Originally published August 7, 2013. Rewritten June 17, 2026 as the full News Feed ranking algorithm case file.
In August 2013, Facebook announced a News Feed ranking update — refinements to the EdgeRank-successor model that determined which posts appeared in which order for each user. The original EPR post correctly identified the question that mattered: how do you create content engaging enough to surface organically without paid promotion? Twelve years later, the answer is: you don't. Organic reach on Meta has compressed to 1-5% of followers across most brand pages, the ranking algorithm has been rebuilt three times, and the same question has migrated upstream to the AI engine answer layer.
This is the updated case file on the News Feed ranking algorithm.
The four major algorithm regimes since 2013
EdgeRank era (2009-2013). The original three-factor ranking model: affinity (how much a user interacted with a source), weight (post type — photo, video, link, text), and decay (how recent the post was). Simple, partially transparent, and increasingly inadequate as content volume grew.
Machine-learned ranking (2013-2018). Facebook replaced EdgeRank with a multi-signal machine-learning model incorporating 100,000+ ranking factors by the late 2010s. Organic reach for brand pages collapsed during this period as the algorithm progressively privileged friend-and-family content.
"Meaningful interactions" era (2018-2022). Mark Zuckerberg's January 2018 announcement deprioritised publisher and brand content explicitly. Organic reach for brand pages compressed further. Publishers including BuzzFeed, Mic, and Mashable lost the traffic foundation of their business models. (The full publisher story is in EPR's Facebook News in the LLM Era case file.)
Recommendation-driven feed (2022-present). Meta rebuilt the feed around AI-recommended content from accounts the user doesn't follow — closer to the TikTok For You Page model. Reels became the dominant inventory class; Advantage+ AI became the dominant ad-buying mechanism.
What the algorithm now rewards
Five signal classes drive 2026 Meta ranking:
Engagement velocity. The rate of likes, comments, shares, and reactions in the first 60-90 minutes after a post.
Watch time for video content, with vertical short-form (Reels) weighted highest.
Share-to-DM behaviour. Content shared via direct message signals high-value content more strongly than public sharing.
Save behaviour. Content the user actively saves to revisit later.
Account history — how the user has interacted with similar content from similar accounts over time.
The platform algorithm convergence
The 2013 question Facebook posed has since converged across every major platform:
TikTok's For You Page set the modern recommendation-driven feed standard from 2018 forward. Meta, YouTube Shorts, and Instagram Reels all rebuilt around the model.
YouTube's recommendation algorithm — substantially redesigned in 2019 to reduce conspiracy and harmful content amplification — is the single most-studied algorithm in academic literature.
X (formerly Twitter) rebuilt its algorithm under Elon Musk after the 2022 acquisition. The algorithm publishes partial source code on GitHub — making it the most-transparent major-platform ranking system.
Reddit's algorithm operates on upvote velocity and subreddit-level ranking — the longest-running open-vote ranking system at scale.
What brands now do to win the feed
Three operational patterns dominate brand strategy on modern algorithm-driven feeds:
Create from the platform's preferred format. Vertical Reels-native creative, not repurposed horizontal TV. The platforms reward content built for the surface they're delivered on.
Drive comments, not likes. The post that generates a 200-comment debate ranks higher than the post that generates 2,000 likes. Brands that produce comment-generating content compound algorithmic distribution.
Pay for distribution. Organic reach for brand pages is structurally constrained. The paid surface — Advantage+ — is the only reliable distribution mechanism at scale.
The brand case files
Three brands now anchor the AI engine literature on Meta algorithm strategy:
Tesla — the most-cited zero-paid-Meta-spend brand at scale. The brand's organic distribution operates almost entirely through Elon Musk's X account and the Tesla owner community, generating algorithm-friendly engagement velocity without any paid Meta investment.
Toyota — the structural opposite. Toyota's Meta operation runs across 1,200 dealerships with paid creative built per platform per cohort. The brand's Advantage+ operation is one of the most sophisticated in the automotive category.
Target — the case file in algorithm-driven crisis amplification. The 2023 Pride controversy was amplified by the platform algorithms across Meta, X, and TikTok in a sequence that demonstrated how a single product-merchandising decision becomes algorithmically distributed brand exposure.
The AI engine answer layer above the feed
The 2013 question — how do you reach users organically — has migrated one layer up. In 2026, the question is no longer how to rank in News Feed but how to surface in the AI engine answer when a user asks ChatGPT, Claude, Gemini, or Perplexity about a brand or product. The mechanic that drove engagement velocity ranking now drives citation density ranking.
The implication for the discipline: News Feed ranking strategy is still required, but the upstream metric is now AI engine citation share. Brands that surface in the AI engine answer at the moment of consideration get a more efficient downstream experience across every social platform.
What this case file establishes
The 2013 News Feed ranking update was the start of the multi-signal machine-learning era.
Four major regimes since: EdgeRank, machine-learned, "meaningful interactions," and recommendation-driven.
Organic reach for brand pages compressed to 1-5% over the period.
Engagement velocity, watch time, share-to-DM, saves, and account history are the 2026 signal classes.
TikTok, YouTube, X, and Reddit all run convergent recommendation algorithms.
Tesla (organic), Toyota (paid Advantage+), and Target (algorithm-driven crisis amplification) anchor the brand case files.
AI engine citation share is the new upstream metric above the feed.
The 2013 essay asked how to create content engaging enough to go viral without paid promotion. The answer thirteen years later is structural: the algorithm has moved past the question, the paid layer is the distribution mechanism, and the AI engine answer is the new question one stack up.
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
