Optimizing the Online Business Channel With Web Analytics

Optimizing the Online Business Channel
With Web Analytics
Christopher McFadden
Page 1 of 19
Published: 2005
Author contacts:
Christopher McFadden
@cristoirmac
Abstract
The Web has gradually evolved from a great experiment to a legitimate business
channel. Web analytics has concurrently progressed from simply measuring visits, page
views, and revenue to a reliable process used to optimize a web site. This paper discusses
some of the fundamentals of Web analytics and how to align measurement with business
goals through the use of Key Performance Indicators relevant to the type of web site.
Some practical considerations are discusses including key best practices, top challenges
for implementations as well as some tactical ways to use Web analytics.

Optimizing the Online Business Channel with Web Analytics
Introduction
What is Web Analytics?
Web analytics refers to the collection, analysis and reporting of Web site usage by
visitors and customers of a web site. This information is used by those responsible for
the success of the web site to better understand the effectiveness of online initiatives and
other changes to the web site in an objective, scientific way through experimentation,
testing, and measurement. This understanding and knowledge can be used wisely to
optimize the web site so that it more effectively accomplishes the goals of the business.
The optimization process can occur in any number of areas such as site content and
media offerings, product and merchandising, site navigation, creative design, internal
search and the checkout process.
As the preceding brief definition indicates there are many facets to Web analytics.
The purpose of this paper is to provide an overview of the fundamental concepts as well
as some of the more practical considerations of Web analytics to ensure that those
responsible for the success of a web site – web masters, marketing, merchandisers,
content producers, and IT – are able to get the most out of their Web analytics. The focus
of Web analytics should be on optimization rather than simply measurement. In the not
so distant past Web analytics was considered to only refer to the analysis of web site log
files to determine basic web traffic data including the number of visitors to the site and
the number of page views. This paper intends to provide some insight into how to go
from measurement to optimization by aligning metrics and analysis with the strategic
goals and vision of the business.
The Web as Businesses Channel
The web has progressed from a cool experiment of the 1990s to a strategic business
channel. The web is now seen as a key revenue driver and a reliable way of attracting
and retaining customers. According to a recent study by Shop.org online retailing will
grow by 27% and almost eight out of ten online retailers will be profitable, regardless of
the retail category. Sites are learning how to increase operating margin and increase
conversion rates (Shop.org, 2004). Sixty percent of top online marketers expect their
total marketing and media budgets will increase in 2005. (Millard Brown, 2004). Fourth
quarter spending in 2004 increased 25% over the previous year to $23 billion (Harris
Interactive, 2004). Businesses now realize that their web-sites are no longer something
they should have simply because everyone else is doing it. To the contrary, the web is
now a strategic business channel that must justify its existence by demonstrating value
for the bottom line. According to David Keller and Josh James this shift of the web from
independence to integrated business channel has brought Web analytics to the forefront:
“Businesses now realize that they must assess the economic efficiency of their
online initiatives. The Internet as a business medium is no longer independent in its
Chris McFadden Page 2 of 19 11/26/2014

role within the corporation; it is now a viable source of revenue and customer
relationship management. And, as such, it must be viewed as any other component of
the business model – one that must be measured and validated in its worth. It is
within this framework that Web analytics provides its greatest benefit.” (Keller,
2003).
Without Web analytics most aspects of site design, function, and organization are
based on the subjective judgments of those responsible for the web site. This was the
way the web was for many years. In some cases site design was seen as a strictly creative
process. With this mode of thinking the goal was to have the site as visually pleasing as
possible. The subjective “cool” factor was how the quality and success of the web site
was measured. For other sites there was an emphasis on functionality and therefore it
was important to have as many buttons or links on a page as possible – the more the
better.
As the web has become a bone fide business channel it has become necessary for web
site managers to make the gradual change from subjective to objective measurement of
success criteria. It is apparent that it is difficult to determine what needs to be improved
and what is working well without having an accurate and objective way to measure,
analyze, and report on user behavior. According to an E-tailing Group Inc 2003 study
14% of respondents to their survey of online retails did not know their conversion rates
and 43% didn’t know their shopping cart abandonment rates. The most frequently used
measurement methods were based on sales results and site activity/traffic reports, many
of which are overly simplistic in nature. (Eisenberg, 2004) It is apparent that while web
has become an increasingly important business channel for ecommerce, ad sales and lead
generation not all businesses have a deep understanding of how customers use their site
and how a site can be more effective at meeting the needs of the customer and the
business.
A Brief History of Web Analytics
The Web analytics market got its start soon after the introduction of the first web
browser in early 1993. Since the beginning there were two basic types of analytics
solutions offered:
• Web log analysis software
• Browser tag based analysis services
The business model and data collection methods for these two solutions are significantly
different. Web log analysis software vendors sell software that can be run in-house
against the voluminous logs that are generated by the web-site. Reports are generated on
what is usually a daily basis. Browser tag based solutions are externally hosted
applications and they follow an Application Service Provide (ASP) business model.
During the period of rapid growth in the Internet sector there was a corresponding
growth in the Web analytics market. In 1999 the Web analytics industry saw growth of
200%. The predominant technology was the enterprise-class software solutions that were
used to analyze huge web server log files. Some of the leading vendors included
WebTrends, NetGenesis, and Accrue. The market for ASPs was limited to mostly small

