Free Diagnosis
Home / Blog / The Feed’s 67% Share Boost: Why Owned Chan
Platform Risk

The Feed’s 67% Share Boost: Why Owned Channels Are Your Only Defense

The algorithm doesn’t care about your expertise—it cares about outrage. Here’s why the smartest operators are building their own distribution instead of fighting the feed.

P5
Omar Catlin · Project5Pi
July 13, 2026 · 6 min read
OUTRAGE MULTIPLIER
67%
Each out-group word raised the odds of a share by about 67%—analysis of 2.7M posts (Rathje, Van Bavel & van der Linden, PNAS 2021).

I’m looking at a dataset that makes the platform’s incentives brutally clear—and it’s not on your side.

67%
increase in share odds per out-group word (Rathje et al., PNAS 2021)
as often for content attacking an out-group (Rathje et al., 2021)
689,003
unwitting subjects in a feed-emotion experiment (Kramer et al., PNAS 2014)

The 67% Share Boost: What the Feed Actually Rewards

An analysis of 2.7 million posts found that content attacking a political out-group was shared about twice as often as everything else (Rathje et al., PNAS 2021).

Every single out-group word you add? That raises the odds of a share by about 67%. The platform’s algorithm looks at that signal and says, more of this. Quality and usefulness don’t even enter the equation.

THE ALGORITHM IS NOT YOUR FRIEND

The feed is designed to boost engagement, not accuracy. The result? A 67% higher chance of sharing for every out-group word you include—and visibility that’s controlled by the platform, not you.

When 689,003 People’s Emotions Were Shifted Without Their Knowledge

In 2014, researchers tweaked the emotional content of the feed for 689,003 users—and successfully shifted their emotions (Kramer et al., PNAS 2014).

That’s the clearest proof we have: the platform, not you, controls what your audience sees and feels. Your carefully crafted content can disappear in an instant because the feed is engineered for engagement loops, not your business outcomes.

WHAT WE DON’T KNOW YET

The 2021 study doesn’t track whether outrage-driven shares lead to long-term audience value or sales. And the 2014 experiment is a decade old. But the principle—that the platform can and will manipulate visibility to its own ends—remains unchallenged.

Owned Channels: The Operator’s Escape Hatch

The only way to remove the feed as a gatekeeper is to build distribution you control. Email lists, private communities, direct messaging—channels where the algorithm doesn’t get a vote.

This isn’t about abandoning social platforms. It’s about using them to funnel your most valuable audience into spaces where you set the rules and the relationship is direct. Owned distribution is the resilience play.

From Omar's desk

I’ve brokered content deals where a single viral post brought in thousands of new followers—only to watch the platform tweak the algorithm and make that audience unreachable. The only channels that kept paying off were the ones where we could talk directly to people who had opted in. Owned audiences aren’t just safer; they’re the only real asset an operator builds.

Questions people ask

Does optimizing for the feed help at all?

It might give you a temporary spike, but the platform can and does change what qualifies as ‘feed-worthy’ without warning. You’re renting space, not owning it.

What’s the first step in building owned distribution?

Start with one primary channel—email or a community—and redirect your best audience there, even if it means smaller initial reach. Consistency beats scale in the early stages.

Isn’t this only for big brands with huge budgets?

No. Even a small, engaged list gives you more predictable visibility than platform dependence. Many operators start with a simple newsletter before anything else.

Stop Fighting the Feed. Start Building Your Own Distribution.

Download our free operator’s checklist to audit your current channel mix and identify the owned assets that will give you true audience resilience.

Start My Free Diagnosis →
Sources: PNAS 2021 (Rathje/Van Bavel/van der Linden); PNAS 2014 (Kramer et al.)
Project5Pi does not claim these formats improve search rankings; this article describes agent/AI visibility, a separate capability.