So, the new Google reader is pretty cool, although I have to say that I liked the old look better.

But here’s what’s most intriguing to me…

…Did you notice that as you scroll through the lists of posts that the number of unread items goes down? Now, I’m totally daunted at the thought of trying to deconstruct Google’s JavaScript so I haven’t even attempted to ‘peek under the hood’. However, from a purely presentation point-of-view, it would seem that they might be using ‘onFocus’ to mark each item scrolled/scanned/read as read.

Why is that intriguing? Well it takes me back to a discussion I took part in back at Emetrics in April. The point of the conversation, to put it melodramatically, is that ‘the page view is dead’. One of the discussion participants declaimed that he wasn’t interested in page views because a single page might include multiple news items that he wanted to measure.

So if Google is using something of the sort I describe, couldn’t they easily hook that interaction into Google Analytics with a whole new metric? Something called ‘Post View’ or something less prosaic? Isn’t that we harp about when complaining about measuring Web 2.0/AJAX/RIA?

Here are some questions that pop up in my head while thinking about this:

1. Does onFocus equate to a person actually reading the post? When Eric Peterson launched his new vendor discovery RSS feed over the weekend, I quickly scrolled through about 50 posts in Google Reader but only actually read about 5 because I wasn’t really interested in which random site was using which random analytics tool
2. Are there any studies that show what the average time to read 50, 100, 200, 500, etc. words online is?
3. If such data as the above exists or could be executed with some rigor, could a combination of onFocus and time spent on that focus more accurately measure ‘engagement’ with an individual post?
4. What kind of KPIs might we drive out of the above scenario?

I guess if Peterson ever gets the “Virtual Web 2.0 Measurement Working Group” organized I can ask these questions there, but until then I leave them for you to ponder.

Ok, so it’s more like what I’ve been keeping my eye on and this is a desperate attempt to make sure that I actually write something this week, but still here are the things that have held my interest.

• Manoj Jasra over at Web Analytics World had four guest posters this week, Avinash Kaushik, Jennifer Laycock, Gord Hotchkiss and Rand Fishkin. They all cover different topics and each one is worth the read.
• Robbin Steif has an excellent post on what web analysts do all day at her Lunametrics blog, so if your bosses, peers, family, or friends have a difficult time understanding what you do – point them to Robbin.
• Both Stephen Few and the guys over at Juice Analytics pointed out this new tool for creating sparklines and bullet charts (among other things) in Excel, so I’ve been playing with it. By the way, it’s called Microcharts by BonaVista Systems
• Of course, the Hewlett Packard BoD fiasco has been absolutely fascinating (and horrifying) – Scoble is keeping tabs on it: Day 1, Day 2, Day 3, Day 4
• Mom’s back to blogging, this time from Egypt where they’re on a 2 year teaching stint – check out the new masthead for the blog which is the view from her classroom!

Pushing on, let’s take a look at where the page views in Part 1 came from.

FYI, this data comes from Google Analytics which I’ve had up and running since April 20 so it’s missing roughly six weeks of data but I didn’t feel like buying the premium service from StatCounter.

It boils down to this: “Eric T. Peterson is GREAT for business!” Since historical referral data has been available, Eric is responsible for 28.9% of all referrals.

Notice too, that even though Avinash has only been blogging for about a month at this point, he is already in what I’d call the mid-tier of Referrers.

Organic searches from Google also do quite well at 8.2% but we’ll explore the Google segment at a later time.

Interestingly, two of the folks that I identified as my key ‘Sneezers’ in the last post show up in the small tier – maybe I just like Jeremiah and Robbin so much that I wanted them to be my key sources.

Finally, we can see that directly referred visits are the number one category (just barely beating Eric) at 29.8%.

Oh! And I was surprised at the volume of incidental traffic received from Blogger; which since I am blocking self-generated traffic, must be coming from the ‘Next Blog’ link in the Blogger chrome. I guess I’ll need to segment that blogger traffic and see if there is any value in it.

---

So, how do my chosen sneezers do on a trended basis? Let’s take a look.
A couple of things to note about this chart:
1. Web Analytics Demystified and Directly referred visits are on the left (primary) axis and everything else is on the right (secondary) axis2. As discussed here, I’ve been playing with Excel 2007 and these charts were created in it. Unfortunately, Excel 2007 is still a bit buggy and it kept dropping the x-axis data points so the date range is April 20 – June 14.Okay, with the caveats out of the way, what’s going on here?I’ve got nice smooth growth both from Eric and Direct but my other picks have this odd stair-stepping quality to them.My hypothesis is that, for whatever reason, there is a great commonality between Eric’s audience and those who find this blog useful so I get a relatively steady stream of referrals from his site.The stepwise growth from my other ‘Sneezers’ most probably is a function of when they post something that directly refers to Instant Cognition.The latter is pretty simple to prove – I just need to go look at when folks like Avinash referred here in a post and see how it correalates to the steps in the chart.

