Meta weighs turning Facebook into a TikTok-style feed

Meta Platforms (META) is weighing a fundamental reorientation of Facebook’s core product, according to an internal memo obtained by The Verge — one that would abandon the social-graph model that has defined the platform since Mark Zuckerberg launched it in 2004 and replace it with an algorithmically driven content feed resembling TikTok’s For You Page.

The document, a six-page strategic overview distributed in April 2022 by Facebook App chief Tom Alison, frames the shift as building a “discovery engine” — a system that surfaces content users are likely to engage with regardless of whether the person or Page producing it is already in their network. The memo does not mention TikTok by name once, but the competitive target is unmistakable. At stake is the basic bargain Facebook has offered for two decades: you connect with people you know, and the feed reflects that social graph. Replacing connection-based curation with pure behavioral prediction would affect more than three billion monthly users and reshape the economics of digital advertising and publishing simultaneously.

From social graph to signal-driven ranking

The social graph — the web of friendship and follow connections that has always determined what Facebook users see — is the foundational architecture that distinguished the platform from earlier, interest-based web communities. Alison’s memo proposes that Facebook’s Home feed “balance both connected content and unconnected content,” specifically prioritizing content from outside a user’s network when the AI judges it more engaging. The Verge quoted Alison describing a feed that highlights interesting content “regardless of whether it was produced by someone you’re connected to or not.”

The internal reaction was not uniformly enthusiastic. Growth Rocket, citing The Verge’s reporting, noted that one Facebook employee warned in written feedback:

“I think there’s a real risk in this approach that we lose focus on our core differentiation (the social graph and human choice) in favor of chasing short-term interests and trends.” — Facebook employee, as reported by The Verge

Alison reportedly told stakeholders the platform would continue surfacing shareable content from friends, but would lean increasingly on the discovery engine for volume. That hedge has not fully held: by 2025, AI-recommended content — posts and videos from accounts users do not follow — accounts for more than 40 percent of the average Facebook News Feed, according to data tracked by Hashmeta. On Instagram, Meta’s other flagship app, the figure has already reached 50 percent, according to Yahoo Finance’s reporting on the company’s internal strategy documents.

Meta’s incremental approach to this transition is deliberate. The company first tested recommendation-driven ranking inside Reels, the short-form video product it launched on Instagram in 2020 to compete directly with TikTok. Inserting algorithmic content into a new feature carries less risk than rewiring the main feed — a distinction Yahoo Finance’s coverage of the rollout described as a key internal debate. Once Reels demonstrated that users would watch algorithmically recommended video without abandoning the app, Meta extended outside-content recommendations to the broader Facebook feed beginning in late 2021.

What this means for users, advertisers, and publishers

For ordinary users, the practical effect of a full pivot would be a feed that looks familiar but operates differently. Friends’ updates and family photos would not disappear, but they would compete against a much larger pool of content — viral clips, creator videos, and brand posts that Meta’s ranking model predicts will hold attention. The experience would shift from a curated social timeline toward a passive, scroll-optimized entertainment surface, matching the posture TikTok has normalized in the sub-35 demographic.

Advertisers would face a more complicated recalibration. Facebook’s ad targeting has historically relied on the social graph’s rich first-party signals: who you know, what groups you join, what Pages you follow. A discovery-engine model trained primarily on behavioral engagement metrics — watch time, replays, shares — generates a different signal set. It may broaden reach for awareness campaigns but dilute the precision that has made Facebook’s ad platform, still Meta’s primary revenue source, attractive for direct-response marketing. It is unclear how Meta plans to preserve targeting fidelity under the new model; the company has not publicly addressed that question.

Publishers face the sharpest exposure. Facebook referral traffic to news and media websites has declined steadily since Meta began deprioritizing link-based posts in its feed around 2018. A full shift to a TikTok-style, video-first discovery model would likely accelerate that trend, as native video content designed for on-platform consumption competes directly against outbound links in the ranking algorithm. Social Media Today noted that the strategic overview from Alison signals “major changes coming to your Facebook feed,” but the company has not specified what, if any, concessions to publishers or news organizations are included in its plans.

The competitive pressure behind the pivot

Tom Alison told The Verge in 2022 that he saw “TikTok chasing Facebook,” not the reverse. By 2025, that framing had become difficult to sustain. According to What’s Trending, which reviewed Meta’s own user data, the company recorded a net loss of 20 million users across its family of apps in the first quarter of 2025 — a figure that underscores the urgency behind a strategy that, only three years earlier, its own app chief characterized as unnecessary.

TikTok’s For You Page (FYP) — its core recommendation feed, powered by a behavioral ranking model that learns from watch time, replays, and interaction patterns rather than social connections — has reset user expectations about what a feed should feel like. Younger cohorts in particular have shown lower tolerance for the friction of a social-graph model, where feed quality depends on whether your specific friends happen to post interesting content on any given day. An interest-graph or engagement-graph model sidesteps that dependency entirely.

Meta has not confirmed a timeline for any formal product change, and the company has not publicly endorsed the specific framing in the Alison memo as current policy. It is unclear whether the six-page document represents a live strategic plan or an internal position paper that has since been revised or shelved.

Social Media Today, Tubefilter, Yahoo Finance, and Growth Rocket provided additional reporting and context on the internal memo and Meta’s broader algorithm strategy. The Verge’s original reporting on the Alison document remains the primary sourced account of the company’s internal deliberations.

The next visible milestone will be Meta’s Q2 2025 earnings call, where analysts are expected to press executives on user-retention trends and whether any formal feed redesign is planned for the fiscal year’s second half.