Chasjo Mitch's List: Analytics

• To be fair, report integration can get difficult other analytics vendor offerings   as well. My colleague Phil   Kemelor points out that WebTrends   and other commercial web analytics vendors can also be rigid in this regard.   He's right, although I've seen WebTrends report or data integrations across   various CMS packages, but have never seen anything similar with Omniture reports.
• As our Web Analytics research   subscribers know, Omniture has a reputation for being particularly closed when   it comes to exporting data or reports. It wants to become your web data warehouse   and central analytics hub
• Omniture would have you log into their   dashboard, but many casual web managers find that too complicated, too remote,   and not actionable. Web managers then tend to turn to (the probably over-worked)   in-house Omniture guru. That seems wasteful. In these economic times, analytics   are too important to be left only to analysts.
• (One increasingly evident alternative -- resorting to the lightweight traffic   metrics possibly built into your CMS -- doesn't solve the problem either. These   CMS-driven systems typically deliver sub-standard metrics that even their vendors   won't fully stand behind.)

2 more annotations...

• Conversion Variable
• As previously discussed, Traffic Variables allow you to segment or breakdown Page Views, Visits and Unique Visitors, while Success Events capture metrics around conversion actions taken by site visitors.
• So what if we want to be able to show the percentage of Lead Generation Form Submissions broken down by Language?
• Perhaps we want to see Shopping Cart Additions broken down by Zip Code or Campaign Tracking Code. 
• There are lots of cases where you will want to segment/breakdown Success Events in a similar manner that Traffic Variables allow you to break down Traffic Metrics.  
• Conversion Variables to the Rescue!
   The purpose of Conversion Variables (also known as eVars) is to allow you to breakdown Success Event metrics in a similar manner that Traffic Variables allow you to breakdown Traffic metrics
• While this sounds easy, Conversion Variables are actually one of the most confusing topics for SiteCatalyst customers because there is a lot to learn about how they behave (which is why this is only Part I).
• Unlike Traffic Variables, Conversion Variables are persistent meaning that once a site visitor gets assigned a value, that value sticks with them until you (SiteCatalyst Administrator) tell SiteCatalyst to clear it out (unless the user deletes their cookies or uses a different computer). 
• Conversion Variables have a direct relationship to Success Events.  I like to think of the two as “married” since they complement each other. When a SiteCatalyst Success Event takes place (i.e. visitor takes an action that results in a Success Event being set), SiteCatalyst assigns credit for the Success Event to one value in each Conversion Variable repor
• As many more Success Events take place, different Conversion Variable values (Cities in this example) would get credit for the various Online Course Demos that take place on the site such that, over time, the report would have many cities and the associated number of Online Course Demos that took place for each.  But, if Conversion Variable 2 represented the current visitor’s Age, then at the same time each City is getting credit for each Online Course Demo in Conversion Variable 1, a specific Age value (say “18 Years Old”) is getting credit for the same Success Event in the Conversion Variable 2 repo

8 more annotations...

• When the newspaper division of Gannett transitioned from WT log file analysis to Omniture beacon tracking, we experienced a drop in metrics such as uniques, visitors and page views.
• Were you using WT's log file product or beacon product?
• We have finally completed our WRC to OM transition, and visit numbers  > decreased by half! What we are looking at is a long and tedious number  > reconciliation (if this is even possible). >  > We were using Webtrends 6.1b, log version of course. We had filters to  > block spiders and bots, and I honestly have no idea why the visits  > would decrease so much. >  > Has anybody seen a similar drop? Any advice?  >  > Explaining this to the client will not be pleasant. I am concerned  > that they will force us to go back to Webtrends for the sake of keeping  > their "good" numbers.

1 more annotation...

• We have finally completed our WRC to OM transition, and visit numbers  decreased by half! What we are looking at is a long and tedious number  reconciliation (if this is even possible).  We were using Webtrends 6.1b, log version of course. We had filters to  block spiders and bots, and I honestly have no idea why the visits  would decrease so much.  Has anybody seen a similar drop? Any advice?   Explaining this to the client will not be pleasant. I am concerned  that they will force us to go back to Webtrends for the sake of keeping  their "good" numbers.

