Coffee, Sun & Technology

This 812,934th article on Google Analytics was written up a little too quickly… Favorite quote:

The launching of the free Urchin Web Analytics sends a message to large SEO or analytics companies such as WebSideStory, Lunsford and Efficient Frontier. The three companies, including small SEO businesses, specialize in optimization by relying on complex algorithm to analyze their clients’ web traffic. All three large analytics companies have suffered falling stocks on November 14, as soon as Google’s Analytics software became free.

(Jeff Lunsford is WebSideStory’s CEO, Efficient Frontier isn’t public, they don’t compete with Google Analytics, etc…)

Is there a correlation between a post’s popularity and its length? Maybe. I tried to prove this by pulling out a Page Metrics report from my favorite Web analytics tool and ranking posts by the number of sessions to it. This is what I found:

My theory is that the longer it takes for people to read a post, the more captivated they are, the more likely they will recommend it to a friend - the higher the session count for the post.

Definetly a simplistic theory but I have to run with this assumption right now and continue to grow my base of readers with longer, higher quality posts.

[Read other posts for this series: part 1 part 2 part 3 part 4 part 5 part 6 part 7]

Firefox 1.5 looks amazing and I am relieved to see cookies are treated right. I mention this because one of the new Firefox 1.5 features looked pretty scary on paper for all the cookie-monsters of the world:

Clear Private Data: Protect your privacy with the new Clear Private Data tool. With a single click, you can delete all personal data, including browsing history, cookies, web form entries and passwords.

Luckily upon clicking on the new Clear Private Data button, you can see that cookies aren’t deleted by default.

Perhaps due to the fact that Google - a big supporter of Firefox (to say the least) - heavily relies on cookies for personalization purposes and tracking (via Urchin/Google Analytics). Whatever the real reason, Firefox is sending the right message: cookies aren’t evil.

I came across this nice write up from Peter Rip on Web 2.0 business models, and realized how deeply the Web analytics industry will need to transform itself to accomodate the upcoming Web 2.0. As content gets more and more granular, more automated, and more “mashed-up” - reporting needs will radically change over the next 2 or 3 years (for content sites as well as retail sites).

The hacking game has just begun for Google Analytics. Meanwhile the Google guys are accepting registrations again after some initial problems.

When are we going to see some Google Analytics-based Mashups?

A few days ago I met with a Web analyst who reminded me of a very important point - the most insightful A|B tests are the ones contrasting two radically different approches to a page or process. If generally speaking it’s hard to tell the difference between version A and version B then results will be identical, garanteed. If you are worried about seeing a revenue drop because of a risky challenger (understandable), then start your test with a 90/10 split and gradually increase to 50/50 over a few days’ time.

Using my favorite Web analytics tool, this is what I saw this morning on my home page (labels indicate click-through rates):

Nice confirmation for something we already know: first item gets most clicks, second gets about half of first’s, third about half of second’s… etc. The interesting thing about this is that the content has limited influence over what gets clicked on. Historically, the KPI link (second reference on the screenshot) always outperformed the Omniture link by a 2:1 when presented independently. Not on this list however.

Topped the 100-session mark yesterday (108 to be exact).

Hopefully the positive trend will continue as I’m refocusing my optimization efforts.

Some visitors are more valuable than others. I tried a few things to increase the number of dialy visitors to my blog, but not all of them worked:

Action (see part 4): “I bought some Google Adwords to acquire new visitors interested in analytics. Ironically, I opened up an account on Saturday with a few generic keywords – then on Monday the Google Analytics announcement hit. That was the perfect opportunity to secure tons of new keywords at a very cheap price (my budget for Adwords is $1/day).”

Results: Mixed. For $4, I got more than 20,000 impressions according to Adwords, 99% of them for the keyword “google analytics”, which generated about 40 clicks. Clearly an awesome deal for the money, however, (a) managing this process was time consuming (adding keywords, creating new banners), (b) the exit rate on the landing page is at 57%, and (c) less than 10% of these 40 clicks came back to the site within 7 days.

Action: “I decided to buy a few keywords on Google France and see if there is any benefit going international for lead-gen. Generally speaking, I was surprised to see many great keywords were available for less than $0.10/click. In truth, volumes are fairly low but it’s worth a try.”

