The Interest Graph Has Replaced the Follower Graph
The Interest Graph Has Replaced the Follower Graph — And Most Content Strategies Haven’t Caught Up
The old rule of content marketing was simple: build an audience, publish consistently, and your audience sees what you post. That rule no longer holds on any major platform. Distribution today is decided by an interest graph, not a follower graph, and the marketers still optimising for follower count are quietly losing reach to strangers who happen to share a topic.
This shift didn’t arrive as an announcement. It arrived as a series of quiet engineering decisions, made on separate platforms, in separate years, that all converged on the same principle: rank content by what a specific piece of content is about, and by how a stranger responds to it, rather than by who already chose to follow the account that made it. If you run a content operation — a blog, a brand page, a founder’s personal profile — this single change rewrites almost everything you thought you knew about growth.
Where the shift started
TikTok is usually credited as the platform that broke the old model first. Rather than surfacing video based on a user’s follow list, its For You feed ranks primarily on watch behaviour — what a person watches, and for how long — which is why a brand-new account can reach millions of strangers before it has a single follower. Sprout Social’s 2026 breakdown of the TikTok algorithm confirms the platform has been explicit that follower count is not a ranking input, and that this behaviour-first model has only become more entrenched as the platform has matured.
Meta followed the same logic on Facebook and Instagram. A 2026 review of cross-platform algorithm data found that more than half of the average Facebook feed now comes from accounts a user has never followed, and that automated recommendation systems now determine over 80% of what appears across major social platforms, according to a 2026 cross-platform algorithm statistics review.
The pattern in numbers: follower count is confirmed as a non-factor in TikTok’s ranking system, more than 50% of the average Facebook feed is unfollowed content, and automated systems now decide over 80% of what users see across major social platforms.
LinkedIn’s turn — and why it matters more for B2B marketers
LinkedIn was the last major holdout, and its shift was arguably the most consequential for anyone doing professional or B2B content. On 12 March 2026, LinkedIn rolled out 360Brew, a 150-billion-parameter model built to read a post the way a human editor would rather than matching keywords, replacing the patchwork of ranking systems the platform had run for over a decade, as detailed in this 2026 explainer on 360Brew.
The practical effect shows up clearly in Richard van der Blom’s Algorithm Insights Report, built from an analysis of well over a million LinkedIn posts. Creators who published consistently within a narrow, declared topic saw their share of platform-wide reach roughly double, moving from about 15% to 31% since 2022. Creators who published across many unrelated topics saw their share collapse from 57% to 28% over the same period. That is a near-complete reversal of who gets seen.
Reach concentration, 2022–2026: focused creators’ share of reach rose from 15% to 31%; broad, unfocused creators’ share fell from 57% to 28% — a reversal driven by topic-authority ranking rather than audience size, per this analysis of LinkedIn’s topic-authority data.
Follower count and reach have effectively decoupled. An account with 8,000 tightly focused followers can now out-distribute one with 80,000 scattered ones, because the platform is inferring topic authority from headline, publishing history, and consistency — not counting subscribers, as this breakdown of the relationship-to-interest graph shift lays out.
Search followed the same road
The same behaviour-over-following logic has reshaped search and discovery. Google’s AI Overviews and AI answer engines now decide which brands get cited before a click ever happens. Conductor’s 2026 AEO/GEO benchmarks report, based on an analysis of 3.3 billion sessions, found that direct AI referral traffic sits at just 1.08% of total web traffic — small in volume, but outsized in influence, since it is deciding who gets cited before a searcher ever lands on a page.
Why this breaks the old content marketing playbook
Content marketing, as an industry, was built on one quiet assumption: publish consistently, grow a following, and your following will see what you publish. That assumption powered two decades of “post more” advice. It’s the assumption that just broke.
The interest graph runs on a different premise entirely. It matches your content to strangers who share a topic, based on what the platform’s model infers you’re actually about — not who already subscribed to hear from you. The metric that used to matter, follower count, has been replaced by non-follower reach, dwell time, and saves, because those are the only signals that prove a stranger who had never heard of you actually stayed.
Content marketing isn’t dying. The assumption it was quietly built on is.
