consequences. An observation about how people moved through early websites, which were slow back then, became a rule for everything, from reports and board materials to strategy documents, trainings and internal emails. Then it became a belief about people themselves. We treated not reading as a human factor to design around, rather than a capability worth building. Instead of teaching people that reading and understanding are necessary for sound decisions, organizations built everything around the assumption that they wouldn’t read, and in the process taught them they didn’t need to.
That has consequences well beyond marketing. In October I wrote about the disappearing middle, the risk of automating away the work that taught people how to think. Reading a long, careful argument and working out what it means is that kind of work. When every document is designed as if no one will read it, people lose the practice, the muscle, if it was ever there, starts to atrophy, and leaders lose the habit of expecting it.
What Gets Lost on the Way Up
Adrian Wolfberg and I spent a good part of our recent two-part conversation on The Human Factor Podcast on exactly this. Adrian spent nearly forty years as an intelligence officer, including briefing senior leaders in the Pentagon up through the Chairman of the Joint Chiefs of Staff, and his career has been built around one lesson: knowledge that can’t cross the boundary between the person who has it and the person who decides with it might as well not exist.
We talked about how a piece of information starts at the edge of an organization. Someone on a sales call hears a customer hesitate. An analyst notices a number that doesn’t fit. A nurse sees a pattern on the floor that the chart doesn’t show. As that information moves toward the person who decides, every hand it passes through summarizes it, tidies it up and drops the caveats, because the caveats make the slide messy. By the time it reaches the top of the building, the evidence is gone and only the conclusion is left. James March and Herbert Simon gave that a name in 1958: uncertainty absorption, and it remains one of the oldest findings in organizational research. We have done remarkably little about it since.
One of the most painful examples comes from the space shuttle Columbia, which broke apart on its return to Earth in February 2003, killing all seven astronauts aboard. The board that investigated the accident wrote that it was surprised, again and again, to receive presentation slides from NASA officials in place of technical reports, and it called the endemic use of PowerPoint briefing slides instead of technical papers an illustration of the problems with how technical information was communicated. About one key chart on the risk of foam striking the shuttle’s wing, the board wrote that “the uncertainties and assumptions that signaled danger dropped out of the information chain,” and that the choice of headings and the size of the bullets “served to highlight what management already believed.” Most of our decisions will never carry stakes like that. But the mechanism is the same one at work in a quarterly business review.
AI is now part of that chain, and it can make things better or a lot worse. In the episode, I described a senior leader who took months of careful, grounded strategy work, ran it through a language model and sent the polished result up to the leadership team without checking it, and then couldn’t answer a single question that followed. Adrian’s view was that a machine in the chain makes the telephone game we discussed in the episode so much worse, not better, and that the questions he learned to ask as a naval officer matter more now than ever. Where did you get this information? How did you interpret it? What assumptions did you make between the original material and what you are telling me now? And what would I need to see to prove it wrong?
He also drew a distinction between apparent confidence and earned confidence. There is nothing wrong with confidence, and teams need it from the people leading them. The problem comes when the confidence is empty, when it rests on a polished summary rather than on understanding the problem. A three-to-five-bullet briefing produces apparent confidence very efficiently. Earned confidence takes longer, and it usually requires someone to read the longer version.
Authority Starts Inside
This is where the outside and the inside of an organization turn out to be the same problem.
Outside, the evidence suggests AI systems reward depth, evidence and real expertise, the earned kind of authority. Inside, many organizations reward compression, the clean conclusion and the confident slide. Those aren’t two separate issues. They are the same muscle being used in two directions. An organization that only lets conclusions through internally will tend to publish conclusions externally, confident and polished and empty of the evidence underneath them. That is apparent confidence turned into marketing, and it is exactly what the new machines appear to pass over.
There is a second connection, and it may matter even more than one might think. The people inside an organization who hold its real expertise are very often the same people whose caveats get dropped on the way up. The engineer who knows why the product fails in certain conditions, the membership coordinator who hears every week why people leave, the analyst who noticed the number that didn’t fit. Their knowledge is the organization’s authority. And in most organizations it stays inside their heads, filtered out of the briefing and absent from anything the public ever sees. Earlier this year I wrote about structural silence, the ways organizations train people not to speak. Silence has a cost inside the building, and now it has a cost outside of it too.