and mid-sized businesses. Coremetrics and Websidestory served this small market,
among others.
With the bursting of the Internet bubble the Web analytics market experienced
dramatic changes. In 2001 the Web analytics market actually shrank by 7%. The
numbers of vendors in the market was almost cut in half. Many went out of business
while others dropped their offerings. The ASP solutions also became more attractive
since they offered an “on-demand” or “pay as you go” service offering that could be
classified as an operating expense and which negated the need for large infrastructure
investments and support staff. The ASP solutions also offered advanced web-based
reporting interfaces. Web-log software-centric vendors have since almost completely lost
the hold they previously had on the market. Additionally, since 2003 the remaining
leading ASP vendors - such as Coremetrics, Omniture, Websidestory, and WebTrends –
have adapted their offerings to include the best features offered by their competitors such
that they have achieved near feature parity. (Ballardvale, 2004a)
Web Channel Performance Management
Aligning Metrics with Business Goals
Web analytics solutions today produce an abundance of information on web site
usage. It is important for web site managers not to be content with simply using the Web
analytics tool to generate a multitude of reports for their own sake. If a metric in a report
does not correspond with the strategic goals of the site it is worthless and can in fact
distract from what is truly important.
A successful web site has a clear purpose and vision that all responsible for the
web site are conscious of and constantly striving to achieve. All members of the web
team can work toward this common goal even though they may be engaged in separate
activities such as: site design, content creation, marketing, advertising, and
merchandising. The web site and the business must be viewed holistically together. The
web site does not exist independent of the business – it is not a new business model but
rather a new mode of delivery and communication. As such the time-tested rules of
business apply as much as ever. The web site must have clearly defined and understood
vision, strategy, goals and tactics. (Sterne, 2004a)
This gets to the heart of the problem with many Web analytics implementations
that provide little benefit to the company. If the reports are simply giving a snapshot of
what has happened in the past there is only data – no knowledge. According to the Web
analytics guru, Jim Sterne, no matter how much data is collected and no matter what
collection method is used “all of this is useless unless it’s actionable – and that means (1)
the business goals must be clear, (2) technology, analytics, and the business must be
aligned, and (3) the feedback loop must be complete”. (Ballardvale, 2004)
Johnathan Becker makes a case for what he calls “Alignment-Centric
Performance Management”. He says metrics in use today mostly reflect past behavior
and have limited use in helping companies improve performance. An alignment-centric

performance management solution calls for the distillation of the number of metrics to a
few Key Performance Indicators (KPI) that help identify and track high-level progress
towards strategic goals. The four key aspects of his approach (the four Ms) are as
follows (Becker, 2005):
• Motivate – communicate organizational objectives in a way that is
relevant and actionable to everyone.
• Manage – provide strategic context and encourage collaboration on the
milestones critical to organizational goals.
• Monitor – track progress via KPIs explicitly linked to objectives;
proactively resolve problems and seize opportunities.
• Measure – drill deeper to identify the root of issues and test assumptions
inherent in the strategy.
Daniel Sapir outlines some of the expected high-level benefits to the business of
what he calls Online Business Performance Management (OBPM). The most important
benefit is the ability to make better and faster decisions and the ability to handle changing
conditions better. The business managers will have more confidence in the analysis and
reports that results. They will also be more accountable since they will be more actively
involved in the establishment of tactical their goals and the metrics used to track
progress. In addition, due to clearly established processes and automation it is possible to
make available high quality information throughout the business in a very timely manner.
(Sapir, 2004)
The lesson to be learned is that measurement is simply a means to an end.
Without tying measurement to business objectives there is only data, no knowledge and
little business value. Aligning the business goals and measurements together will help
the organization keep its eye on the prize and consequently achieve its strategic
objectives.
Key Performance Indicators (KPI)
KPIs are quantifiable measures that help decision makers define and measure
progress toward organizational goals. KPI metrics translate complex measures into a
simple indicator that allows decision makers to assess the current situation and act
quickly. For KPIs to be effective, however, they need to be actionable – meaning they
should report on the metrics that matter to the organization. KPIs help decision makers
define and measure progress toward organizational goals. KPIs should also establish an
expectation for performance by using business-relevant comparisons over time.
(WebSideStory, 2004a) According to Eric Peterson of Jupiter Research, KPIs are metrics
that allow business managers that are not directly involved with the operation of the web
site to understand if the web site is contributing to the businesses bottom line. That
management is able to understand these KPIs means that the Web analytics becomes less
of an IT tool and more of a strategic business driver. (Ballardvale, 2004)
Additionally, a valuable KPI is capable of revealing a meaningful change in
activity for a selected time period. As Allen Crane of Dell Computers said “The Holy
Grail is a suite of quantifiable, actionable e-metrics that capture behavior patterns and
accurately relate them to the key transactional business levers of units, revenue and

