To anyone sending emails (including posterous), I'd recommend going beyond A/B tests. You should know the optimal time to contact each user.
For new users you can intentionally send emails at various times (and days of the week). Then you can track a time-specific clickthru rate for each user. Eventually you'll converge on one or more optimal times/days. You can accomplish this with a machine learning algorithm or you can simply track the timestamp of each user's last N clicks and use that to decide when to email them. Precision isn't of the essence here, but I've found that clickthru rates can be dramatically affected by picking the right time of day and day of week.
You need an awful lot of clicks to get time-specific CTR, and that strikes me as unlikely with most services, but it takes zero clicks to identify one time the user is at the computer ("the anniversary hour:minute of their signup") and one click to identify a time at which they open the email (the time at which they opened any email from you, causing a download of your web bug, or clicked from the email to your website).
Silverpop offers this as a feature, though I think the algorithm is pretty naive. Basically, it aims to send around the time you normally open messages.
Comments
To anyone sending emails (including posterous), I'd recommend going beyond A/B tests. You should know the optimal time to contact each user.
For new users you can intentionally send emails at various times (and days of the week). Then you can track a time-specific clickthru rate for each user. Eventually you'll converge on one or more optimal times/days. You can accomplish this with a machine learning algorithm or you can simply track the timestamp of each user's last N clicks and use that to decide when to email them. Precision isn't of the essence here, but I've found that clickthru rates can be dramatically affected by picking the right time of day and day of week.
You need an awful lot of clicks to get time-specific CTR, and that strikes me as unlikely with most services, but it takes zero clicks to identify one time the user is at the computer ("the anniversary hour:minute of their signup") and one click to identify a time at which they open the email (the time at which they opened any email from you, causing a download of your web bug, or clicked from the email to your website).
Silverpop offers this as a feature, though I think the algorithm is pretty naive. Basically, it aims to send around the time you normally open messages.