Thursday, June 24, 2010

How to effectively use Web Analytics

A Web analytics program is like an airplane pilot's navigation system. As pilots look to their dials and scopes to steer the plane, site owners monitor the data from their Web analytics software to help them run their site.

An analytics program — like those by WebTrends, ClickTracks, Omniture, Google and others — outputs ever-changing data about a site's traffic patterns, unique visitor numbers, bounce rates, referring URLs, and other user trends.

But, like a pilot's dashboard, no one ever said using a Web analytics program is easy. These programs display torrents of data, row upon row of numbers. Even analytics experts admit that they're complex and can be confusing.

Big, well-financed Web sites have no problem with this complexity. Top e-tailers employ full time analytics experts (sometimes full teams of them), specially trained to interpret analytics data. The big boys know that constantly tracking user trends is critical to the bottom line.

But smaller and mid-size sites have struggled to catch up. Smaller merchants are busy handling everything from order fulfillment to accounting. After a busy day, digging into the inscrutable data from an analytics program is a daunting task.

To shed light on how to best use an analytics program, E-commerce Guide spoke with Avinash Kaushik, a top expert who's respected throughout the analytics community. Kaushik talked about getting started with analytics, key reports to run and essential concepts for understanding your site's data.

Getting Started

  • You don't have to buy a top package (there are good ones available for free)
For merchants new to analytics, Kaushik says, "if you haven't yet bought a top program, please don't — you have other options.
"There's such a steep learning curve [to the top packages] that you're better off playing with a free tool and figuring out things about your site before spending your money."
He recommends that owners get the free package from ClickTracks or Google
These free packages are extraordinarily easy to set up. For ClickTracks, you download the software and point it toward your log files; for Google's program, site owners insert Javascript tags into their pages' footer files. Ten minutes later, the data is flowing.

Kaushik is fond of both free programs, though he leans toward ClickTracks because it's "an extremely simple tool. It only does six reports, and it's a fabulous way for a small business owner to feel things out."

Lets Start with the Rules of Engagement

Start with Your Desired Outcome — Not with Reports

Let your goals for your site be your guide to understanding the data. So rather than opening your program and getting distracted by the 'geo' report or the 'trends of unique visitors,' instead focus on factors that are actually producing the outcome you want. Look for reports and analysis that will help you understand the whys and whats around your desired outcome.
"Why am I selling X number of figurines, and not Y [a higher number] number of figurines? What pages are the people who buy things reading? If I'm running promotions, which Web sites are sending me traffic with people who make purchases? "If you start with the outcome, you will automatically ignore the distractions and focus on the data that will provide you with meaningful insights."