margin”. The use of KPIs is really the only meaningful link between the reports
generated and the goals of the business. (Eisenberg, 2004)
To identify the most relevant KPIs for a web site it is important to understand
what type of site it is. Many web sites have multiple purposes, which can make it
difficult to identify how to measure the success of the site. For example a software
company’s web site may include an e-commerce element to support direct sales of
software to consumers. It may also contain content relating to the industry. It may
generate leads to be followed up by the sales force. In addition, there may be an on-line
knowledge base that allows customers to get answers to support issues. KPIs can vary
drastically depending on the industry and the type of web site. Most web sites can be
classified in one or more of the following site types (WebSiteStory, 2004a and
GrokDotCom, 2005):
Commerce
In this environment, the goal is to get customers to buy some goods or services
directly on-line. Examples of these sites are Discovery Channel Store Online, Amazon,
eBay, and Expedia.
Online retail sites have always been interested in the bottom line – how much
revenue does the site generated. Revenue is usually the primary objective of the site.
The KPIs for a commerce site should include the following
• Conversion rates: there are multiple ways to measure conversions. The most
common is order conversion rate - the ratio of visitors to orders. Equally
important is checkout conversion rates – the ratio of people that start checkout to
orders. One of the goals of advertising and marketing is to drive traffic to the site.
It is important to benefit from this traffic by actualizing real revenue. When
tracked at the individual marketing promotion or campaign level it is possible to
see which promotions are working and which are not.
• Average Order Value: this is the ratio of revenue to orders. This is greatly
dependant on the product mix of the site and the merchandising that encourages
customers to buy certain things or to buy more than one item. The AOV has an
impact on profit margin therefore it is important to entice customers to buy more
expensive items and to buy more than one item at a time.
• Visit Value: this is the ratio of all visits to revenue. This is also an important
benchmark in measuring the effectiveness in driving qualified traffic to the site.
• Customer Loyalty: the ratio of new to previously acquired customers.
• Stickiness: the ratio of visitors that arrive on a key landing page, such as the
homepage or a promotional page, to those that stay on the site (go to a subsequent
page) or make a purchase.
• Search Engine referrals: the ratio of referrals to the site from search engines such
as Google and Yahoo compared to industry average. It is also possible to track
referrals and conversion rates for paid search terms.
Lead Generation
In this model, the goal of the site is to get visitors to submit their contact
information so that the company’s sales representative can follow up with them later.
Examples of such sites include New York Life, WebSideStory, and Siperian. These

companies focus on sales of specialized products and services that are not purchased
directly online.
The KPIs for a lead generating site can be somewhat similar to a commerce site.
Both are conversion driven and as such they have some KPIs in common. The objective
of a lead generating site is to get visitors to provide their information so they can be
solicited at some later date. Some useful KPIs are:
• Conversion rate: the ratio of site visitors to leads generated. Leads may be
captured in the form of direct requests for a follow up sales contact or via requests
for content downloads, signing up for mailings, and subscribing to newsletters.
This is especially important to track at the campaign level to see which
promotions work and which do not.
• Cost per Lead (CPL): the ratio of marketing/promotional costs incurred to leads
generated. This is important in measuring the success of promotions.
• Single Access Ratio: the ratio of single access pages to entry pages. This
measures the effectiveness of getting people past the entry pages.
• Traffic Concentration: the ratio of visitors to a page or content area to overall site
visitors. This is a good indicator of which areas of the site are generating the
most interest.
Content/Media
Content sites are generally based on the on-line advertising business model.
Alternatively, the purpose of content sites is to build affinity with the off-line content
(TV or print), which results in more off-line ad sales. Examples of such sites include
ESPN.com, Discovery.com, WebMD, NewYorkTimes.com and Fox.com. The goal is to
keep visitors coming back and to stay longer on the site. Content should be engaging and
be updated frequently.
Since the objective of content sites is to increase readership (to bring in more ad
revenue) the level of interest in the site and the length of stay is of most importance.
Some of the most relevant KPIs are as follows:
• Depth of visit: measures the ratio of page views to visits. This measures the level
of interest in the content provided. The more engaged visitors are the more page
views consumed – directly affecting ad revenue.
• Returning Visitors: the ratio of visits to unique visitors. This measures the
effectiveness of the site at bring visitors back to the site. This is also one factor in
determining customer loyalty.
• New visitor percentage: the ratio of new visitors to unique visitors, which
measures the success of the site in attracting new viewers. This can mean
different things depending on whether the site is new or attempting to attract large
number of new visitors or is more established and is focused on retention.
• Page/content depth: the ratio of page views to visitors at the page or content level.
Key site content should be measured separately to measure interest in content
areas over time. This can also be critical in determining the navigational
effectiveness of a single page or content area.

