Retrospective, arbitrary construction of funnels. If you need it and you know you need it, their implementation is jawdropping.

If you are tracking a single funnel which doesn't change much, that probably isn't that spiffy for you. For example, you configure Opens Trial Page, Trial Signup, Opens Plan Page, Signs Up For Plan as a funnel once and you're done.

However, if you're doing more sophisticated things, it can be a lifesaver to not have to tell the engineering team "Yo, we're reconceptualizing our funnels: you guys now have to do annoying modifications all over the site." Lets say you have a Facebook game with a Valentine's Day promotion and you want to see whether people who sent their bonnie lass a cow to show their affection are better or worse customers two months later. No problem, with a few clicks on the interface you can whip up a funnel that tracks purchasing the pink cow as a stage and then watch conversions after that. You don't have to tell engineering about it, since they already fired the "bought a pink cow" event. You don't have to wait a few weeks after the experiment to collect data (most funnel analysis software I've used requires a do-over any time you create a new funnel or alter an existing one except by concatenation) -- you have the data, you're just slicing it a new way.

The notion of preventing a turf war between the analytics side and the engineering side is really popular with some companies. They also typically have lots and lots of money to spend.

Nothing. I looked at KISSMetrics ages ago when the private beta was first leaked. Similar to crazyegg it focuses on a single aspect of analytics, in this case conversion funnels. So it is simple to setup funnels for every possible conversion point (which if your website is effective should be almost every page). However, you can do this on GA too, which is also a hugely more powerful and integrated tool. Someone mentioned previously how expensive enterprise analytics is... this is _not_ enterprise analytics. I'm really not sure who they're aiming this product at.

We (nearlyweds.com) are a beta customer of Kiss. We use GA and Kiss right now. As much as I love GA, they do a crappy crappy job at tracking via conversion funnels, in fact it's almost useless for our business. The real story is that Kiss tracks people, not just visits. This gives us clarity and insight into our conversion funnels, which paths convert the best, which referral sources convert the best, and then on the back-end, because they track people, which sources produce the subscribers that last the longest, etc. We can set up funnels for just about anything.

There is also a psychological element at play here that has been interesting for us. When the product was free, we didn't prioritize optimizing our funnel aggressively. Now that monthly fees are looming, we are having to fish or cut bait and get real serious about which funnel projects to pursue or not to pursue. It's a very healthy restriction on our business (a la 37signals).

It was purely the simplicity of the user interface for visualising the steps in the sales funnel and seeing exactly where you were losing conversions. I think you can construct something similar in Google Analytics but KISSmetrics is a much more focused tool (but not worth the difference between free and $149/month!)