We have been recently asked here at SocialReport.com about adding Twitter reach and impressions to our analytics. We had all the data already so it was a no brainer - we went ahead and added it. We did however take a moment to see what these numbers actually meant as it seemed like these two measurements (reach and impressions) were widely used and accepted by many.
According to TweetReach.com - reach and impressions are defined as follows:
Simply put impressions are a total number of people that have potentially seen your message and reach is a number of unique people.
We should emphasize the "potentially" part. Love it! Hypothetical marketing! Perhaps payments for these marketing efforts should also be hypothetical :)
Twitter feed is a drive-by. If you happen to be there when the message is passing by - you will see it. If you are not, you will never even know about it. As a matter of fact even if you try to find it later - it will take you enormous amount of time. Twitter has little in terms of historical search. It is a though twitter only wants you to see "the present".
It seems that the premise for reach and impressions is coming from web impression measurement. Just like there are impressions in the web analytics world - these social metrics are trying to do the same. These are hardly analogous though. Web impressions are a result of one's physical action - clicking on the link, typing the address, etc. When someone tells you that you had 1000 impressions on the web - you know what that means - 1000 actual people (give or take a few bot hits) saw your page. When you are told that you had 1000 twitter impressions - you can't even begin to estimate what that truly means. Clearly only a fraction of these folks really saw it. How many? No one knows.
Perhaps the only way to truly figure this out is from twitter. Hopefully one day they will make Twitter Insights available to all. They certainly can tell if a tweet has been served into a user's timeline.
All other attempts at measuring impressions and reach are completely off the mark.