Support/Self-Service
The support or the self-service model has a goal of providing customers with the
ability to find the answers they need regarding their products or services. There are two
business goals - to increase customer satisfaction and decrease call center costs. Some of
the KPIs that can help measure the success of the site in meeting these goals are as
follows:
• Customer satisfaction metrics: these are collected through use of actual online
surveys or can be collected by allowing users to rate support content.
• Length of visit and content depth: in contrast to media sites the goal should be to
make it as quick and easy for a user to find the information they are looking for or
to perform the self-service task.
• Percent of visits under 90 seconds: the site should be efficient but not difficult to
use. A high percentage here may indicate that users get frustrated and cannot find
what they are looking for.
• Top internal search phrases: not a traditional KPI but important in identifying
areas that may require better navigation or more content to meet customer
demands.
(WebSiteStory, 2004a and GrokDotCom, 2005):
Practical Approaches to Web Analytics
Key Best Practices
As the industry continues to mature there are still many web sites that are
continuing to have trouble getting value out of their Web analytics. At the Emetrics.org
conference in 2004 the best and the brightest of the field shared their stories of success
and failure. Ballardvale Research provides an excellent summary of some of the best
practices to come out of the conference. Jim Sterne also has many of his own valuable
practical insights into how to get the most out of Web analytics. The following is a
synthesis of these best practices, with some additional observations. See (Ballardvale,
2004) and (Sterne, 2004b) for more information.
Identify Key Stakeholders
One of the first steps is to identify all those that have a vital interest in the success
of the web site. This group of people will not be from any one group but will consist of
online content producers, merchandisers, site developers, online advertising sales,
marketing, research, etc. Web analytics should not be something that is forced upon
people but should be seen as a vital new tool to help improve the business processes of all
parties.
Additionally, a key contributor to success is a “Chief Web Officer” – a single
person responsible for a company's online presence. This person will be in charge of
establishing the web site’s vision and help resolve disputes between the various
departments. Without such a role in the company the web site will fluctuate between

chaos and paralysis due to committee style leadership. The person in this role must take
responsibility for the web site and be duly held accountable.
Define Primary Goals for Web Site
Sometimes the biggest hindrance to optimizing a web site is not having clear
goals and directives. It is important to document what the key stakeholders see as the
primary goals for the web site from the point of view of the corporation, the department,
and even as individuals. The list of business goals may include any combination of the
following: improving revenue (selling more products or advertising), cutting expenses,
improving customer service, and building brand loyalty. There will invariably be
political considerations when it comes to evaluating whose goals are most important. It
is necessary to identify which goals are important to which people and be as clear and
honest as possible about the business value of goals.
All goals and objectives should be put on the table so that they can be discussed
and prioritized with the end result being a clear set of goals that everyone accepts as valid
and worthwhile. Without doing this there will be too many unstated assumptions and
conflicting business goals preventing the identification of meaningful metrics and KPIs.
Without common goals there will also be a lack of synergy between the different groups
and individuals responsible for the web site. However, if the organization is able to agree
on what is important it makes the entire Web analytics process much simpler with a
higher probability of success.
Identify the Most Important Site Visitors
This may not be so obvious but for many sites a relatively small base of visitors
bring a disproportionate amount of value to the business. The visitors that are the most
valuable over time are the site’s bread and butter. These might be the people that come to
the site the most often, generate more page views, spend the most money, or sign up for
the most newsletters. As a rule of thumb, if the web site managers can determine what
these people want and try to make their experience better then the experience of all
visitors will improve. A common way to better understand who these people are and
what they are doing is through segmentation.
Determine the Key Performance Indicators (KPI)
Once the goals of the web site have been clearly defined and agreed to by all
stakeholders it is necessary to identify which metrics can be used to track whether the
web site it getting closer to those goals or is moving further away. Each goal needs to
have a set of clearly defined KPIs that can measure progress toward the goal over time.
It also important to reiterate the need to only report on what is important and to
ignore those things that are not. As a general rule, if a report is not actionable it does not
contribute any business value and therefore should not be produced in the first place. As
with many things, the focus should be quality not quantity.
See the preceding section on Key Performance Indicators for more detailed
information on this topic.

