This session is going to look at the most important information you need for your website, and how most analytics packages don't give you that information.  You have a digital creation on a server which has 2 goals.  First goal is to have the search engines find it and determine relevance rank.  Second goal is to persuade visitor to take action.

Machines are recording all of this.  Everything that is between us and what we want to know is a machines, so we buy analytics packages that give us chartagraphy and data.  They try to give you a broad overview of what's going on in your website, but nothing specific. The only way you can get the right information about your business is to ask questions, because a lot gets lost in translation.

Concepts - Log File vs. Page "Tagging".  The Log file on your server is very portable and another program can help read it.  With Tagging it's a piece of javascript that monitors information that writes information to a different log file.

The log file lives on your server.  Page Tagging lives elsewhere.  Remember with log files, the data is still yours.  With page tagging - all that data is with them, so it's tough to move the data from one program to another and keep the history.

High end analytics programs use page tagging.  Usually they tell you that log files are notoriously bad.  However if you use a browser w/o java script or search engine spiders will not be tracked by page tagging.  There will be no consistency between analytics packages.  In order for an analytics package to define data, it needs to follow a set rules.  The first place these programs get thrown off is the search bots.  So it's really a case that no program is fully accurate.

So you need to really know what you are - log file or page tagging.  They both get the same kind of information.  Who came, when they came, how long they stayed and where they went.  Machines record both human visits and machine visits.

Accuracy - no such thing when it comes to web analytics.

What we can track is Trends.  As long as you use the same package to track the trends you can track trends accurately.  Using different packages every month to get the biggest number you are loosing data integrity.

Unique Visitors - can only be tracked by 2 methods.  Cookies (dropping on a browser) can be session and / or persistent.  Persistent cookies stay until you flush it from your system.

Visitor Session - The number of times visitors view your website (how many pages they saw while they were visiting).  Total visitor sessions made up by 2 groups:  1) visitors who visited me once 2) visitors who visited more than once.

Search Engine Referrals
- The number of visitors generated by search engines.  Search Engine Referral Rate (%) = Search Engine Referrals / Unique Visitors generally if you don't optimize generally you get a 12% referral rate of people finding you on your business name.

Exit / Bounce Rate - The number of users viewing a page and then leaving.

Conversion Rate - The number of visitors who make the intended actions.  You define what the conversion rate.  Additional Details:  Product Literature Downloads, Bookmarked Pages, Contact Forms / Emails / Information Requests, Order / Purchases.  Average conversion rate is about 1.8%.

Its easy to analyze if the visitor only had one thing to do and one way to do it (this is called a Macro action).  Most people don't have this.  We all have hundreds of Micro Actions to take before doing the Macro Action.  Example of this is Amazon - Macro Actions is "Buy the Book".  There are over 26 Micro Actions to get to that one Macro Actions, fortunately for Amazon the next steps are clearly defined.  Lead / Cart actions - see who got to the page, how many go to step 2, 3 .. etc. and who got to the end.  You start to see the holes of where you are loosing people.  These holes should be big red flags. 

Stats are not just about numbers.  It's about finding out how to improve you website.

Pay Per Click Analytics
31% of companies do not track PPC at all.  41% only track click throughs.  72% of business are buying PPC and not tracking it.  Only 11% of advertisers were tracking detailed ROI analysis and this does not include tracking profit!

Hits are irrelevant.  Even though hits are a big number.  Its the rest of the information that's important but not unless you add context, so you need to forget everything you know now.

The problem is we need much more Context.
Visitors who searched for:
Who entered at page:
Stayed on the site for
more than:
less than:
Analytics Pyramid:
Data (foundation of pyramid), Information (middle of pyramid), Knowledge (top of pyramid)

We need wisdom & understanding - no analytics package will give you it.  You need to provide it.  Unless you apply yourself into it won't help.  10% should be spent on the package and 90% is spent on the person to what the data means.

Reporting or Analysis?

"Web analytics works best when measurement expectations are clearly defined in advance, no after the fact"... Mark Peterson

What is the ONE thing you want people to do on your site.  If they do not do that  - then what? 

