-
Web Analytics
Web Analytics, as defined by the Web Analytics Association, is the measurement, collection, analysis and reporting of Internet data for the purposes of understanding and optimizing Web usage.
The goals of web analytics will vary by web site and by company – essentially what analysis is needed to help create the site that helps the organization achieve it’s goals. Those goals may range from share information with many visitors; to gather emails/leads from many visitors; to get the credit card number from many visitors. These goals are most effective when they lead to key metrics or key performance indicators (KPIs). Most vendors discussed in this Knol gather similar data and thus can answer similar questions. However, knowing your specific goals and KPIs will help determine the best type of solution and then the best vendor to meet the needs of your company.
There are three main ways to track web visitors
- Enterprise Logfile Analyzer Software
- Software as a Service (SaaS) Web Analytics Services
- Panel Software
This Knol will explain each method as well as identify the advantages and disadvantages of these methods.
Enterprise logfile analyzer software
Most web servers have the ability to record all their transactions in logfiles. The data in these visitor logs can be parsed and the data can then be analyzed to study the web visitors. The first commercial logfile analyzer was released by IPRO in 1994.
Software as a Service (SaaS)
In the mid 1990s, web counters were commonly seen on web pages. These were images included on a web page that showed the number of times the image had been requested. This concept evolved to include a small invisible image (web bug) instead of a visible one.
JavaScript was introduced to collect information from the visitor’s browser and pass data about the page and the visitor back to a specified server. Just as with logfiles, the data in these visitor logs can be parsed and the data can then be analyzed to study the web visitors. The key difference is that the data is stored not on your web servers, but by servers owned by a web analytics company. You access all your data via a web based interface to the vendor’s servers.
The demand for SaaS offerings occured as web visitor growth occured. Companies were quickly adding multiple web servers and this meant that many different log files had to be aggregated. Also the growth in traffic mean larger and larger log files.
Complex page tagging vendors charge a monthly fee, typically based on the number of pageviews per month. However, the economics of the SaaS service changed in 2005 when Google Analytics begain to offer page tagging solutions for free. Following Google’s lead, Microsoft Analytics is in beta (2008) and Yahoo purchased IndexTools (2008).
Panel Software
Panel software are essentially a large survey of data – not an actual tracking of visitors. Data is collected using a variety of methods principally via:
- Packet sniffing which is based upon looking at the network traffic passing along the Internet. Packet Sniffing is used by some of the largest Internet Service
Provider (ISP) who than sell this data. - Panel data is collected when users voluntarily agree to have all their Internet usage tracked. This typically involves a download of software to their PC or PCs to allow for their behavior to be tracked.
Factors to consider when choosing an approach
Analytics Data Quality- Sample Size
- Logfiles will give you a 99-100% sample of visitors. In theory, the logfiles should get you 100% of the visitor information.
- SaaS tools will give you a 95-998% plus sample of visitors. There
are a myriad of reasons why you will not get 100% of the data from the
way cookies are handled; to the way JavaScript is handled; the way data
is transmitted over the Internet from the visitors PC, to your web
servers to the SaaS vendor’s servers. - Panel tools will give you a 1-5% sample of visitors. By definition
of the way they collect data from select ISPs and/or use a sample panel
you will always be basing your analysis on a sample.
Comparisons / Transitions
- Consider the different ways of collecting data between logs, SaaS
and panels and add in the differences between vendors. Because of
this, it is virtually impossible to compare data across systems.- If you switch approaches – you will lose the historical trending of data.
- If you use different tools on different sites – you will never be able to compare between sites with 100% accuracy.
Data Currency
- How quickly do you need to see visitor counts? – Some of the
systems can provide data in real time, some are on a few hour delay and
some aggregate and deliver the next day. - If you are going to be doing extensive site testing where you try
out new pages/ideas, real time testing can be very valuable as you can
quickly react to ideas that looked better on paper than in reality!
Size of your web site
- How many pages do you have? Using javascript you have to tag every page on your site.
- How many visitors/page views do you have/estimate? Unless you use
a free SaaS service this will impact your price. If you use a logging
solution this will impact your data storage needs.
Privacy
- Is privacy an issue – some companies have rules against external hosting of data. If so – then you need Enterprise logging software.
IT resources
- Are you concerned about minimizing your internal IT time – SaaS is a HUGE value here
- A subset of IT resources is time to manage ongoing software updates
- a SaaS vendor will typically handle that for you and in the
Enterprise logging world your IT team has to manage. - How much data are you going to keep and for how long do you need to
keep it? – For example if you plan on year over year comparisons -
consider the IT requirements of storing all the data in enterprise
logfiles versus the SaaS storing that data for you.
Quality of the vendor
- Anytime you consider working with a SaaS vendor you have to realize
that their fortunes, or more importantly their misfortunes can impact
you immediately. If a SaaS vendor goes out of business – you can lose
access to your data in hours so be sure you trust the business
condition of the SaaS vendor.
Integration
- Do you have email marketing campaigns and if so how is that data
tracked? Will you want to integrate the data with your web analytics? - Do you have search engine marketing campaigns and if so how is that
data tracked? Will you want to integrate the data with your web
analytics?
Comparative Data
- Are you most interested in comparisons to other sites – than Panel software is of huge value.
Reporting
- Do you want reporting and analytic s tools – SaaS products have, by far, built in better reporting tools
Marketing
- Do you plan on doing A/B testing of your site and search engine optimization – here the SaaS vendors will have more support/options than the Enterprise logging approaches.
SaaS Vendors
- Omniture – www.omniture.comLeading enterprise level vendor – particularly with the acquisition of WebSideStory and Visual Sciences. They have a large user community and growing integration with their Genesis plaform.
- Coremetrics – www.coremetrics.com
- Google – www.googleanalytics.com (FREE)Leading vendor in terms of number of users, but mostly tiny sites (blogs) to small/medium web sites.
- Lyris HQ ClickTracks -www.lyris.comA product I was unfamiliar with till writing this summary. Looks to have great integration with the Lyris family of products which include email and search marketing tools. A real strong contendor if you are looking for a suite of products.
- Webtrends – www.webtrends.comWebtrends has it’s roots as a Enterpise log application (which they still offer) but branched into the hosted option as well. Solid product – but the company has had a rough history. They survived an acquisition by a company that nearly put them out of business. More recently much of the senior team departed en masse.
- IndexTools – www.indextools.comBought by Yahoo, in limbo right now as to paid/free future.
- ClickTale – www.clicktale.comClickTale uses a different approach to website analysis and optimization. While the traditional web analytics providers shown above provide aggregated visitor data across web pages, they provide movies of browsing sessions. Similar to the panel software below, ClickTale is best used for sampling. Trying to watch movies of 1 million visitors would be a bit tiresome. Typically, just several hundred user sessions are “recorded” to allow you to see exactly how these visitors behaved – mouse click, by mouse click.
Enterprise Log Analyzer Vendors
- AWstats – awstats.sourceforge.net
- Webalizer – www.mrunix.net/webalizer
Panel Software Vendors
- Hitwise – www.hitwise.com
- Comscore – www.comscore.com
