An Awesome Post i found through @danmartell & watchingwebsites.com about the power of google anayltics and alerts
Google Analytics Alerts: the start of a complete view?
Google Analytics recently added a new feature, called Alerts. At first glance, it’s an elegant way to show someone when a KPI on their site has changed significantly from what’s expected. It’s baselining, applied to all KPIs — even the ones you’re not looking at.
This is a great idea for folks who forget to check their analytics data, because now they can find out about significant events. It tricks you into being a better analyst. It encourages baselining, segmentation, and thinking about your business. But we think it’s the start of something bigger, once it incorporates the things Google and others know about your online presence.
Details, and some juicy UI mockup speculation, after the jump.
Baselining, even when you didn’t know you should
Beginner web analysts treat analytics as accounting. They use it to report the news, not make the news. It’s only the more advanced analysts that see analytics as a means for optimization, using things like A/B testing to learn whether a change made things better. And to do that, you need a baseline.
The new feature learns what normal is, then shows you deviation. This encourages experimentation: “I tried something new today, and I can see the results.” Google’s already introduced comparative rankings, showing you how you’re doing against others; now, they make it much easier to identify significant changes to your site, even if you don’t know where to look.
Imagine, for example, that you change your website. You don’t see an appreciable shift in traffic volume, so you decide it didn’t have an effect. But hidden in those traffic numbers is the fact that there was an increase in European traffic at the expense of US traffic. The new functionality would show you this, allowing you to tailor content to specific geographies.
Making segmentation easy to try
The new functionality tries to find chunks of traffic that have “broken away from the pack.” It does this for known metrics and segments — such as geographic regions — as follows:
Notice that little “create segment” at the end? It makes it easy to carve out a slice of traffic you should care about, which then means you can start to play and experiment with it. Segmenting traffic is a sign of web analytics maturity, but until recently, it’s been something few people play with. Now, Google Analytics is essentially telling you, “hey, dummy, have a closer look at this.”
You can use custom segments in lots of cool ways–for example, as the analysis above shows, I now know that returning US visitors are more likely to download content from the site, but first-timers aren’t. Once you’ve seen a segment that Google found for you, you’re more likely to create your own because you understand how they work.
Thinking about your business
You can also set up custom alerts within the system to tell you when something’s gone out of kilter. We know lots of companies who use revenue or transactions per second as the first sign that something’s wrong on the website — this is a great top-down approach if you can manage it, because it means everyone in the company is focused on what actually pays the bills.
The new functionality lets you look for specific occurrences even before they happen. Consider @alexbfree’s recent post on Twitter Retweeting, which got picked up by Dave Winer. You can set up an alert to see if Dave sends you traffic:
Overall, these are excellent enhancements to the product. They’ll improve engagement — because the system will tell you when things are happening, rather than waiting for you to log in. They’ll encourage good behaviors like baselining and segmentation. And they’ll also satisfy the less business-centric, more hobbyist segment that just wants to know when the world is thinking about them.
What I really want: a holistic view
It’ll be more useful (and in keeping with the Complete Web Monitoring philosophy) when it includes other kinds of data:
- A timeline of posts created, based on Feedburner statistics or blog history
- A series of Google Alerts showing when some search criteria on the web is met
- A volume of followers or friends obtained through the APIs of social platforms like Twitter, Facebook, and Flickr
- Performance data from synthetic or real user monitoring
- Voice of the Customer feedback through systems like Kampyle
Here’s an example of what that could be like, for a content creator/blogger.
That’s a pretty intimidating amount of information. Most of it, Google already has; some, we’d get from elsewhere. We borrowed concepts from:
- Bit.ly’s historical views (over a longer time period) with a rollover for individual links on a given day
- Google Labs’ News timeline
- The dashboard of WordPress
- Postrank’s content scoring system (we spent time with these folks this week)
- Feedburner RSS stats
- Email subscription management stats from a mailing list provider
- Moni.tor.us performance monitoring
- Trendistic’s timeline graph of Twitter (with a rollover of Outwit.me’s realtime tag cloud)
- Google Alerts, which come in by mail but could be turned into a timeline with rollovers
Anyone with a bit of time and some spreadsheet know-how can assemble this manually; it could also be done in Greasemonkey with a bit of work, using Google’s new views as the anchor.
Admittedly, this is still “reporting the news” — the real insight comes from observing correlations, such as what kinds of posts increase subscriptions or what news drives follower count. And this is targeted at a specific kind of site (media/community) whereas other businesses more focused on SaaS or e-commerce revenues probably want something that shows productivity or conversion rates.
Unfortunately, there isn’t a lot of money in giving tools to bloggers. We’re a cheap bunch. So while there’s great multivariate testing for online retailers, a content creator has to cobble together many different views and data sources to paint a complete picture.
By alistair – November 26, 2009