Identify and Implement the Right Solution
If he company knows what the goals for the web site are and has a list of KPIs
identified to measure the success of the web site in achieving those goals, selection of the
right technology and vendor becomes a lot easier. The factors in evaluating the right
vendor or solution should come down to a few basic considerations:
• Does the solution capture, track and report the necessary metrics?
• Does the tool provide ways to drill-down deeper to better understand the
contributing factors – such as path analysis and segmentation?
• How does cost compare with the budget? Be sure to have allocated a reasonable
budget to ensure a successful project and be aware of the total cost of ownership
(TCO).
• Flexibility to meet unknown future needs.
• Ease of use and other user interface issues.
It is always a good idea to conduct a pilot test of at least the top two solutions before
final selection. This way it is possible to get a better understanding of the
appropriateness of the solution for the particular web site and business.
Use Multiple Technologies and Methods
Most Web analytics packages today provide powerful analytics and reporting
based on click-stream data. However, that will ultimately only give you a piece of the
puzzle when it comes to understanding what works and what does not. The most
successful web sites also use more subtle approaches such as focus groups, online
surveys, usability studies, and customer services contact analysis to better understand the
way customers use the web site and to constantly find ways to improve the site. These
disparate analysis methods, if done in a rigorous way, will yield a more holistic view of
how the site is performing and where it can be further optimized.
Make Improvements Iteratively
As a general rule among the best web sites, the only way to know what adds value
and what does not is to make changes to the site gradually. If after a full-scale redesign
of the web site there are noticeable improvement or things are actually worse it is very
difficult to pinpoint what specifically caused the change. It is a common best practice to
change only a small set of variables at any given time while measuring changes in KPIs
over time. This evolutionary approach leads to the gradual and steady improvement of
the web site.
The opposite approach would be to spend huge amounts of time and effort on
implementing a site redesign without any way of knowing if it will help the web site
accomplish its goals or not until after the redesign is rolled out. In the case of a dramatic
redesign consider at least performing some A/B testing before rolling out the changes for
the entire site.
Hire and Empower a Full-Time Analyst
It can be hard to become an expert at Web analytics if that is a part-time role. It is
difficult to be looking at the big picture if there are many other high priority tasks that

need to be addressed on a daily basis. For that reason it is recommended that a single
person be assigned as a full-time Web analyst.
According to James Maguire “The biggest Web analytics decision you face is,
who it going to be responsible for it? What set of eyes is assigned the task of following
the data, week after week, sharing it with the proper personnel”. (Maguire, 2004). He
further outlines (with input from Eric Peterson of Jupiter Research) some of the qualities
of a good analyst:
• Understands the business needs and can communicate well with all the
stakeholders – web site designers, marketers, merchandisers, and the IT staff.
• Has a good grasp of both technology and marketing – with basic web skills
sufficient enough to understand how things are put together as well as a good
head for business and marketing.
• Is capable of leading detailed analysis – must be logical and analytical with
desire to find the deeper meaning in the data. This involves more than simply
running reports.
• Has respect, credibility, and authority – the analyst must carry enough weight
in the organization or report to someone that does, such as the Chief Web
Officer.
• Is an existing employee of the company – since Web analytics is still
relatively new it may be more effective to train someone already familiar with
the business then trying to recruit a hard to find veteran analyst.
Peterson, an expert in Web analytics, states that some companies are hesitant to hire a full
time analyst but he believes that “if they’re the right people, they’re smart, and you
empower them, the return on investment should be very strong”. (Maguire, 2004).
Establish a Process of Continuous Improvement
It is easy to implement a Web analytics solution and then forget about it. The
hard truth is, however, that successful Web analytics does not run on autopilot. Even if
the tool is collecting all the right metrics and has the best user interface in the world it is
worthless if nothing is done with the acquired information. Web analytics must be
viewed as a process, not a tool. If there is no accountability and responsibility and no
conscious effort to establish and follow a process of continuous improvement the value to
the business of Web analytics is questionable – if not actually a waste of time and money.
Web analytics should not be relegated to something that needs to be done once in a while
or a way to get some numbers at the end of the month (in case anyone asks).
Instead, every change to the web site, every design decision, every new feature
should be analyzed in comparison to what the actual state of the web site is and what the
expected future impact on the site will be. Once the change is made it is necessary to
measure the impact of the change and determine how it compares with expectations. Did
the change make things better or worse? Did it add value or not? Did it contribute in a
positive way towards meeting the business goals of the web site? Eric Peterson states
that “fundamental to the continuous improvement process is the notion that no changes
are made on the Web site without a reason for making the changes and expectations
about what effect the changes should produce.” (Peterson, 2004)
It is only though this process of hypothesis, experiment, testing, and checking
results that Web analytics can help web site managers use scientific methods to

constantly improve the web site. Jim Sterne provides some insight into the importance of
establishing a solid process:
“Liberte, Egalite and Fraternite may have fueled the French Revolution, but the
more mundane accountability, responsibility and integrity will determine whether
your Web analytics efforts are going to pay off….
“Don’t go through all this effort just to say ‘Yes, we do Web analytics and we
have the reports right here to prove it!’ Instead, make sure those reports are an
integral part of a process of continuous improvement. Then you’ll know whether
the Web site is working or not.” (Sterne, 2004b)
Top Challenges
According to industry practitioners gathered at the 2004 Emetics Conference there
are a number of key problems and limitations of Web analytics that should be recognized
and accounted for both before and after implementation. These issues are elaborated
below (Ballardvale, 2004):
Site complexity
The more complex the site the more difficult Web analytics becomes. If the site
does not have a small set of clearly defined goals or if fits into more than one of the four
basic types of site (lead generation, commerce, content or self-service) it may become
increasingly difficult to determine which KPIs are relevant.
Complexity will also make it more difficult to identify appropriate customer
segments. Usability and accessibility may be more difficult to analyze since there may
be many different paths users can take and different ways users can relate to the site. If
there are fluctuations in usage patterns over time or other aspects of the site change may
change frequently it may be difficult to establish necessary benchmarks.
Additionally, if the site content is not organized in a logical way or if it does not
have a documented information architecture it makes it exceedingly difficult to actually
measure which page is which and how a page relates to another page. Any useful path or
content analysis becomes intractable.
Technology and Training
Although Web analytics technology has matured there are still some hurdles to be
overcome. It is a common assumption that Web analytics tools have all the answers.
However, without the strategic business planning required the tools will not provide the
promised benefits. Even though user interfaces for the tools has made significant
advances it is still critical that business users and web site managers get adequate training
so that they can derive the maximum benefit.
Additionally, contrary to vendor promises, implementations can require
significant effort to ensure that the right data is being captured. It is important to
understand that it is best to phase in the implementation rather than trying to accomplish
everything at once.