Results: Poor. I went a step further and bought the same list for Spain, Germany, Italy and Portugal. I am generating less than 200 impressions a day and 2 or 3 clicks.The good news is that you can buy very high quality keywords (websidestory, omniture, coremetrics, fireclick, webtrends, WSSI, etc) and with a decent placement for very little money.

Action: Post on hot topics, making sure I cater to a wider audience and … make trying to make my posts as interesting as possible (not always easy!).

Results: Good (I think) – I do not have a lot of data but I am getting a lot more referrals from Digg, Del.icio.us, or (very important) other bloggers as a result of this. My hypothesis, to be tested this week, is that such referrals tend to “snow ball”, i.e. they generate a lot of repeat visits AND new referrals.

Conclusion: First, defocusing from Adwords, clearly, and second, work on the content. I will also continue to work on the site layout, introducing some very light personalization touches.

[Read other posts for this series: part 1 part 2 part 3 part 4 part 5 part 6 part 7]

Since Google Analytics came out there’s been a lot of expert talk about what this would do to the Web analytics and online marketing industry. Like thousands of other bloggers, I installed Urchin on Coffee, Sun & Analytics (my personal blog) on Monday morning. Before I give my impressions, I need to disclose that I work for Fireclick, a leading vendor in the Web analytics space. We compete against WebSideStory, Omniture, Coremetrics, Webtrends and Clicktracks – and therefore are affected directly by the Google announcement. Please keep that in mind as you read this.

I will divide my assessment in 3 parts: general overview, technology, analysis and conclusion. Let’s start with the general overview:

a) Implementation. Installing GA on www.coffeesunanalytics.com was extremely easy. I logged to my Adwords account, clicked on the new “Analytics” tab, and signed up for the service. The main tag is just 3 or 4 lines, a simple copy and paste. Wordpress is my CMS and they have a plugin available for download to include the Google tags. The process literally took 5 minutes.

b) Getting Started. The setup, or initial phase was a little more frustrating. It took about 36 hours for the first reports to populate, and I realized that Urchin likely doesn’t offer any real-time reporting. Also, as I am writing this I only have data for Tuesday (Wednesday through Saturday are still missing). They obviously have a few bugs to work out.

c) The UI. When my first reports arrived I took a first look at the interface: a very nice Flash/Javascript UI that gets the job done even though it’s a little slow, a little clunky and not too interactive. The geo reports are kind of nice, the pre-defined dashboards look pretty good too.

d) Report set. On the reporting side there are about 80 something reports, which is a little light, with 1/3 of them are about search performance – the rest is focused on the basic stuff (visitors, referrers, customer system, etc). I really liked the contextual menus on each of the reports, which allow to get single campaigns to be charted over time in 1 click, or cross-referenced (segmented) with session-level variables such as referring domains.

e) Commerce analysis. I like the simplicity of their funnel analysis. Funnels need to be pre-defined but they are reasonably insightful.

f) Merchandising analysis. I did not see any product analysis capabilities.

g) Marketing analysis. Great tracking capabilities for Search and nice banner testing reports (as well and A/B testing reports). I don’t know how emails are tracked though, and I suspect the UI doesn’t work too well when you have multiple types of campaigns with thousands of records for each.

h) Content analysis. I was very unimpressed with the content analysis capabilities, as they did not allow for multi-dimensional analysis (important for content sites to get reports by section, sub-section etc).

i) Path analysis. Their path analysis (navigational analysis) tool is primitive at best, in my opinion 2 or 3 years behind the rest of the pack. Worse, it doesn’t appear to be segmentable – i.e. I can’t get path analysis for “first time users” for instance.

j) Excel. Excel support was very basic as well, I could not find any Excel plugin module to pull and manipulate data in Excel.

k) Browser overlay. I think I briefly saw a browser overlay tool but for some reason it disappeared on Tuesday.

l) Segmentation, Ad-Hoc reports. It’s easy to segment an item in a report by some session-level parameter, and easy to look at an item over time. But that’s about it – you can’t go deeper and you cannot define your own segments.

Next will be the technology.

Web analytics, Google analytics