As we noted in Part 1, Web analytics is a subject that even the most experienced online retailers find confusing.
And for good reason. Web analytics software outputs a blizzard of data about your site: most visited pages, top referring URLs, popular search keyword terms and many other metrics.
These reams of data are tough to digest if you're not a statistician. Many merchants wonder: What are all these numbers, and what do they mean?
To help crack the code, we'll take a guided tour of an actual analytics program, guided by the man who helped design it. The program is ClickTracks and our guide is John Marshall, the company's CEO. Also helping will be Michael Stebbins, the company's vice president of marketing.
Additionally, we'll talk to a leading analytics expert about best practices in monitoring your site traffic.
First, we'll open up eight key reports in ClickTracks and examine the visitor patterns of a fictional site. (Editor's note: although we're looking at ClickTracks, equivalent reports are found in all analytics programs.)
1) Navigation report
"Customers coming to a Web site basically vote for what they want by clicking on things," Marshall says. "The job of the analytics program is to express those votes." The navigation report, which displays a page as visitors see it, expresses these votes by recording two key facts:
- The percentage of visitors who click on each link
- Where those visitors come from (MSN, Google or other)
This not only allows merchants to gauge the relative popularity of each link, but also to monitor return on investment on their pay-per-click budget by comparing traffic levels from various engines.
This report "encourages business owners to put themselves in the visitors' shoes and understand the site from their view," Marshall says.
|The navigation report reveals the popularity of each link.|
(Click for larger image.)
Marshall says he encourages merchants to use this data by following the user paths it records. Follow the most popular links though the site, go to the shopping cart and find out, "What is the behavior how many people abandoned it?"
2) Segmentation Report
The segmentation report includes a "label wizard" tool, which allows merchants to track the behavior of specific groups of visitors. For example, it reveals the navigational path of customers who made a purchase, those who came from Google, and those who left the site quickly.
"People who come in from paid search behave differently than people coming in through organic [unpaid] search," Marshall says. "Paid search is used by people who are in a hurry and know what they want. Organic is used more as an authoritative voice."
Using the label wizard, a merchant can compare the behavior of users who visit via paid and organic search. What percentage of each goes on to make a purchase on your site? The "magic" of segmentation is that it shows not just your top referring URLs, but the referrers who disproportionately get people to the checkout page. This enables you to adjust your ad budget (or your relationship with your top referring URLs) to maximize sales.
|The label wizard enables site owners to track the page-by-page site usage of specific groups.|
3) Keyword Analysis Report
The keyword report shows the search terms that brought visitors to the site, where they came from, and how long they stayed. The amount of time that people stay on your site, per keyword, "is a very good leading indicator of positive ROI," Marshall says.
This report can also be used to do what's called "day-parting," which tracks usage by time of day. A merchant, for example, might use day-parting to compare her traffic between 9 a.m. to 5 p.m. with traffic outside of regular business hours.
|The keyword report records traffic levels from each search engine. It shows, for example, that the search phrase 'fresh fruits' brought users from Google who stayed an average of 22 seconds. |
(Click for larger image.)
"If I know the robot visits every four days, I may want to push my changes just before they come back, to get the best ranking," Stebbins says. The Robot report also reveals problems with your URLs. If a link has such a highly complex URL that it confuses the spiders meaning it never gets indexed this will show up in your Robot report.
A merchant might choose to boost the number of anchor keywords on certain pages to encourage the spiders to crawl it, then check the Robot report to see if this works.
|The Robot Report shows how frequently the search engines crawl your pages. |
(Click for larger image.)
5) Funnel Report
True to its name, the Funnel Report shows the movement of visitors through the site, displaying the ways these groups narrow as they move from page to page. For each page, the report displays what percentage of users who moved on to a given subsequent page.
The report displays "the level of persuasion or influence of each page," Stebbins says. Those pages that are more persuasive send a higher percentage of users on toward the final goal of checkout.
The Funnel report reveals the quirks of user behavior. It shows that "people bounce all over your site, they perform like an unpredictable tornado throughout the site," Stebbins says.
This report can be segmented to track the way users funnel from various sources: from your newsletter, pay-per-click, organic listing and other sources. Marshall notes that in his own monitoring of the ClickTracks site, he notices big differences between which pages are most "persuasive," based on whether their traffic comes from the ClickTracks PPC buy or its newsletter.
|The Funnel report lists pages that link to one another and shows the percentage of users who clicked to the next page. |
(Click for larger image.)
6) What's Changed Report
"I come to this report to find out what I didn't predict," Stebbins says. The What's Changed report "is probably the exception to the 'I have a decision in mind, I'm opening up ClickTracks,'" approach, Stebbins says.
This report's data tends to provide all manner of surprises, including what pages and referrers are on the way up or are headed toward oblivion.
The What's Changed report displays data that might not show up elsewhere. A referring URL that sends you minimal traffic but that grew disproportionately wouldn't show up in your list of top referring URLs, but would rank highly in the What's Changed data.
