I used to use a system like this for my software, and unfortunately I found that it did a terrible job of educating users. People would work through the entire tutorial (~5 minutes), get to the end, and then email/call me with questions that were covered in the tutorial.
I switched to just showing a few short videos demonstrating things and it works much better. It's disappointing because I was really excited about the potential of an interactive guide (and I spent a long time building it), but I can't argue with the numbers. My guess is that users were so focused on completing the steps that they didn't bother actually looking at the software, so they didn't retain anything. I can confirm that this is what happened with me when I clicked through to the link above. I clicked through all of the bubbles and finished the tutorial without reading a single word of text on the actual page.
Great feedback, I don't know what your tour looked like and what the level of text was but I would hope that by keeping the individual points very simple and visually attractive they would stand a higher chance of being read?
The guide itself was definitely being read. I built it directly into the app, so users could only advance the guide by completing the step described (like, they'd have to actually create a new contact before the guide would advance). A surprisingly high number of people completed the entire thing (~45% if I remember correctly) but very few of those people seemed to actually retain anything. We received way more support requests, and the conversion from trial to paying customers decreased (relative to just showing a video).
I don't mean to suggest that interactive tutorials can't ever work, but they were a huge failure for me. The only upside was that several people contacted us specifically to mention how awesome they thought the tutorial was. However, most of those people seemed to be other software developers that were interested in doing something similar. They didn't seem to actually be any more likely to subscribe.