• Like any software product you use, there are a few key elements that you need to understand to be successful.  In working with clients I have found that many do not have a good understanding of the three fundamental building blocks of SiteCatalyst: sProps, eVars and Success Events.
• However, over time, I came to understand that these variables are the foundation of all SiteCatalyst reporting, so if I wanted to use Omniture reports to measure my website success and/or justify my re-design theories, I had better suck it up and learn about these three variable types. 
• Traffic Variables
   Traffic variables (known to old-timers as sProps) help to track page-by-page site traffic activity.
• The primary purpose of sProps is to allow you to breakdown Page Views, Visits and Unique Visitors into meaningful buckets. 
• Without sProps, you would not be able to see such things as which pages are the most popular or what percentage of pages were viewed in the USA vs. Canada, etc…. 
• An important thing to know about sProps is that they are not persistent.  This means that they do not retain their value from one page to the next, a concept that often confuses Omniture customers
• easiest way to understand sProps is to think of the most commonly used sProp: Page Name.  Let’s assume a visitor to your site starts their visit on the Home Page and the value “Home Page” is passed into the Page Name sProp.  From the Home Page, the visitor clicks on a link and is taken to the “Contact Us” page.  You would certainly not want the value of “Home Page” to persist and be passed into the Page Name sProp on this next page or it would look like the Home Page had two Page Views instead of each page having one Page View.
• Another important thing to know about sProps is that they are used for Pathing. Pathing is the ability to view the order in which values are passed to a particular SiteCatalyst sProp for a specific Visit. 
• f Pathing is enabled for that sProp, SiteCatalyst would record the order in which the values were passed and allow you to view reports that show the percentage for which all site users went from Page A to Page B, etc…
• Visits, Daily Unique Visitors, Weekly Unique Visitors, Monthly Unique Visitors and Quarterly Unique Visitors can be enabled for any sProp (for an additional fee)
• SiteCatalyst provides several “out-of-the-box” pre-defined sProps including: Pagename, Site Section, Server, Browser, Country, etc…  These are similar to all other (custom) sProps with the exception being the Pagename sProp which is somewhat special in that it captures the page URL if no value is passed to it.
• • Any two sProps can be “correlated” or broken down by each other
   
  • Pathing can be enabled for any sProp
   
  • Any sProp can be “classified” using SAINT
• To accomplish this, the client would pass the language that the current page is being viewed in to a custom sProp.  For best results, this sProp should be populated on every page so that the total Page Views in this report matches (or is close to) the overall number of Page Views for the same timeframe.

11 more annotations...

• What are Classifications?
   So what exactly is a Classification?  Technically speaking, when you “classify” a SiteCatalyst variable you are establishing a relationship between a variable and meta-data related to that variable.
• lassifications are most frequently used in the Campaigns area so I will use that as a way to explain them.  Most clients send campaign traffic to their site using a tracking code.  This tracking code is an identifier that may represent a specific keyword purchased on Google, such as “goog123.”  This identifier is passed into the s.campaigns variable so you can see what site success events take place after visitors come to your site from that campaign code.
• But what if, instead of viewing Campaigns just by the tracking code, you want to see campaign results by Search Engine or Keyword or Campaign Channel?  Do you have to create a new conversion variable for Search Engine, another for Keyword and yet another for Campaign Channel?  If so, you would use up many of your fifty variables on Campaigns alone!  Thankfully, you can use Classifications to make your life easier! 
• Since each tracking code could have a Search Engine, Keyword or Campaign Channel, you can simply create three Classifications of the Campaigns variable to represent each.  You are essentially telling SiteCatalyst that there is a direct relationship between the Campaigns variable and these three other “meta-data” values.  By doing this, SiteCatalyst will allow you to slice and dice site Success Events by all four variables with no additional tagging!

2 more annotations...

