Tải bản đầy đủ - 0trang
Chapter 9. Tracking Results and Measuring Success
These are all valid objectives that relate directly to the business. Missing from this list are things
such as measuring rankings, measuring the total number of links, and measuring the average
cost per acquired link (organically obtained links still have a cost: the cost of the marketing
campaign to get them). These other measurements may be interesting, but they are not the
Closely connected with this are direct drivers of results. These include parameters such as:
• Total site traffic
• Traffic to your most popular pages
• Keywords driving the traffic
• Keywords driving conversions
The mechanics of measuring results in terms of the main business goal should always be the
first set of measurements you put in place. From there you can diverge and look at other metrics
that might help you diagnose problems with the site or give you additional insight into how
to set up a campaign.
Why Measuring Success Is Essential to the SEO Process
Although quantifying deliverables and measuring progress are important for external reporting
purposes, it is just as important for SEO practitioners to measure the efficacy of their own efforts
to make timely adjustments as necessary. As you will see in this chapter, numerous tools are
available to aid in this process.
At the beginning of any SEO project it is wise to establish baseline data points for the website.
This includes the following:
• Quantifying organic search traffic by search engine and keyword
• Quantifying a baseline of the major keywords that are driving traffic by search engine
• Quantifying a breakout of what sections are getting the current organic search traffic by
search engine and keyword
• Quantifying data on conversions broken down by search engine and keyword
• Identifying poorly performing pages
• Tracking search engine crawler activity on the site
• Determining the number of indexed pages
• Identifying 404 error pages and external sites linking to these pages, if any
Remember: you cannot methodically improve what you cannot measure.
Defining and mapping the path toward concrete goals are crucial aspects of the SEO process—
and over time, some goals may change. Therefore, it is also important to make sure the data
you capture helps you understand your progress against these goals.
In the world of web analytics, this is referred to as picking Actionable Key Performance
Indicators (or Actionable KPIs). The best data measurements are those that potentially result
in an action being taken in response. Think of this as data-driven decision making.
The Tracking Cycle: Produce, Launch, Measure, Refine
In summary, the basic process usually looks something like this:
1. Define an SEO campaign and set goals.
What are you going to accomplish, and what is the strategy for accomplishing it? How will
you measure progress?
2. Discuss your strategy.
The marketing and business development teams are your allies here—you want to ensure
that your SEO objectives are based on the overall business and site objectives, both longand short-term.
3. Establish a baseline.
Now that you are about to start and you have decided how you are going to measure
progress, establish a baseline by recording the current stats prior to beginning work. Make
sure you don’t get a false baseline due to seasonal factors or some other unusual event.
4. Proceed with your project.
Implement the new pages, the site changes, the link-building campaign, or whatever you
have planned. Put it in place and execute.
5. Collect data.
Collect the newest data for each metric you decided to focus on. Since this is SEO and SEO
can take many months to show results, make sure you wait long enough for your efforts
to have an impact. Of course, if you are a student of the process, you can take more
frequent measurements so that you can see how things begin to progress over time. For
many on-page changes, 60 to 90 days are enough, but for link-building campaigns it may
take six months or more to see the full impact. Many factors could influence the length
of time you should wait. Here are some of them:
• If your site is brand new, it may take longer for your changes to take effect.
• If the scope of the change is drastic (such as a complete redesign), the time to see results
will probably be longer.
• Sites that get crawled at great depth and frequency will probably yield results faster.
• Sites seen as authoritative may also show faster results.
6. Compare the baseline data to the new data.
The new data has little meaning unless it is compared to your baseline. This is the time
when you can really assess your progress.
TRACKING RESULTS AND MEASURING SUCCESS
7. Refine your campaign.
Now that you have compared your old data with your new data, you can make some
decisions. Is the campaign a bust? If so, abandon it and move on to the next one. The old
business axiom “Fail quickly” applies here. The faster you diagnose a failure and move on
to the next thing, the better.
You may also find you are getting mediocre results. Examining the data more closely may give
you some ideas as to how you can improve those results. And if you are achieving great results,
look for ways to scale the effort and drive even more volume.
Using Analytics As a Business Case for SEO
You can use a properly structured plan as the business case for an SEO project. The way to do
this is to express the target results of an SEO project in terms of financial impact. You could
include a variety of metrics in a business case, such as:
• Lead generation
• Branding value
• Other action triggers (newsletter sign-ups, contact requests, demo requests, free-trial
acceptance, viewing a specific piece of content, etc.)
Measuring such things requires that you tie organic search engine visits to the revenue and
other conversions that result.
Measuring Search Traffic
Classic web analytics data is an incredible asset to SEO. Here are three examples of ways to
utilize this data for SEO purposes:
• Look at your daily referral reports to detect newly received inbound links (the great
majority of webmasters click on a link after implementing it, to make sure it works)
• Look at the search terms people use to come to your site to spot long tail search
• Measure the results of your campaigns by tracking the increase in conversions you are
driving over time
Web analytics are a must-have for any web publisher.
Your hosting company most likely provides a free web analytics solution, such as AWStats,
Webalizer, or something similar. Although these tools provide valuable data, they are very
limited in scope, and other tools out there provide significantly more data. Here are six of the
• Google Analytics
• Core Metrics
• Unica Affinium NetInsight
• Yahoo! Web Analytics
Web analytics platforms track your site’s traffic in two major ways. The older of the two
methodologies is to use software that analyzes your web server logfiles after traffic has
occurred. Setting up this type of solution generally involves installing the software on an
internal server that is able to gain access to the logfiles.
website. Provided that you have a basic template for your pages, this generally is a fairly
and builds a data profile reflecting that activity.
Selecting the Right Analytics Package
strengths and weaknesses. The biggest advantage of the logfile method is that you can track
The second big advantage of a logfile-based solution is that you run the software in-house, so
no third party has a copy of a logfile with your proprietary traffic data on it. This distinction
can be a big security issue for some organizations.
tracking, or gather pages into logical groupings in a manner that cannot be done in logfilebased applications.
Some companies, such as Unica and Webtrends, will offer you both options, or they will even
offer the option to implement a combined solution. This kind of approach can bring you the
TRACKING RESULTS AND MEASURING SUCCESS
Making this decision is only the first step in picking an analytics package. We listed six of the
more popular vendors earlier, and there are many more vendors than that. Each of these
packages has different strengths and weaknesses. Not only that, they all do their counting a
little bit differently. The chart from Stone Temple Consulting’s Web Analytics Shootout
report, shown in Figure 9-1, helps to illustrate the point.
FIGURE 9-1. Web analytics accuracy
The chart in Figure 9-1 shows the results of seven different analytics packages (listed on the
right). The traffic data for each vendor is reported across five different sites (represented by the
acronyms on the bottom).
On the AMD site, the lowest reporting package (HBX Analytics) shows a little less than 750,000
unique visitors, and the highest reporting package (Clicktracks) shows about 1,050,000 unique
visitors during the same period—almost 50% more!
These differences result from different decisions in how the analytics packages conduct visitor
tracking. None of them are right or wrong, they are just different in the exact thing they are
HBX Analytics, which is listed in Figure 9-1, is no longer available, as WebSideStory
was acquired by Omniture. In addition, IndexTools was acquired by Yahoo! and is
now called Yahoo! Web Analytics.
The more important component of this is whether the functionality of the web analytics
software fits your needs. Making this more difficult to understand is the fact that you often do
not know what your requirements are until you have used analytics for a while. As you engage
with analytics, you will continually learn more things you want to investigate, and develop
For many companies, one of the best solutions is to start with a free analytics package such as
Google Analytics or Yahoo! Web Analytics and then look to buy a higher-end solution once
they have pushed these packages to their limits. By pushing the limits of these free analytics
products first, you will end up developing a set of requirements you can use in deciding where
to go next.
None of this is meant to say that you should not brainstorm your requirements in detail before
selecting an analytics package. You should (and must). Just expect that you will develop new
requirements along the way. Web analytics is a journey that unfolds over time.
Based on the requirements you establish in your upfront brainstorming, you may find that
you require a set of features that the free packages do not provide. Use that as knowledge to
select the right package to start with.
Valuable SEO Data in Web Analytics
You can extract all kinds of data from web analytics. Here are a few of the more interesting
types of information you may want to extract.
Traffic by search engine
One of the first things you may want to know is the breakout of traffic by search engine.
Figure 9-2 provides an example of such a report in Google Analytics.
Notice how small the traffic is on Yahoo! compared to Google (about 2.5%). This may be
indicative of a problem with the site in question and Yahoo!. This site owner might want to
spend some time exploring why the site traffic from Yahoo! is so low.
Traffic by keyword
One of the basic data points of interest for an SEO practitioner is what search terms are bringing
traffic to the website. This provides a quick way to see where the SEO campaign is going well
and where it is not going so well. You can also use this to spot opportunities where a key search
term is providing some traffic, but not as much as you would expect if you were ranking highly
for that term.
You can then look to see where you are ranking for that term. Perhaps you are in a lower
position on the first page, or on the second page of the SERPs. If so, it might make sense to
focus some attention on this term. With a little effort, such as a focused link-building campaign
for the page in question, you may be able to move up several positions and obtain a traffic boost.
TRACKING RESULTS AND MEASURING SUCCESS
FIGURE 9-2. Traffic by search engine
A traffic-by-keyword report can also show you the long tail of search as it relates to your current
site. Figure 9-3 depicts a snippet from the organic search phrases report of Yahoo! Web
Analytics showing some of the small-volume terms for Stone Temple Consulting’s website.
Notice that a lot of names are showing up in the list, in addition to the rather interesting who
is the author of ask.com. Looking into this query data can give you a broad perspective on
opportunities for long tail search.
Notice also how two of the names combine Google with a person’s name (jack ancone google and
google carter). This could indicate an opportunity to make sure the person’s organization name
is a part of the title of the articles in the Stone Temple Consulting interview series.
FIGURE 9-3. Long tail keywords
Segmenting Search Traffic with Multiple Parameters
Next, you can consider putting these things together. Even the free tools provide substantial
capability for building out custom reports. Figure 9-4 depicts a screen shot from Yahoo! Web
Analytics that shows the search phrases just for Google, and the pages of the site that Google
sent users to when users clicked on your link.
This is a key improvement to your SEO research for two reasons:
• If you are looking into what terms can bring fast traffic benefits as a result of some
additional optimization, you are going to want to know in which search engine you are
• If you are going to optimize a page to rank higher, you will need to make sure you are
optimizing the right page!
It is interesting to look at a referring site report for a number of reasons, but one of the more
interesting SEO reasons to do so is to spot when you receive new links. You can often see those
new links in these reports first, even before the search engines report them. Figure 9-5 shows
a sample portion of the referring sites report from Google Analytics.
TRACKING RESULTS AND MEASURING SUCCESS
FIGURE 9-4. Search phrases from one search engine
FIGURE 9-5. Referring sites report
Expanding on this example, consider the site circled in Figure 9-5, edwardbeckett.com. If this is
the first time you have ever noticed this site in your referrers, it can be a leading indicator that
you have received a new link. This is of interest as it can help you to detect new links that
result from your link-building campaigns, and therefore help you measure which of your linkbuilding campaigns are yielding the best results.
Using Analytics Dashboards
In analytics terms, a dashboard is a single-page view that contains your most critical metrics all
in one place. Of course, your most critical metrics are different from those of the next publisher,
because the needs of different sites vary greatly.
In addition, multiple dashboards may be required in any given organization. For example, the
CEO of a large public company probably wants to see different data (and a lot less of it) than
a senior business analyst.
Each analytics package provides methods for implementing a custom dashboard. Figure 9-6 is
an example of one from Unica’s Affinium NetInsight.
FIGURE 9-6. Custom dashboard
As you can see from Figure 9-6, a dashboard can be quite visual. What is most important,
though, is that it provides the data that is most important to the person for whom the dashboard
TRACKING RESULTS AND MEASURING SUCCESS
was designed. As an SEO practitioner, you can implement a dashboard to show progress against
the goals you set for your SEO campaign.
Providing this type of visibility has two important benefits:
• The person viewing the report will appreciate the fact that she does not have to work hard
to do a quick health check on the progress of the SEO efforts. As we suggested at the
beginning of this chapter, the fact that you agree to measurable goals will be a great
comfort to management.
• You will know what data your managers are looking at. When something goes wrong (or
right), or when management wants to discuss some aspect of the business, they will have
started from the dashboard you set up for them.
A Deeper Look at Action Tracking
Action tracking is one step deeper than basic analytics. Rather than simply observing what
pages are visited and how many unique sessions are logged, action tracking allows you to
narrow down groups of visitors based on the actions they take on your site.
In most instances, it requires setting up a code in your analytics program and attaching that
Once you’ve plugged it into your analytics and the website, you can use the action to refine
data you’re already collecting. Figure 9-7 provides a look at how this works.
You can see from Figure 9-7 that:
• SEOmoz’s sign-up form has action tracking applied to it.
• Based on the people who sign up, you can predict which search terms will be better at
converting visitors into applicants.
• The Revenue column is empty, but if you were tracking e-commerce buyers, you could
put their totals into the Revenue column and track high-volume buyers.
• Expanding on this idea, you could also track users by time of day, the search engine they
used, their geographic location, and so on.
So, what types of actions should you be tracking on your site? The answer varies depending
on your business and site structure. Here are some suggestions as segmented by site type.
Add to Cart button
Studies have shown us that users who “add to cart,” even if they do not complete the
checkout process, are more likely to return to make a purchase. This is also a good way to
calculate shopping cart abandonment and make changes to refine and improve the