So I looked around, and liked what I saw from Carbonite. They recently went public, and raised a decent bit of money, and probably helped enrich the lives of a number of software professionals in and around Boston.
Their software was easy to download, easy to install, nice modern user interface. But what struck me about something they did was how well they managed user expectation. Backup technology hasn't really changed much since I worked at Connected, and all the problems that Carbonite must struggle with are the same problems we struggled with. One such problem was the duration of the first backup. (First backups are "full" backups, everything else will be block-level incrimenetals, which means that unless you download or create several GB of data a day, your second, third, and etc. backups will be quick, especially compared to the first one...)
We always fought against the sales objection of "the first backup takes too long." We put significant work into speeding that process up, invented block-level checksum analytics to backup "symbolic links" to common files, to reduce the size of the first backup. We sweated this problem, and made it better by small percentages over many long years.
Carbonite took a different approach. They put, front and center - in their Buy page, in their FAQ, in their EULA, in their application - text to the effect that THE FIRST BACKUP COULD TAKE DAYS, OR EVEN A WEEK. Mine took a day. I suspect most take a day. But this is just brilliant expectation management.
Product Development Moral: If something is likely to take a day or two, tell the user it may take up to a week. Then, when it only takes a day, instead of viewing the performance as a reason not to buy, they'll be happy!
I wish I'd thought of that - maybe Connected Corp. would have made a big splash with an IPO!