As the Web 2.0 hype spreads wider, sooner or later it’ll start spreading beyond its trendy enclave and start to penetrate main stream websites. I’m sure that’s already happening in a few places (such as gmail), but by and large web 2.0 is dominated by small start ups. When you get big companies deploying stuff, they want to measure it - and rightly so.
However, web 2.0 is resistant to the normal sort of measurements. What are the traditional measurements? Let’s go through some Web 1.0 measurements and see how many of them are relevant in the Web 2.0 world.
| Web 1.0 | Web 2.0 |
| Page views | Not much use in AJAX applications. Replace with user actions - how many people clicked on button X. |
| Unique users | Still some relevance, however more and more people are blocking cookies, and most modern web browsers make it easy to block 3rd party cookies which many tracking services rely on the maintain session state useful for unique users. Still a relevant metric and it never was precise (a dirty little secret the stats vendors don’t tell you), it may simply become increasingly less precise. |
| Click paths | Click paths are still very relevant, but again, challenging in AJAX applications. A new term I propose is "action streams". What sequence of actions do people take? |
| Conversion statistics | Many people, particularly ecommerce, are obsessed with conversion statistics, as they should be. Which ad drove the most traffic? Which ad converted the best? Which keywords? Which ad copy? Which graphical execution? Which landing page? I just dumped Overture from a campaign I manage as it was generating plenty of traffic but very few conversions relative to other sources. Web 2.0 still needs conversion statistics but needs to take this to another level - see below. |
New to Web 2.0
- Measure onsite buzz. Web 2.0 is more than just technology, it’s also about engaging your customers and partners and allowing them to contribute. So, measure the comments on your blog. Measure the track backs (ok, not strictly onsite). Which subject generates the most comments? Which subject generates the most positive track backs? Which post caused a big jump in the number of RSS subscribers? Which page on the Wiki is getting the most edits? The most page views? The most vandalism? Let your imagination run free, I’m sure I’ve only scratched the surface here.
- Measure offsite buzz. This should be done in Web 1.0 world as well, but I very rarely see it. What are people blogging about you (without a track back)? Posting in forums?
The interesting thing is many of these things are very difficult to do in traditional analytics packages. Show me how you work out in webtrends or websitestory how many people clicked on button X which invoked AJAX action Y. Yes, there are ways to do it, but it’s really hard. Web vendors are going to have to either make a radical shift, or people are going to need to start instrumenting their applications - building measurement into them. I suspect it will be a combination - people will need to write against an API/webservice their statistics provider exposes. That has some pretty major repercussions for both statistics providers and for application developers. There’s certainly a great opportunity here for the right people, and vendor lock in will be pretty enormous - unless someone develops a standardised statistics API (I doubt that’ll happen).