The 3 Digital Gatekeepers: Agents, Authorities, Algorithms
Hidden forces shaping online discovery
Not long ago, discovering content online was a winding journey. You’d Google a subject, stumble upon a creator, check their social profiles, and maybe subscribe to their blog. For creators, there was real serendipity. Share thoughtful ideas, and with some luck, people would find you.
Today, discovery has become concentrated in two primary gateways: AI chatbots and social media feeds. AI, trained on the web, acts as an oracle - summarizing answers, surfacing recommendations, and completing tasks. Whatever time isn’t spent with AI is spent in social feeds engineered for endless engagement. Outside of these systems, unstructured discovery is disappearing.
This shift carries two significant implications:
Content and distribution are inseparable: The best creators design with distribution in mind, considering where and how their work will spread before they even begin.
The new gatekeepers are AI, algorithms, and authorities: If your work isn’t optimized for at least one, it risks invisibility.
In this article, I dig into these trends and what they mean for creators. Anyone not interested in reading the full article can find a summary here.
The Data Transforming Online Discovery
ChatGPT’s debut in November 2022 was a watershed moment that sent ripples throughout the industry. Overnight, almost every tech company found itself on the defensive and recognized that adopting AI directly into its products and operations was imperative to fend off obsolescence.
Google, the 800-pound gorilla in search, responded with AI overviews, designed to provide concise and helpful summaries in response to user queries. AI overviews proved remarkably effective, and Google actually saw a 21.64% increase in search volume from 2023 to 2024.
However, these AI overviews came with a price. A study from Pew Research found that they roughly halved click-through rates. Users who did not encounter an AI summary clicked on traditional search results in around 15% of visits, while those who encountered AI summaries did so in only 8% of visits. In 2024, 58.5% of Google searches didn’t result in a click.
While AI chatbots and search engines are gaining ground quickly, Google is still very dominant. A study by SERanking found that in 2025, AI platforms account for just 0.15% of global internet traffic (with ChatGPT in the lead), compared to 48.5% from organic search.
Thus, based on the data, it’s not the rise of ChatGPT and Perplexity that are most responsible for 37 of the top 50 news sites suffering year-over-year traffic declines. Instead, it is Google’s AI overviews as well as social media sites that are increasingly penalizing and de-prioritizing outbound links in their algorithms.
Interestingly, visitors referred by AI platforms spend 68% more time on websites than those from traditional organic search. In a world in which web browsers are becoming increasingly AI-powered (e.g. Perplexity’s Comet browser or Atlassian’s recent acquisition of the Dia browser), website traffic won’t die off. But, it will increasingly be filtered and funneled with AI agents as an initial gatekeeper.
Any website traffic that comes outside of AI is likely to originate from social media, where each successive generation is spending more time than the previous one.
Optimizing Content for AI
Answer Engine Optimization (AEO) is a burgeoning field with an increasing amount of research and frequent discoveries. I created this checklist as a comprehensive guide on how to optimize content for AI. Although it is a few months old, there is likely fresher research to add to this.
For those who want a more straightforward answer, here is a streamlined four-step approach:
Audit how AI sees your site – Use a tool like Firecrawl to understand how agents access and interpret your content, then identify gaps.
Optimize structure and schema – Add an llms.txt file, keep paragraphs to 2–4 sentences, use bullet points and lists, create a strong FAQ page, and make sure your site loads quickly.
Expand distribution – Boost visibility through PR, Reddit, Quora, and other channels where your content can gain traction. This article by Sparktoro highlights the resurgence of PR.
Track performance – Use WorkGPT’s Google Sheets integration (video below) or HubSpot’s AI analyzer to monitor how often your content gets referenced for relevant queries. I wrote about this here.
Optimizing Content for Authorities
When I speak of optimizing for authorities, I’m primarily referring to creating high-quality content that is likely to resonate with cutting-edge, deep thinkers or leaders at well-known institutions. Consider writing a piece that gets quoted by a writer at The Atlantic or referenced by a Substack or Beehiiv writer with over 50k subscribers.
This article itself is a good example: it is unlikely to make waves on social media and isn’t structured to be perfectly digestible by AI, but a thoughtful reader in my space is likely to appreciate it.
While every space is different, a few suggestions for writing content that appeals to authorities include:
Write on Subjects They Care About → Do research on the most cutting-edge subjects in your industry and add some fresh insight or perspective to the conversation.
