Outbound mail traffic from the Barracuda Email Security Gateway undergoes Bayesian scoring just as inbound mail does. For details on how Bayesian scoring works, see Bayesian Analysis Inbound. Note that to 'train' your Bayesian database to most accurately determine what you or your user(s) consider to be spam or 'not spam', at least 200 messages of each kind of inbound mail (spam and 'not spam') must be identified. Users can use the Barracuda Outlook Add-in with their MS Outlook client which provides convenient buttons to click for identifying selected messages as either spam or 'not spam'. See the Barracuda Outlook Add-In Overview for details.