Before ClickTracks started its Click Fraud report, merchants used What's Changed to see if any PPC ads zoomed dramatically with no corresponding move in conversion rate.
|The What's Changed report often displays unexpected shifts in user trends. |
(Click for larger image.)
7) Campaign Report
Merchants who want to monitor their return on investment of their pay-per-click campaign will eagerly scan their Campaign Report. It pulls in keyword pricing data from search engines to build the report, allowing you to bid more judiciously.
The Campaign Report also shows the percentage of visitors who spent less than five seconds on the site, meaning they saw nothing of interest.
This is a critical metric. In measuring the effectiveness of keyword buys, Stebbins says he finds the time-on-site metric to be more accurate than the ROI metric. "It's much less prone to error and gaps in time in terms of sales cycles."
This report also provides a metric called ROAS, or return-on-advertising-spend. ROAS is a ratio between the cost of a campaign and the revenue it generates. "If you end up with a number greater than one, you're making money, if it's less than one you're losing money," Marshall says.
|The Campaign Report answers a critical question: how cost effective is your pay-per-click expenditure?|
8) Site Overview Report
For the busy merchant with limited time, the Site Overview would likely be the most important report. It allows a quick overview of the behavior patterns of various groups throughout the site, such as those who purchased, or those who came from Yahoo.
It shows, for example, the keywords used by visitors who actually made it to the shopping cart. "If those keywords are dramatically different from my pay-per-click keywords, than I'm doing something wrong," Stebbins says.
Most important, the Overview calculates the cost of visitors (based on PPC buys) and the revenue from customers.
|The Site Overview provides a global picture of your site's user trends.|
(Click for larger image.)
Web Analytics Best Practices
Eric Peterson, the author of "Web Analytics Demystified," has worked in the analytics field since its infancy. He spoke about the essentials of using an analytics program to monitor online sales objectives.
Define Objectives, then Find Indicators
Define your business objectives of your site, then look for KPI (key performance indicators) in your analytics package that relate to this goal.
Is your objective getting people to the product page? Getting them to download a PDF? Finding the KPIs that support these objectives is "often an 'ah-ha' moment for merchants," Peterson says. Two particularly important KPIs are average order value and conversion rate.
Avoid 'Analysis Paralysis'
Peterson finds "analysis paralysis" to be an all-too-common phenomenon throughout online retailing: businesses get bogged down in too much analytics data. They try to digest too much of it.
In his experience, the companies who have overcome this overload have found six or so (perhaps more) critical KPIs to track, and they limit their focus to this handful.
Never Be Satisfied
When you open your analytics program and view your numbers, "Organizations need to not be satisfied with the numbers they're getting. They need to look at their conversion rate and say, 'three percent just isn't very good.'"
Create A/B Tests to Monitor Change
With your goals for improvement in mind, "start changing things on your Web site," Peterson says. "Get visitors to go through a new process, a new checkout, use different buttons." Then use your analytics program to measure the effect of these changes.
Be Aware of Your 'Stickiness Rates'
A site is "sticky" when its users want to stay on it (or "stick to it") instead of leaving quickly. Peterson defines a site's stickiness rate as the percentage of users who see more than one page.
"If your home page is 75 percent sticky, that means that three quarters of the people stay and view at least one more page and one quarter don't." Your analytics program will tell you your site's stickiness rate. This is especially important for traffic that comes from expensive pay-per-click campaigns.
Make Sure Your Analytics Work is Assigned
Is there someone in your business who's assigned to monitor analytics data on a set schedule or does monitoring happen on an ad hoc "we'll do it when there's time" basis?
"The companies that are successful with analytics attribute their success to having dedicated staff," he says. That is, they track user trends on a daily basis.
It's All About Comparison
Getting the most from your analytics data means being aware that today's numbers don't mean anything by themselves; they must be compared to last month's and even last year's. "The most important context is week over week," Peterson says.
Essential Indicators: Pageviews Per Visit and Visits per Visitor
As important as it is to draw more total visitors, it's also essential to drive the number of page views per visitor. An engaged shopper one likely to make a purchase looks at plenty of pages.
Also, individuals who return often to your site are more likely to buy, so you want a high visits (or 'sessions') per visitor figure. "Are people coming back enough to reasonably make the purchase? If you're getting one visit per visitor, no wonder your sales are low."
Don't Worry About Other People's Numbers
If your goal is growing your business, it doesn't matter how many page views Best Buy gets, or the total unique visitors of Wal-Mart.com. The point is to compete with yourself.
The numbers you're trying to beat are the first numbers you got when you started measuring. Or last month's number, or last week's.
"I often scratch my head at the studies that say industry-wide conversion rates are three percent," Peterson says. "So the sites that convert at four percent feel great, while the sites who convert at two percent get despondent. The problem is: if you're converting at four percent, why aren't you converting at five percent? the only benchmark you can work against is your own."
|Do you have a comment or question about this article or other small business topics in general? Speak out in the SmallBusinessComputing.com Forums. Join the discussion today!|