An organization’s authority is also no longer defined only by its website or its search ranking. It is increasingly shaped by the whole public record around it. That includes what its leaders say in keynotes, on podcasts and in white papers, what its experts publish, and what its customers and employees say in reviews and forums. Remember, AI is looking for the library, and it’s a library it can build from the sources of information available. If the official message is safe, focus-grouped and polished while the real expertise lives somewhere no one, including AI, can see, the record that AI systems and curious people draw on will be thin, and someone else’s voice will fill the gap when a question is asked.
From the Megaphone to a Body of Work
So what does it take to move from an organization built to be found to one worth quoting? It is less about a new set of tactics and more about a shift that reaches well past marketing, into how leaders take in information and what they choose to put their names on.
It starts with an honest look at your public footprint. Ask the AI tools your customers use about the problems your organization solves and see whether you show up and what gets said. Then look past the website at what your leaders actually contribute in public. Are they offering real insight from their experience, or repeating the same safe lines that could have come from any organization in your field? The move is from broadcasting messages to contributing knowledge, and that usually means putting the people with real expertise out front, including some who have never before been asked to take on the responsibility.
It also means making room for depth again, inside the building first. Transformation can’t be bullet-pointed. Whether the subject is governance, culture change or a shift in the market, complex problems need longer, well-reasoned thinking, and leaders signal what matters by what they ask for. Amazon is one well-known example. Jeff Bezos did away with slide presentations in senior meetings years ago, replacing them with six-page narrative memos that everyone reads silently at the start of the meeting. In a 2004 email, he explained why, writing that “the narrative structure of a good memo forces better thought and better understanding of what’s more important than what.” Not every organization needs to adopt that practice. The point is that full sentences make gaps visible in a way bullets don’t, and asking Adrian’s four questions of every summary, whether a person or a machine wrote it, is a practice and a habit any leader can start next week.
And it means connecting what you know to what you publish. Podcasts, newsletters, research, presentations and the knowledge that lives with your experts shouldn’t sit in separate silos as one-off campaigns. They should add up to a connected body of work that cites its sources, builds on itself over time and says something valuable, specific and with context. When an AI system, a prospective member or a curious customer looks into your field, that body of work is what they will find, or not find.
None of this happens through a single initiative, a new strategic hire, a new consultant or a new agency contract. It’s a change in what the organization values and rewards, from the boardroom to the website, and that is why it reaches well beyond the marketing department.
Back to the Recipe Page
Go back to that recipe page for a moment. Someone wanted to know how to make something new for dinner, and they had to dig through a page written for a machine to find the answer. Now they ask an AI, and the AI goes looking for a source that actually knows how to cook. It delivers the list of ingredients and the steps to take, and even offers other information to consider. Twenty years of pages built for the crawler won’t count for much in that search, because so much gets stripped away or disregarded. That means the user, the customer or the student never sees what was left out. Even a well-sourced summary is still a summary, and what it leaves out depends on what it had to draw from.
The organizations that hold up best in the years ahead won’t be the ones that find the newest optimization trick or shrink their strategy into fewer bullet points. They will be the ones that rediscover the value of real information, respect their sources and listen to the people who know the most.
There is also reason for some hope. If the summaries people increasingly rely on are built from sourced authority and show where they came from, they may help bring back something we lost along the way: the habit of asking where a conclusion, a piece of information or a set of facts actually comes from. Right now, too many leaders pick up a sound bite from almost anywhere, a headline, a post, a conversation in a conference hallway, and repeat it in the next meeting or the next conversation as fact. A summary that carries its sources with it gives leaders something solid to stand on, and a reason to go read the longer version when the decision really matters. Whether things go that way will depend a great deal on what leaders, and really all of us as a society, choose to ask for.
When an AI looks across your field, or a leader inside your organization goes looking for the real story behind a hard problem, what do they find?
A polished conclusion with the evidence stripped away, or something actually worth quoting?
