If you look at the underlying data, described in the linked NiemanReports article, it shows that Twitter-sourced traffic dropped by over 50%. It's just that NPR

If you look at the underlying data, described in the linked NiemanReports article, it shows that Twitter-sourced traffic dropped by over 50%. It's just that NPR's Twitter-sourced traffic only comprised 2% of their total traffic before the change, so getting rid of their account only dropped it by 1% overall (though that was 50% of their total Twitter-sourced traffic). So it might not have made much of an impact for NPR (though no publisher wants to lose 1% of traffic), but for a publisher with more traffic coming from Twitter shouldn't nuke their account, thinking that it won't matter for them either.