Resources
Since many Web analytics solutions are provided as a service there is often the
misunderstanding that few additional resources are required. There is always needed
someone to implement the instrumentation or tracking tags on the web pages. It also may
not be sufficient to simply add a tag to track click-stream data. Usually, additional
changes are needed to get more valuable data on certain pages. Additionally, in order to
derive benefit from the Web analytics solution it is important to have at least one
educated and experienced analyst that can examine data. The web managers must also
allocate enough time and energy to using the knowledge acquired towards optimizing the
site. If the end result of Web analytics is reports and not action then there is little benefit.
Politics
Political or management problems can sometimes be the biggest roadblock to
success. This may be in the form of inaction on the part of management, lack of
education and backing by corporate leadership, or the fact that there is no one ultimately
responsible for the web channel. It is also problematic if the web site business
management views Web analytics as a responsibility of the IT department. Since Web
analytics is primarily a business analysis tool it must be implemented with the goals of
the business in mind, not IT.
Metrics
Often there are serious problems getting metrics aligned with the strategic
objectives. Sometimes the most difficult problems resolve around identifying the most
suitable KPIs for a web site. This a hurdle that must be overcome before Web analytics
can help the web site managers optimize the site. It can also be difficult to establish and
maintain clear definitions of KPIs that everyone can agree on – for example what is
meant by conversion? What is the definition of revenue?
During the transition period legacy metrics can be problematic because people are
used to the old measurements, as insufficient as they may have been. It is important and
sometimes difficult to get rid of these legacy metrics. Another problem is how to
properly measure the value of content. Content sites are not as easily understood as
retails sites that have very much more defined purpose. In this case even more thought
must go into defining the sites goals and the key metrics to track progress in meeting
those goals.
Data Integrity and Quality
The accuracy of the data is still a key concern. There is certainly a need to
manage users expectations of accuracy as well as actual data collection problems. Such
things as client/proxy caching, web spiders, visitor identification errors, etc easily throw
off web log analysis. Accuracy of ASP browser tag solutions is dependent on client side
cookies, which people may not have turned on or they may clear on a regular basis
(McGann, 2005). Ad blocking and anti-spyware software is also becoming more of a
concern since this may interfere with the tracking. Due to the nature of the web, which is
inherently indeterminate, there are always going to be some level of inaccuracy. This is
an area where solution providers can continue to improve upon.

It is also difficult to get a consolidated view of the customer across channels, and
to integrate external data sources, such as surveys, to get a more complete view of how
the web site is performing. Without a compete view of the customer from all touch
points it might not be possible to completely understand the needs and wants of the site
visitors.
Tactical Uses of Web Analytics
There are a number of practical ways to use Web analytics to get a better
understanding of the web site and how to optimize it further. The following is a
summary of a number of these specific tactical approaches. In most cases it would be
necessary to do more research and training on each topic before it would be possible to
understand these tactics enough to implement effectively.
Visitor Segmentation and Clustering
Segmentation isolates the behavior of certain types of important types of online
visitors. This can help in the analysis and understanding of how important groups of
people use the site by looking at the page-by-page activity of the group without being
distracted by activity from other types of users. This method can be used to identify who
is buying, browsing, staying the longest, coming back the most often, visiting particular
parts of the site, viewing certain content or using certain features. It can also isolate the
behavior of visitors based on how they get to the site, be it a special marketing campaign,
a search engine, or a partner site. Segmentation can also be performed based on
demographics such as geographic location, gender, age group, and whether the user has
broadband connectivity (some of which depends on the collection process for this
information). For more detailed information on customer segmentation and clustering
see (Lenzen, 2004).
Cross-Sell/Up-Sell and Merchandising Optimization
Web analytics can help in identifying the optimal merchandising offers for
product pages, the shopping cart as well as email offers. It is relatively straightforward to
determine which products people most often purchase together. It is also possible to
measure the performance of merchandising choices to determine which site design and
navigation scheme contribute the most to maximizing merchandising offers.
For content sites it is possible to do similar analysis to determine which content
areas have the most affinity with others. It is generally a good idea that if people move
from one content area to another to make it easier for them to do so. This will keep
visitors on the site longer.
Fall-out (Funnel) Analysis
For many sites, most particularly ecommerce sites, one of the areas of most
concern are getting people to “convert”. For retail sites this means getting visitors to add
to the cart then proceed through each step of checkout until the purchase is complete.
Conversion can be described more generically as measuring and analyzing the rate at
which visitors go from page A to page B. For non-retail type sites there are many areas