Six Key Reports — And What to Do with the Data
The analytics industry is full of horror stories of eager site owners who churn out dozens (in some cases, hundreds) of reports per week. Before you fall prey to report-mania, realize that some reports are more important than others. In particular:
•Key Report #1: Referring Search Terms From Search Engines
The report that tells you which search phrases people are using to find your site tells you a lot about your users. For most of the people who come to your site, it's very hard to understand their intent. Yet there's no more effective way to discern this intent than knowing these top search terms — it's like mind reading. They reveal specifically what people want from your site.
Check your unpaid search engine ranking for these terms. If you're not ranked highly, do more search engine optimization — make yourself number one. Also important: develop more content and/or products that relate to these terms. Catering to your users creates more traffic.
•Key Report #2: Referring URLs
Look at the report that tells you which Web sites are sending you traffic. Does this correspond to your expectations?
Use this list of your referring URLs to boost your user base. If you see that there are sites that should be sending you traffic but aren't, contact them — can you form an alliance with them?
And if you find Web sites that are sending you traffic, but not as much as you want, do something for them. For example, if you think your mother-in-law's site should be sending you traffic and it's not showing up, why? 
Included in the data about referring URLs is a still more important data set: referring URLs for all people who reach a certain goal, for example, people who make a purchase. (Breaking data down between various user types and actions is called "segmentation" in the language of analytics programs.)
Wouldn't anyone love to know which Web sites are sending traffic to your site who meet this goal? Once you know this, you'll understand a lot more about the traffic and what's going on in your site, rather than looking at generic trends of visitors.
•Key Report #3: Content Popularity
It's essential to view the list of Top 10 (or 15) most popular pages on your site. Knowing what content is being consumed can lead you to so many insights. "What are people coming to my Web site for? Are the things that I want to promote actually the things that people are looking at?"
This list offers plenty of surprises. Sometimes it validates that visitors are doing the 'right' things, but more often, it shows that visitors are doing all these other things on your site that you didn't anticipate — and that's great food for thought
Turn these 'surprises' to your advantage. "You get ideas for what you should promote on your Web site." And, if appropriate, make sure you run your most compelling promotions on your top 10 most popular pages.
•Key Report #4: Percent of Visitors Who Visit the Home Page
This metric often shocks site owners. They think that everyone sees the homepage, so they put their maximum energies and promotion there. But since search engines display a site's internal pages, most users enter a site far from the home page.
People have benchmarked across so many e-commerce sites, and on the extreme end, only 10 percent of people see the home page, on the other end, 50-60 percent see the homepage.
Knowing home page usage levels enables you to better calibrate the resources dedicated to your homepage. In many cases, sites spend far too much effort on their home page and far too little on popular internal pages.
•Key Report #5: Site Overlay
Wouldn't you love it if you could open your site and see exactly where people are clicking? With the Site Overlay report you can. It displays your actual pages — just as they look to users — with a click level indicator next to each link. It shows the number of people who click on each link.
Kaushik calls the Site Overlay report "one of my favorites. I recommend doing Site Overlay reports for at least your top 10 pages." Site Overlay might reveal, for example, that the gorgeous hyperlinked graphic you spent four hours building gets no clicks at all. Indeed, this report might hurt your ego.
Site owners, by segmenting their analytics data between "visitors" and "purchasers," can learn which links are most popular with people who ultimately make purchases. These links can then be positioned for still greater visibility.
To improve your click rates on key links, "do simple experiments" with site structure and navigation. Also, look at well-financed sites to see how they lay out their links. "Get inspiration from the big boys, they've invested millions in studying user trends."
•Key Report #6: Site Bounce Rate
The Bounce Rate report reveals the number of visitors who stayed just a few seconds. These are the people who came to your site but didn't engage. In short, your bounce rate is your failure rate.
When you calculate your site bounce rate, you'll be surprised by how many people leave your site in five seconds. The exact number of seconds that represents a failure varies site by site (and even page by page within a site).
Put another way, sites should be aware of a critical number: what is the shortest visit time (in seconds) for a site user to be considered "interested"? To find this out for your site, ask an average person to run through a typical scenario on your site, and time it with a stopwatch.
"Get your grandmother to do it, then get your 16-year-old to do it". "And then you'll have your benchmark — somewhere in between those two is the optimal time."
Site owners can dig deeper into their Site Bounce Rate report to find out which referring URL or pay-per-click keywords draws traffic with high bounce rates. "That traffic is the wrong traffic," he notes. This data obviously suggests changes in your PPC buy.
In the case of referring URLs with high bounce rates, take a look at that referring site. Are you being linked to in a way that doesn't accurately reflect your site? Why are visitors from this referring URL so disappointed when they land on your page?
Your Program's Blind Spot
Your analytics program tells you which search terms are sending traffic your way, but it can't tell you which popular terms you should be using but aren't.
For example, if you sell Tom Cruise ceramic figurines — but no Tom Cruise glass figurines — you're not aware of all the people searching for 'Tom Cruise glass figurines.' If you get no traffic for a term, your analytics program will give you no clue to this horde of users.
However, there are two ways to discover these "hidden" relevant keywords to draw this additional traffic:
1) Subscribe to a service like Hitwise
Hitwise lets you scan lists of popular keywords. For instance, it'll show you the top 1,000 terms in a given industry, or the list of terms from any search engine in a certain category. Hitwise also reveals which sites people go to for those terms.
But Hitwise is a pricey tool. A free tool is Overture's Keyword Selector Tool, though it won't show which sites people flock to from given terms.
2) Simply type in a large number of relevant terms to Google
As a business owner, you probably know (or can guess) the most frequently used search terms that might draw you traffic. It costs nothing to type them into Google's search box and find all the sites that come up.
"Look at the sponsored links that show up". These businesses (who are likely your closest competitors) have probably figured out all the best contextual key words. He recommends visiting these sites and doing a View Source look at their home page's code. Within the HTML code, look for the list of keywords embedded in their meta tags.


SEO vs. PPC: Using Analytics to Save Money
The science of Web analytics is married to the practice of buying pay-per-click advertising. The reason merchants huddle for hours with their analytics package is to find out if their PPC expenditure is actually driving them traffic.

Indeed, the frightful cost of PPC makes analytics all the more important. For those merchants considering PPC, analysts caution that they first must have full knowledge of their analytics program. "If you want to do PPC, first learn which key phrases work for your goal," otherwise you're wasting your money.
"PPC is a very fast way to lose a lot of money if you're not careful with your analysis. Very few people do it right."

In fact, sponsored links, though they drive traffic, are only one part of an effective strategy. "Increasingly, people who search are beginning to realize that these are sponsored links — businesses are paying to get in here. And organic placement [showing up in the free results] is providing more credibility." For this reason, resources spent on search engine optimization — which produces unpaid search rankings — tends to be a better long term investment than pay-per-click.

No comments: