Nate Weiner

This is an old archived post from my former blog The Idea Shower. It's where I cataloged my product explorations and releases, one of which ultimately became Pocket.

This post was published back in 2007. It may not function as originally intended or may be missing images.

Google Analytics Markers

November 16, 2007

The next target in my continuing series of "I wish it had this" ideas is Google Analytics.

What I'd like to have is the ability to tag days with information about what happened. So that when peaks occur I can mark them with things like 'Frontpage of Digg' or 'Mentioned on Lifehacker'.

When looking at recent weeks it's not too hard to remember why you had peaks in traffic, but not when you roll back and look at traffic over the course of months or a year. The graph may look pretty but it doesn't really help you understand what influenced the patterns you see.

This technology isn't anything new. In fact, Google already uses it on their stock pages (Example). They do this to let traders get a better feel for what caused the stock to jump or drop without having to individually research each and every day. Well, analyzing stock graphs is not much different than analyzing traffic volume, both are highly influenced by day to day events.

A Step Further

In the same vein, I think that some of these markers on peak days could be generated automatically. Almost all spikes in traffic are a result of one page, and generally are from one source. So when there was a large spike in traffic, Google could slap on a default marker including the top content page for the day and the top referring source. So if for example, you wrote a blog post about '10 Puppies Wearing Hats', and it got picked by The Association of Puppy Hat Wearers who linked you and sent you 10,000 visitors; Analytics could then mark the day with your puppy post and the referring source and you'd be covered.

Even further, if Google simply added the top content page and the top source of the day to the information when you moused over any day, you would be able to very quickly get a grasp of what was causing your peaks and what was popular during certain trends.

What's Out There?

As usual, if you are aware of any stat packages that already have this feature or one similar to it, please let us know in the comments below.