where conversion is worth analyzing. For example, in analyzing the behavior of people
that use an internal search tool the ratio of people that search to people that click on a
results page would be valuable. Another example would be analyzing the ratio of people
that click on a link to register to people that complete the registration. Any “success
event” that adds value to the business can be analyzed by looking at the steps it takes to
get to that success event and how many people fall out at each step along the way.
Some important ways to do this is through Fall-Out or Funnel analysis and
reporting. All leading Web analytics packages offer this capability. Pick the entry page
for the process you want to analyze (or start with all site visitors). Then pick the
intermediate steps along the way and include the final success page or event. The fall-out
report will graphically represent the rates at which people are dropping out of the process
at each step along the way. By keeping a close eye on this it is possible to determine
what impact any changes you make have on the conversion rates for your key processes.
A/B Testing
A/B or split-run testing got its start with direct mail marketing. It is very easy to
send two slightly different marketing pieces – coupons, trial offers, catalogs – to
randomly selected samples of customers and track how the two marketing pieces perform
in relationship to each other. By using slight variations it is possible to determine which
minor differences make the biggest difference. A/B testing is a natural fit for the web
since small variations are easy and it is possible to randomly present the variations to
visitors and measure the results. Any part of the web site can be A/B tested to determine
the effectiveness of changes – the homepage, landing pages, product pages,
merchandising campaigns and promotional offers. See (Anderson, 2004) for more
complete information on this important topic.
Optimizing Internal Search
According to DoubleClick more people are using the internal site search to find
what they need. On ecommerce sites there was an increase in the number of sales that
came through the internal site search from 6.6% in Q3 2003 to 9.3% in Q3 a year later.
Conversion rates and average order value for those visitors were also up dramatically.
(DoubleClick, 2004) Internal search is critical not just for ecommerce sites but it is also
very important for self-service sites and content sites. For any site it is important that
visitors are able to find the product or content they are looking for. If they are not able to
find what they are looking for the most common result is a lost customer – and lost
revenue. Web analytics can be used to closely monitor how internal search is performing
and to analyze how changes to search affect the performance of the site.
Search Engine Optimization (SEO)
For many web sites the largest amount of traffic may be coming from search
engines such as Google, Yahoo, and MSN. Some studies have shown that more than
70% of Web traffic comes straight from these search engines. Referrals may be either
from paid advertising placements on these sites or from so-called organic search. The top
ten search engines are responsible for 90% of all search referrals. Also it has been shown
that half of all search engine users will only click on the top ten results. (Obrey, 2004).
There are two basic approaches to increasing the referral rate from search engines. The

first is to use marketing dollars effectively to buy the keywords that will generate the
most qualified traffic to the site via paid search. The second is to optimize the site for
organic search by ensuring the site gets indexed and highly ranked by the search engines
– if the site does not get indexed or appears on page ten of the results there will be few
referrals. Web analytics tools can be used to manage and measure the success of your
efforts. There are vendors that specialize in helping sites get the most out of search
engines referrals – just be sure to check references to ensure that these vendors are able to
add real value.
Optimizing Homepage and Landing Pages
As important as it is to get people to come to your site is the need to ensure that
the key landing pages of the web site draw people into the rest of the site. It has been
shown that only 20% visitors coming from search engines will click through to the next
page. (Obrey, 2004) Consequently, much analysis is required of the homepage and any
of the other top entry pages into the site. Web analytics can be used to measure and
analyze where people are coming from, which pages are the top landing pages and where
people go after they get there, including those that go no further. By optimizing these
landing pages there is the possibility for a huge return on investment. Some ways to
optimize landing pages are making it clear that what the person was searching for can be
found at the site. Sometimes this is not so obvious to a visitor. By calling out in bold
ways the content that the person was looking for the visitor is more likely to go deeper
into the site with the potential of either completing a purchase or generating more page
views and ad revenue.
Online Marketing Performance Management
Any money spent on ads on external affiliate or partner sites can result in a much
greater return on investment by closely measuring and analyzing the effectiveness of
which banners generate the most value, which sites and campaigns produce the most
revenue, and on the reverse side which do not. It does not make sense to blindly spend
money on advertising without knowing what the return on investment is. Most Web
analytics packages have very effective tools for tracking all campaigns and referrals to
determine which work and which do not. The goal for each campaign should be that it
generates more revenue than it costs – the campaigns that do not meet this criteria should
be abandoned, those that do create additional value for the business are candidates for
additional marketing effort. Without adequate measurements and tracking this is
impossible.
Performance Dashboards
One of the causes for ineffective use of Web analytics is the fact that many
managers do not review reports on a regular basis. The benefit of a dashboard is that it is
possible to assess the performance of the online business by having all KPI reports in a
single location. Most Web analytics packages allow a user to graphically view the sites
critical visitor, sales, conversion, and other key performance metrics on a single page. It
may also be able to include other information such as monthly targets and projections.
The dashboard can be quickly accessed and may also be emailed and shared with other
individuals interested in the performance of the online site.