And who in your organization knows the most and is heard the least?
Related Reading
The Visit That Never Comes (Issue 290). Last week, on how AI answers are replacing the search, the click and the visit, and the larger transformations behind that shift.
How Sure Are You? (Issue 289). On the illusion of validity, and why a tidy story feels more certain than it should.
Structural Silence. Why organizations train people not to speak, and what never reaches the top as a result.
Who Owns That Number? Why organizations can add a measure and never subtract one, including the traffic dashboards built for the search era.
Smart Marketing Is a Mirror, Not a Megaphone (Issue 225). On marketing that reflects the audience rather than broadcasting at it.
The Disappearing Middle (Issue 284). What happens when we automate the work that taught people how to think.
The Importance of Language in Organizational Transformation (Issue 157). From 2024, on why shared language matters so much to how change is understood.
On The Human Factor Podcast
Who Leads When AI Thinks?, Part 1, Season 3 Episode 039. With Adrian Wolfberg, on uncertainty absorption, the telephone game with a machine in the chain, and apparent versus earned confidence.
Who Leads When AI Thinks?, Part 2, Season 3 Episode 040. On why a problem needs to be understood before it gets an owner or a tool, and judgment as a practice rather than a trait.
Research Sources
1. Aggarwal, Pranjal, Murahari, Vishvak, Rajpurohit, Tanmay, Kalyan, Ashwin, Narasimhan, Karthik, and Deshpande, Ameet (2024). GEO: Generative Engine Optimization. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, August 2024. Source of the finding that citing sources, adding quotations and adding statistics improved visibility in AI answers by up to 40 percent, the weak result for keyword stuffing, and the larger gains for lower-ranked sites, including the fifth-ranked site’s visibility more than doubling while the top-ranked site’s fell.
2. Pew Research Center (2025). Google Users Are Less Likely to Click on Links When an AI Summary Appears in the Results. July 22, 2025. Source of the 8 percent and 15 percent click rates and of Wikipedia, YouTube and Reddit as the most cited sources.
3. March, James G., and Simon, Herbert A. (1958). Organizations. Wiley. Source of uncertainty absorption.
4. Columbia Accident Investigation Board (2003). Report, Volume 1. NASA and Government Printing Office, August 2003, pages 191 and 201, as quoted by Edward Tufte in Columbia Accident Investigation Board: The Boeing PowerPoint Slide. Source of the board’s statements on slides in place of technical reports and on uncertainties and assumptions dropping out of the information chain.
5. Bezos, Jeff (2018). 2017 Letter to Amazon Shareholders, as reported in What Jeff Bezos Learned From Requiring 6-Page Memos at Amazon, CNBC, April 23, 2018. Source of the narrative memos, the silent reading at the start of meetings and the 2004 email quotation.
6. Wolfberg, Adrian. Who Leads When AI Thinks? How Leaders Frame Decisions and Exercise Judgment. Springer. With The Human Factor Podcast, Season 3, Episodes 039 and 040. Source of the knowledge boundary lesson, the four questions and the distinction between apparent and earned confidence.
7. Nielsen, Jakob (1997). How Users Read on the Web. Alertbox, October 1, 1997. Source of the “They don’t” answer, the study with John Morkes in which 79 percent of test users always scanned and 16 percent read word by word, and the advice to use bulleted lists, put the conclusion first and use about half the words.
Connect With Us
What leadership challenges are shaping your decisions right now? Share your experiences and join the conversation.
Go Deeper: Human Factor Podcast
From resistance and identity to the frameworks that help leaders navigate transformation. Available wherever you listen to or watch podcasts.
Kevin Novak
Kevin Novak is the Founder & CEO of 2040 Digital, a professor of digital strategy and organizational transformation, and author of The Truth About Transformation. He is the creator of the Human Factor Method™, a framework that integrates psychology, identity, and behavior into how organizations navigate change. Kevin publishes the long-running Ideas & Innovations newsletter, hosts the Human Factor Podcast, and advises executives, associations, and global organizations on strategy, transformation, and the human dynamics that determine success or failure.
