Facebook says 10% or ~207M accounts are estimated to be duplicates, up from 6% and fake accounts total ~60M, up to 2-3% from 1%, as detection methods improve
- Facebook quietly increased its number of estimated duplicate accounts from 6% to 10%. — Estimated fake accounts were raised to 2-3% from 1%. Tweets: @alexeheath and @shiraovide . Thanks: @stevetweedie Tweets: Alex Heath / @alexeheath : There are now roughly 267 million duplicate/fake Facebook accounts http://www.businessinsider.com/ ... Shira Ovide / @shiraovide : Hmmm. Facebook now thinks 10% of its monthly users are people/organizations with more than one account. Up from 6%. Thanks: @stevetweedie Expand More For Next 3 Unexpand More For Next 3
Context & Ripple Effects
In late 2017 Facebook quietly revised its own integrity math upward: estimated duplicates went from 6% to 10% of monthly users and estimated fakes from 1% to 2-3%, with the company attributing the jump to better detection rather than a sudden influx. The revision matters because these percentages sit inside the user counts advertisers buy against.
The years after this disclosure show the same problem at industrial scale: Facebook reported deleting 1.3B fake accounts in six months by late 2018, banned 2.19B in Q1 2019 alone, and by 2020 credited an ML classifier with 6.6B takedowns in a year. A 2019 analysis also found the active-fake figures in filings ran well below the takedown numbers, keeping the true base rate contested.
First-order effects
- Advertisers paying on Facebook's audience metrics are now billed against a base that includes roughly 207M duplicates and ~60M fakes — about 267M accounts that represent no unique human reach.
- Facebook's own disclosure hands critics and auditors a documented gap between its headline user counts and its real-user counts, straight from its filings.
Second-order effects
- Fake-account creation scales faster than deletion — the pattern visible from the 2017 estimates through the multi-billion takedown reports — forcing Facebook into a permanent arms race built on automated detection rather than manual review.
- Rival platforms face the same measurement question, since ad budgets priced on verified-human reach put pressure on every network's reporting methodology.
Third-order effects
- If the pattern holds, platform integrity becomes a disclosed-metrics business — quarterly takedown counts and estimate revisions as standard filing items — inviting regulator scrutiny of how user numbers underpin ad pricing.
- The durable endgame is proving each account maps to one person, which is why verification and proof-of-personhood approaches keep resurfacing across the industry.
The trend: Social platforms are shifting from periodic self-reported estimates of fake activity to industrial-scale automated takedowns as the primary integrity metric, with the underlying true-user base staying contested.