What actually works once you’ve been matched
Getting matched to the right stranger by an interest graph is necessary, but it isn’t sufficient. Once a platform hands your post to someone who has never heard of you, something else decides whether they stay for three seconds or sixty — and that something is not the topic tag. It’s what I call the human signal: the specific story, the opinion that costs the writer something, the detail that proves they were actually there for the mistake and the years nobody was watching. A language model, by design, predicts the average; it won’t reliably produce that.
Put simply: the interest graph is the door. The human signal is the reason anyone stays in the room once they’ve walked through it.
What this means for your content calendar
Three practical shifts follow from all of this, and they apply whether you’re running a personal brand, a SaaS blog, or an e-commerce content hub:
1. Declare a lane, publicly and repeatedly. Platforms are now inferring topic authority from your headline, bio, and publishing history. A scattered content calendar reads as noise to the model that decides your distribution, regardless of how good any individual piece is.
2. Optimise for the signals that prove a stranger stayed. Non-follower reach, dwell time, and saves matter more now than raw follower growth. If your analytics dashboard still leads with follower count, it’s measuring the wrong game.
3. Build proof into every piece, not just information. A sharp observation, a reusable framework, the named emotional tension underneath the topic, and real proof — research, data, or lived experience — are what separate a “territory” you own from generic content that’s replaceable by the next AI-generated draft.
The bottom line
The platforms didn’t kill content marketing. They killed the assumption that publishing volume alone would find an audience. The moat was never the publishing cadence — it was always the specific human doing the publishing, and for the first time, the ranking systems across TikTok, Meta, LinkedIn, and Google’s AI Overviews are all quietly rewarding exactly that.
Frequently asked questions
What is the “interest graph” in content marketing?
It’s a distribution model where platforms rank and surface content based on topic relevance and behavioural signals — what people actually watch, read, or save — rather than on who follows an account. It replaces the older “follower graph,” where an audience only saw content from accounts they had chosen to subscribe to.
Does follower count still matter at all?
It still matters for direct reach to an existing audience and for social proof, but it’s no longer the primary driver of distribution to new people. On LinkedIn specifically, Richard van der Blom’s research shows accounts with smaller, focused followings can now out-distribute much larger, unfocused ones.
What is LinkedIn’s 360Brew and why does it matter for marketers?
360Brew is the 150-billion-parameter AI ranking model LinkedIn rolled out on 12 March 2026 to evaluate posts more like a human editor would, replacing its older, more mechanical ranking systems. It puts far more weight on topic consistency and declared expertise than previous systems did, per this 2026 explainer.
Is AI search traffic (AI Overviews, ChatGPT, etc.) actually significant yet?
In direct volume, not really — Conductor’s 2026 benchmark study put AI referral traffic at just 1.08% of the 3.3 billion sessions it analysed. But its influence is outsized because these systems decide which brands get cited before a searcher ever clicks through, making early topic authority increasingly important for long-term visibility.
Should a brand narrow its content to a single topic?
The data favours declared, consistent topic authority over scattered publishing — reach share for focused creators roughly doubled while it collapsed for broad ones, according to LinkedIn algorithm research. That said, depth within a topic and genuine range aren’t mutually exclusive; the key is that the platform can clearly infer what you’re known for, even if your content explores that territory from multiple angles.
What replaces “post more, post consistently” as a growth strategy?
A structure built around proof and specificity: a sharp, original observation, a reusable framework, a clearly named emotional tension, real proof (data, research, or lived experience), and a version of the message adapted to each platform’s format. Content that skips these layers reads as generic and easily replaceable.
What is “human signal” and why does it matter more now?
It refers to the specific, personally-costly detail in a piece of content — the real story, the opinion with a stake attached — that proves a human with direct experience wrote it. Because AI models are built to predict the average, human signal is difficult for them to replicate, making it one of the few durable differentiators once a piece of content has already been matched to the right audience by the interest graph.
Does this shift apply outside social media, to blogs and SEO content too?
Yes. The same topic-authority logic underlies how AI Overviews and answer engines decide which sources to cite, and search engines increasingly reward sites with clear, consistent subject-matter authority over ones that publish broadly across unrelated topics.