Strong Frameworks → Create conceptual models (2x2s, taxonomies, “laws of X”) that simplify complexity. Thought leaders often re-share these as teaching tools.
Original Research or Data → If you can generate even small proprietary datasets or analyze existing ones creatively, that makes your work reference-worthy.
Visual Aids → Well-designed charts, diagrams, or infographics that make a point succinctly are highly shareable.
Take a Long-Term View → Go beyond the news cycle. Offer analysis that frames today’s events in a broader historical, cultural, or economic context.
Building relationships with authorities in your space (e.g. by offering consistently thoughtful and value-added comments to their posts) and then contacting them directly about guest posts or sponsorships is a good way to assess whether your content resonates with them. This strategy enabled me to get a shoutout from Maja Voje, a well-known authority in the GTM space.
Optimizing Content for Algorithms
While I don’t consider myself an expert on cracking social media algorithms, it’s well-known that different platforms prioritize and elevate content based on their preferred KPIs (as shown below). If you’re seeking to learn more about this, my friend Vin Clancy is a social media wiz and has written some brilliant content on how to crack algorithms.
The Content Waterfall
These paths are not separate but are intertwined. Some examples of how they can criss-cross:
Path 1: Authority → Algorithm → Agent
You share content with an authority.
The authority amplifies it on social media.
The social post is surfaced by AI agents and discovered by others.
Path 2: Agent → Algorithm → Authority
You create content designed for AI discovery.
An AI recommendation is shared on social media.
An authority encounters it in their feed, shaping their perspective.
Path 3: Algorithm → Authority → Agent
You publish content crafted to win algorithmic attention.
An authority notices and cites it in a long-form piece.
That piece becomes part of the knowledge AI references for others.
Conclusion
Discovery has narrowed. Instead of stumbling across content through search and blogs, people now find ideas through three channels: AI agents, algorithms, and authorities. That means creating content without thinking about distribution is no longer an option.
The opportunity is in learning how these paths intersect. AI surfacing content that goes viral on social, social introducing your work to authorities, and authorities shaping what AI references next. Creators who design for this loop will continue to show up in the places that matter. Those who don’t will struggle to be found at all.
If you liked this content, please click the <3 button on Substack so I know which content to double down on.
TLDR Summary
Discovery online has shifted from open-ended exploration to being funneled through three gatekeepers: AI agents, social media algorithms, and subject-matter authorities. For creators, this means content creation and distribution are inseparable - if you don’t optimize for at least one of these channels, your work risks invisibility.
Key Points
The Collapse of Discovery: Once a winding journey through blogs and search, discovery now happens almost exclusively via AI chatbots and social media feeds.
AI’s Impact on Search:
Google’s AI overviews boosted search volume by 21.6% (2023–2024).
But they cut click-through rates nearly in half - 58.5% of searches in 2024 resulted in no clicks.
AI platforms remain small (0.15% of global traffic), but users referred by them spend 68% more time on sites.
Social Media’s Role: Feeds are prioritizing engagement while de-prioritizing outbound links, further reducing organic discovery.
Optimizing for AI: Audit how agents see your site, improve schema and FAQs, expand distribution through PR and forums, and track AI-driven mentions.
Optimizing for Authorities: Publish frameworks, original research, and long-term analysis that attract citations from influential writers and institutions.
Optimizing for Algorithms: Align content with platform-specific KPIs to increase reach on social channels.
The Content Waterfall: Content flows across AI, algorithms, and authorities in interlocking loops. For example, authority → algorithm → agent, or agent → algorithm → authority.
Conclusion
Discovery has narrowed. Instead of search engines and blogs, audiences now find content through AI agents, algorithms, and authorities. For creators, this means distribution can no longer be an afterthought - it must be baked into the content itself. Those who design for the interplay between AI, social feeds, and thought leaders will remain visible. Those who don’t will fade into obscurity, no matter how strong their ideas.
If I agreed any more with this article it would suggest I paid you to write it :-) Exactly the thesis for Rocksalt.
One variant I would add to the buyer journeys list: buyer hears something on social repeatedly and explores further with AI. Then visits vendor sites that solve the problem they were exploring. I think this is a very common scenario.
Great point, James! For anyone trying to keep up with discovery, this is a must read.