Do I have Data Integrity?  You need to pick one program and stick with it to track your trends.

Goals:  You Define It!

Segmentation - the principle that "people aren't cows"  we don't move in herds, they move randomly.  Segment on what they are looking for..

Red Shirt Phenomenon!
Crew of 430 on Star Trek.  59 total death on the 5 years.  Red shirts had 72.8% of deaths (43 deaths).  Now that's just numbers... what can we do with that number?

Lets dig in.  If you beam down on the planet, 57.5% of the time you die if you go down with Capt. Kirk.  However if Capt. Kirk meets an alien woman, the rate of survival rate increased to 84%. 

Another Example:
Segmentation is different than what brings the person into your site.  Example:  User types in "digital camera" and came in on that, but they are thinking about megapixels, brands, price, size, etc.  Someone types in MP3 - but they are also thinking of ipod, flash drive, cost, gigs, etc.  The thing now to realize is that the conversion rate for each of these is different because the segmentation is different.  Get rid of "overall conversion rate" for a site.  Segment on the phrase!  Then segment further!  But just by segmenting on a first few intial keywords can help you understand if users are finding what they are looking for.

3 C's of Analytics
Context, Comparison, & Contrast.
Context of what people are looking for, who's coming, where they are coming for.
Then do a comparison & contrast.  What's different, what's the same.  Don't care about the industry average or competitor.  Your segments tell you more about you!

Numbers without context are just number.  More data points give more context. 

Building the context.  You need to look at the stats (time on site, pages viewed, conversions, goals), but look at it by Segment (blogs, keywords, referrals). 

6 Initial Reports - by Segment
1.  Conversion Rate
2.  Top keyword "groups" from Search Engines
3.  Top Referring URL's
4.  Content Popularity
5.  Clink Density / Overlay
6.  Bounce Rate

If your analytics can't do segmentation, dump it and get one that can.  Matt says, he can guarantee this will pay for itself in 3 months, if you find one that can segmentation.  Forrester survey showed that companies had an 900-1200% ROI increase by investing in an analyst.

You need to get a value on the cost of the lead.  If you don't attach the value, you don't know what the "R" is.  You need to attach a value to actions taken on your website.

1.  Start with a Question
2.  Segment by Verticals & Acquisition Method
3.  Measure each Segment
4.  Compare the Results
5.  Put the Result in context
6.  Focus on Business Value

The worst thing you can do is have this information and not do anything.  Use this information.

April 21, 2008

Liana "Li" Evans is the director of internet marketing at KeyRelevance Search Engine Marketing. She oversees all social media, online PR and word-of-mouth marketing efforts for a variety of clients within the company’s portfolio. Li also collaborates with the team on all SEO (Natural Search) optimization for clients. Since 1999 Liana has been active in the search marketing arena, becoming well versed in all avenues of search marketing but with particular focus on natural search, image search, and social media. She also has substantial experience in areas of the retail industry regulated by the FTC. Prior to Li’s tenure at KeyRelevance Search Engine Marketing, she helped to design, plan, and implement an Internet Retailer 500 company’s efforts into natural search optimization, successfully revamping outdated navigation and site architecture. Li is an accomplished programmer and database programmer/designer familiar with the demands of large-scale, dynamic retail sites.

Liana holds a B.A. in Information Systems with a minor in Public Relations from Susquehanna University, and an Associates Degree in Mass Communications/Public Relations from Pennsylvania College of Technology. She currently lives in the Philadelphia suburb of Limerick, PA and runs the search marketing blog Search Marketing Gurus.


Looked like a great session. Even learned a little. The PPC stats are scary. These people are spending money and not even checking to see where the money goes. I guess that just means more conversions for those of us that really get our searchers and customers.

Dear Matt,


The article is very informative and challenging.. Could you kindly send me some details about Business Analytics.

Thanks & regards,


Comments closed after 30 days to combat spam.

Search Engine Guide > Liana Evans > Unleashed: The Joy of Analytics by Matt Bailey (Live Blogging)