For the former, I’m not sure how I’d go about testing it – any thoughts?

One thing that I find particularly gratifying about the chart is the nice smooth growth in direct referrals — I’ll have to do a segmentation study to be sure, but — this seems to indicate growth in brand loyalty (boy I hope that’s true) as my visitors are just hitting a bookmark or typing in ‘instantcognition.blogspot.com’ into the address bar.

I thought I would start off this analysis by comparing my blog traffic to Xavier Casanova’s theory of Readership Growth:

The basic cycle goes like this:
(1) Seeding. You start a blog, write 8 or 10 interesting posts, and advertise your blog as much as possible (link on your email signature/forums/etc). Soon enough, another blogger will write about you or point to you (blogroll, etc).

(2) Exponential growth. After reaching some critical mass of readers, readership growth accelerates all of the sudden – typically because a popular
blogger has written about you, or there is a press article pointing to your site.

(3) Organic growth. Your base keeps growing but it’s slooooowww. You’re incapable to sustaining the growth rates you had in phase 2.

So let me caviat this by saying that I am looking at page views and not readers – I don’t have a sufficient sample of readers via Feedburner or Bloglines to make the direct comparison but I think that page views are an acceptable stand-in in this case.

At right, is a chart of page views to this blog during its first one hundred days. You’ll notice that it has both daily tallies as well as a cumulative line. I’ve also included what I think are key events. In most cases, these are when “Sneezers” pick me up and spread the word. Although please note that Microsoft’s acquisition of Deep Metrix caused a spike in traffic to this blog (as it did to most blogs in the web analytics community) around May 6. In addition to the actual stats (daily and cumulative) I’ve added a best fit curve (4th order polynomial) which besides having a great correalation to the actual data, resembles Xavier’s theoretical chart above quite closely.

So, following Xavier’s theory, it looks as though Instant Cognition was in the ‘Seeding’ phase from March 7, 2006 through about April 18, 2006. On or about April 19, this blog entered it’s first ‘Exponential’ growth phase (a low exponent to be sure) that was kicked off by my attendance (and possibly my presentation) at Emetrics. Then, soon after Jeremiah Owyang (a blogging and podcasting heavy) added me to his blog roll and Avinash Kaushik started blogging my exponential growth phase comes to an end and I am now in the ‘Organic Growth’ phase until I get picked up by a new community.

For those who are interested, here are the linear slopes for each phase:

Phase I: 16.21
Phase II: 34.67 (more than double Phase I)
Phase III: 22.43

A couple of things that stand out to me here.

• If you walk along the cumulative line, you see what I consider to be micro traffic cycles. Consider for instance, April 19 through April 22 (Emetrics), in fine, this spike event looks just like the theoretical pattern posited by Xavier. The micro cycle repeats itself between May 15 through May 19 and again from May 28 – June 6 — and these are just a few examples.
• Secondly, observation of the daily page views bar chart bears out the growth trend just by eyeballing the density of high-traffic days during each phase. In phase I there is a paucity of high volume days while in Phase II there are many more and then the high volume days settle back in phase III, they are not as few as in Phase I and the low volume days are at a higher average than before so it looks like Phase II had a positive impact on phase III (otherwise the last phase would look a lot more like the first one).

So, what have I learned from this first part of the analysis?

• Well, if Xavier is right, I’ve got a fairly typical blog and am at least doing some things right. We’ll get into more detail later, but for the most part, I’ve picked up the ‘right’ Sneezers: Eric Peterson, Robbin Steif, Avinash Kaushik and Jeremiah Owyang to name a few.
• Also, it’s hard to get out of your own way. At work I’m dealing with numbers that are many orders of magnitude larger. Climbing down off that mountain to look at my blog traffic is tough.
• Oh! And Xavier’s a really bright guy. Ok, I already knew that (and so did all of you probably) but he is being especially kind by letting me reproduce his charts and theory here and by offering up any other data or help I might need as I go down this 100 days path.

Sources:

• StatCounter
• Coffee, Sun & Technology

Special thanks to Xavier Casanova of Perenety for letting me use his Blog Theory and charts for this comparison.

I noticed (mea culpa) Matt Jacobs’ blog “Digital Media Analytics” this evening.

I just read through a very detailed post on Click-to-Landing Page Drop-Off. Which covers the vagaries of search engine marketing measurement from Click Fraud to Visitor Defection and Measurement. Not only does Matt offer a detailed view of the measurement pitfalls but he also offers solid recommendations on how get the most out of your web analytics package, ad vendor and site monitors to get a very wholistic view of those marketing efforts.

If you are a search marketer, or need to help search marketers figure out their campaigns, I’d encourage you to read this post.