• In this case, if the SiteCatalyst user clicked on the report highlighted above, they would see a report that shows the Page Names that were viewed by visitors having a Browser Height between 750 and 759 and their associated percentages:
• However, when you create a Traffic Data Correlation, you get the ability to see two new reports, not one.  This concept confuses many of my clients.  In the example shown above, we see a report showing Page Names for a specific Browser Height, but with the same Traffic Data Correlation, we can also see the converse, which in this case, is a report that shows all Browser Heights for a specific Page Name:
• While you can see Page View, Visit and Unique Visitor metrics for any Traffic Variable, SiteCatalyst only shows Page Views in Traffic Data Correlation reports (other metrics can be seen in Omniture DataWarehouse or Omniture Discover).
• SiteCatalyst provides the ability to create Traffic Data Correlations with many standard Traffic Variables out of the box such as Browser, OS, Resolution, etc…
• Enabling a Traffic Data Correlation allows you to correlate the two Traffic Variables from that point forward, but does not allow you to see the data correlated retroactively.
• In Omniture DataWarehouse and Omniture Discover, all Traffic Variables are automatically correlated with each other
• One of the items Greco Inc. has implemented is the passing of the Language that each website page was viewed in (”English” or Spanish”) to a Traffic Variable.
• Since every page of the site has a Traffic Variable value for Page Name and Language, enabling a 2-item Traffic Data Correlation will allow us to see Page Names broken down by Language and Language broken down by Page Name.

6 more annotations...

• Chatter is building on the interwebs about Omniture’s recent (and ongoing) latency woes. Looks like both SiteCatalyst and Discover are days behind in processing data (according to messages on Twitter, up to around 5 – 7 days in some cases)
• But do these problems tells us something more about Omniture’s (or any other web analytics company’s) ability to run a viable business? Infrastructure costs are a big part of a web analytics firm’s cost base (at least, those with a hosted offering, which is all of them). And unfortunately, these costs don’t really scale linearly with the charging method that most Enterprise vendors use – charging by page views captured. Factors like the amount a tool is used, and the complexity of the reports that are being called upon, have a big impact on the load placed on a web analytics system, and the resulting infrastructure cost. It’s tricky for a vendor to recoup this cost without seeming avaricious.
• an Enterprise web analytics pay the bills for its own infrastructure cost? Or will all web analytics ultimately need to be subsidized by something else (such as, oh, I don’t know, advertising)?

1 more annotation...

• Web logs are only a small part of the story. What is important is how do customers interact with a site to produce profitable outcomes. It is the customer experience that should be measured, not how well the web site works as a piece of technology.
• At the heart of web analytics is a need to uniquely and correctly recognise visitors. This requires persistence in the item used for identification, and requires that barriers to that accuracy be understood and eliminated.
• By ensuring that visitors are correctly identified it is then possible to obtain an accurate picture of the funnel from browsing to purchasing.
• Most tools, including WebTrends, offer both out of the box reporting and also access to raw data; WebTrends offer a half-way house whereby the events which are aggregated automatically can be analysed by segments which can be user defined and then stored to be run at will, enabling new insights to be specified without needing technical skills.

2 more annotations...

• it’s a bit of a departure from the intricately granular management platform Omniture Search Center provides, but might be a welcome change for SEMs looking for a more high-level strategic solution.
• This approach differs substantially from industry-leader Omniture, as WebTrends aims for that group of search marketers that don’t necessarily want to waste time on delving deeper into statistics and management options. 
• To be fair to Omniture (if I don’t, I’ll get phone calls), Search Center does have a bevvy of tools available to automate several unique processes and manage groups of keywords, but due to certain limitations that I won’t detail here, you can’t say that it’s a simplistic ‘high-level’ solution.  Setting up and understanding those bid management tools and actually gaining ROAS from using them can be a significant investment in human capital.
• The company is employing a strategic, versus tactical, approach to analytics and online marketing management.  Rather than scrutinizing the return on ad spend (ROAS) for various individual keywords, WebTrends Ad Director groups various keywords and tries to achieve the greatest possible ROAS for that given group. There are still big winners and big losers in that group, but the overall effect is a much higher ROAS.

2 more annotations...

• Autonomy Interwoven is the first software provider to offer a WCM solution that includes MVT and closed-loop analytics within a single process and view for online marketers. Leveraging Autonomy Optimost’s website testing capabilities in a self-service model within TeamSite, the new solution combines the intelligence of web A/B and multivariable testing and analysis with the ease and power of the industry’s leading WCM solution. The solution enables businesses to dramatically improve online results, by accelerating the time-to-web and creating a compelling online experience that engages visitors.  “Analytics, content and testing have made the Web what it is today – rich, interactive, and more intuitive for visitors – yet these essential tools have never been tightly linked,” said Michele Vivona, Senior Vice President Global Web Strategy at LexisNexis. “I am thrilled to see that Autonomy Interwoven has coupled web analytics and multivariable testing with their web content management system. This combination aims to make the process of optimizing web content faster and easier for marketers, which will mean a better online experience for visitors.”

• By combining multivariable testing and web analytics within the TeamSite web content management solution, businesses can intelligently optimize websites to enrich the visitor experience and drive online conversions. Marketers can easily review analytics data on the business performance of their site, quickly create and run tests to determine the optimal site experience, and then deliver the most effective content, all within Interwoven TeamSite. This results in several benefits:

   
   
  • Increase conversions and online business performance through testing rather than guesswork
   
  • Make analytics actionable by putting data directly in front of content owners and reducing dependency on IT
   
  • Accelerate time-to-market to quickly take advantage of new and emerging opportunities that are identified as a result of analytics and testing.

• extensive partner support for the Interwoven Open Analytics Framework. The framework is a key component of Interwoven Segmentation and Analytics, a recently introduced solution that gives marketers a powerful platform to deliver more compelling and relevant online experiences that result in increased conversions, greater competitive differentiation, and more interactive dialogues with customers.
• The Interwoven Open Analytics Framework includes pre-built connectivity with Web analytics technologies

  

  through a component-based page tagging framework. This framework greatly reduces manual processes required for tracking and analyzing customer behavior.
• The combined capabilities enable customers to create and deliver targeted content, offers, and experiences – and analyze and act on the results in real-time – in order to differentiate their brands and accelerate customer acquisitions and transactions. This solution also shatters the myth that delivering a content-rich experience is time-consuming and expensive, and ensures marketers can move with the speed they need to drive top-line revenue growth.
• As a result, marketers can measure the success of online campaigns and offers, apply what they’ve learned in real-time, and continually improve their online presence.”
• “The combination of WebTrends Analytics with Interwoven’s dynamic content management solutions offers marketers a streamlined process to deliver content that is compelling, relevant and targeted to visitors’ specific interests and desires.”

3 more annotations...

• The good thing about Omniture is it can do whatever you want. The bad thing about Omniture is it can do whatever you want. It can be a very complex system
• Taking a page out of the Google Analytics book
   Omniture should take a page out of the Google Analytics book and check out the report designs, navigation, and general user interface of this free product. Omniture is a complex interface for the average business user. Google seems to master the ability to create interfaces that most users can figure out without any knowledgebases, tech support, or training. There is power in this and it’s why Google is good at what they do. If Omniture could start adopting some of this thinking their SiteCatalyst UI, it’d be a great start.

• No disrespect to Omniture’s fine product team, but Visual Workstation is unparalleled for sheer analyst-class power, and I’m fairly sure that without modification Visual Workstaion can leverage whatever format Omniture stores visitor-level data to get up and running quickly. This may cause problems from a pure SaaS-perspective, and I could be wrong, but I suspect that most analysts wouldn’t actually mind having to run the software locally in exchange for having the robust data manipulation capabilities that Workstation provides.
• I believe, at least from a software perspective, that WebTrends is finally getting back on track. I really do believe that WebTrends Score is one of the few true innovations we’ve seen in the web analytics marketplace recently, and learned WebTrends users far and wide have commented that they really like the stuff in the MarketingLab2 release.
• But just as I felt like WebTrends was well positioned (along with Coremetrics) to be a strong solution with a great customer base, a good set of features, that was incidentally “not Omniture”,
• despite believing that “it’s not the technology, it’s how you use it” that it’s nice to see innovation from time to time. For this to happen, I believe we need a strong WebTrends, a strong Coremetrics, and at least a small handful of smaller innovators out there in the world (Nedstat, IndexTools, Clicktracks, etc.) nipping at everyone else’s heels.

2 more annotations...

• I’m certain that Internet marketing has become crucial to this industry however as an SEO, I consider myself more of a technologist than a marketer. Search, at its core, is simply a conduit for people looking for specific information. My job is simply to know how search engines access data so that whenever we have information, it can surface on relevant queries. Beyond that, the feedback loop of search data provides us with a lovely little window into the heads of our audience, what they’re thinking about and what words they use to express this in a search box.
• Duplicate content was one of the first and most challenging problems to address when I came to WSJ. It was like a Hydra…every time I cut off one head, another would grow in its place. We implemented the CLE tag immediately after Google announced it, and so far it has been a huge blessing to me in this regard. It’s still a little buggy here and there, but for the most part I’ve been able to reallocate a significant amount of my time away from dupe content issues.
• ongoing training sessions every month so that new staff, people that missed previous sessions or people that want a refresher don’t have to wait long before another session. I have different training sessions for different groups of people…the SEO for my editors/ writers/ bloggers is different from my session for IT. I also have a session for the business team that talks about understanding SEO in the context of syndication deals or microsites as well as a (less-frequent) session for management that keeps them apprised of the big picture so that I can be sure there is always top-down buy-in for SEO.

1 more annotation...

• There are three free tools available to anyone: Compete.com, Alexa.com and Quantcast.com.
• Compete.com is a panel-based measurement service. This means they take a sample of Internet use statistics (called click streams) from users that have opted in to volunteer that information. Their data sample is from over 2 million U.S. consumers. This includes click stream data from a number of Internet Service Providers (ISPs)
• also includes Compete toolbar users: People who have gone through the trouble of downloading a toolbar plug-in from Compete specifically to volunteer their web surfing data for measurement. There is an inherent bias in this portion of their data because it will skew geek.
• third data point is what they call a proprietary panel. (Because what’s a good sales pitch without working in the word “proprietary?”) This panel is a group of users actively engaged with Compete to provide data, respond to surveys, etc.
• The key thing to know here is that even though Compete.com has data from over 2 million users (more than ComScore and Nielsen, to my knowledge, which use samples of 200,000-300,000) they are still projecting results, a process I believe is inherently flawed
• Still, Compete allows users to compare these projected traffic numbers of up to three websites for free. While the numbers are inconsistent and far from accurate (see below), the tool can be useful for site comparisons.
• What that means is the paid solution is only going to be worth your while if you have an SEO specialist on staff or on retainer to do something with it. (If you need one, let me know. Kat and David are really good.)
• I would recommend Compete as a tool to compare the traffic of websites. And if you’re interested in a fairly economical keyword and search engine research tools, at least in comparison to a lot of solutions out there, it’s worth considering even if the price tag seems a bit disproportionate to the value you receive
• Compete takes out any traffic originating from outside of the U.S. – in other words, if a user comes from India, Canada, or any other non-U.S. country to a site, this visit will not be reflected in our numbers. Additionally, we count people only (i.e. eliminate all bots, spiders, and other automated site crawlers)
• Since our approach is not cookie-based, we will only count that person once, which is a reason why the numbers never match perfectly

8 more annotations...

• 
   Compete.com offers a (third party) Site Analytics service similar to Alexa, where you could view traffic readings for competitor sites.
   
   Unlike Alexa, Compete.com does not rely on a tool-bar. Compete.com partners with major ISPs and calculates most of its readings based on aggregate traffic data as reported by the ISPs.

• WTF happened to Alexa? Remember them, we use to clamor over those #s for our sites and others to prove that we are worth some valuation .
• Their traffic was more accurate and reliable than Alexa. Their data comes partially from ISPs which is actual traffic of sites and not as skewed where Alexa’s came from browser and toolbar installs. Which tend to favor IE users.

• Drupal.org is the official website of Drupal, an open source content management platform.

   

  Equipped with a powerful blend of features, Drupal supports a variety of websites ranging from personal weblogs to large community-driven websites.