Conclusions
Web analytics has progressed from the realm of simplistic metrics – visits, page
views, orders – to critical component of the business processes of successful web sites.
Businesses gradually came to terms with the fact that the web is a legitimate business
channel and began to expect it to deliver a return on investment like any other business
channel. Simultaneously, the Web analytics market has continued to improve in their
ability to help those responsible for a web site measure the site’s effectiveness at meeting
the goals of the business. There is also a wealth of information on how to effectively use
these tools as part of a process of continuous improvement. Web sites that are able to
implement Web channel performance management strategies and processes that tie
measurements to business objectives will continue to thrive whereas those that do not
will struggle in determining if the web site is worth the investment.

References
Anderson, Eric (2003) Optimizing Your Marketing Campaigns with A/B Testing
http://whitepapers.businessweek.com/detail/RES/1061923472_193.html&src=TRM_TOPN
Ballardvale Research (2004) Market View – Emetrics Conference: Santa Barbara, 2004
http://www.ballardvale.com/pdf/MV%20-%20Emetrics%202004.pdf
Ballardvale Research (2004a) Market Trends – Web analytics: History and Future
http://www.ballardvale.com/free/WAHistory.htm
Becker, Jonathan (2005) Why Metrics-Centric Performance Management Solutions Fall
Short
http://www.dmreview.com/article_sub.cfm?articleId=1021509
Burby, Jason (2004) Three Reasons Why Analytics Fail Companies
http://www.clickz.com/experts/crm/analyze_data/article.php/3307121
DoubleClick (2004) E-Commerce Site Trend Report for Q3 2004
http://emea.doubleclick.com/WEB_ADMIN/documents/dc_q304ecommerce_04111.pdf
Eisenberg, Bryan (2004) Web analytics for Retailers
http://www.clickz.com/experts/crm/traffic/article.php/3396361
GrokDotCom (2005) Help Yourself to a KPI!
http://www.grokdotcom.com/topics/helpyourselftoakpi.htm
Harris Interactive, et al (2005) eSpending Report: 2004
http://www.harrisinteractive.com/news/allnewsbydate.asp?NewsID=878
Keller, David and Josh James (2003) The Power of Where in Web Analytics
http://www.dmreview.com/editorial/newsletter_article.cfm?nl=dmdirect&articleId=7382
Lenzen, Roman (2004) Customer Analytics: It’s About Behavior
http://www.crm2day.com/library/EpAVkVpFyZoxUYKRLg.php
Maguire, James (2004) Web analytics, Who’s Mining the Store
http://www.ecommerce-guide.com/solutions/design/article.php/3440351
McGann, Rob (2005) Study: Consumers Delete Cookies at Surprising Rate
http://www.clickz.com/news/article.php/3489636
Millard Brown (2004) Marketing & Media Snapshot: 2004
Obrey, Thomas (2004) You Now Have Web Site Traffic, Now What?
http://www.tmcnet.com/tmcnet/articles/2004/033104pm.htm
Peterson, Eric (2004) Web Analytics Demystified
http://www.webanalyticsdemystified.com/about_wad.asp

Sapir, Daniel (2004) Online analytics and Business Performance Management
http://www.dmreview.com/editorial/dmreview/print_action.cfm?articleId=1008820
Shop.org (2004) The State of Retail Online 7.0
http://www.shop.org/research/SRO7/SRO7main.asp
Sterne, Jim (2002) Web Metrics – Proven Methods for Measuring Web Site Success
http://www.amazon.com/exec/obidos/tg/detail/-/0471220728/102-6649907-4508967?v=glance
Sterne, Jim (2005a) E-Metrics.org: 2005 Web analytics Conference agenda
http://www.emetrics.org
Sterne, Jim (2004a) Web Channel Performance Management: Aligning web site vision
and strategy with goals and tactics
http://www.dmreview.com/whitepaper/WID1010070.pdf
Sterne, Jim (2004b) 10 Steps to Measuring Web Site Success
http://www.marketingprofs.com/preview.asp?file=/4/sterne13.asp
Websidestory (2004) Web analytics – It’s Surprisingly Simple
http://www.knowledgestorm.com/sol_summary_77133.asp
WebSideStory (2004a) Use of Key Performance Indicators in Web analytics
http://www.websidestory.com/web-analytics-resources/best-practice-briefs/view-document.
html?docid=031

Optimizing the Online Business Channel With Web Analytics

More Related Content

What's hot (20)

Similar to Optimizing the Online Business Channel With Web Analytics (20)

Recently uploaded (20)

Optimizing the Online Business Channel With Web Analytics