Access Open Source Technology
Innovate w/ IBM and Discover New Open Source Technology Today. Learn More.
Click here to know more
Websites are only effective if they’re attracting the right visitors. Web analytics helps you know them better
Customers are the lifeblood of any business. For businesses that depend on Web visibility for their growth, customers mean visitors on their Web sites—those who come once should be interested, and visit more often. Whether you’re a business Web site, blog, forum, hobbyist site or news site—any site that earns money, in fact—you should care about visitors. And before caring about them, you must know about them, where they come from, and what they want. This, and a variety of other answers, can come from the wonderful art of Web analytics.
As you might have already guessed, Web analytics means number-crunching. It means using tools to collect data on the server side, and then analysing that data to understand where you, as an investor, are getting your expected ROI (Return On Investment). At the more basic level, the data will let you know whether you have succeeded in reaching out to your intended audience.
Consider this: you’re running a business selling diapers, and it’s not a brick-and-mortar business in that you have no outlets—just the Web site. Being a new business, you don’t have a widespread distribution network and for now aim to cater to local needs. As a businessperson, you are necessarily interested in how many people are visiting your online shop. You run a Web analysis where you find that there is only a minority of your visitors come from your city, or neighbouring areas. And what’s worse, there are very few who click on the cart and make a purchase, which in Web analytics jargon is called goal conversion. Your business is obviously not becoming profitable, and you need to do something about it. With Web analytics, you get to know about this in advance and hopefully do something before the bills start arriving and the creditors start queuing outside your doors. Without Web Analytics, you would be in for a rude shock.
(Don’t) Bluff Your Way Through
And now, time for the customary round of jargon-busting. Web analytics being a subject which uses maths, most of these terms describe percentages and numbers which most tools use:
Page views: This is what people mean when they are talking about hits. A page view is the successful loading of a document on a server, and doesn’t include visits by search robots or error messages, requests for sound and video files, or even CSS and images. Page views don’t denote the number of people visiting the site, because a single visitor can visit multiple pages and request a number of documents.
How well you know what your users are looking for, and how quickly you can act on it—is one of the most important parameters that contributes to any online portal’s success”
Suvomoy Sarkar, Head of Analytics, Info-edge
User Sessions: This is the metric used for determining the number of visitors that actually spend time on your site—it is usually recorded for 30 minutes of users’ activity. It accounts for both IP addresses, and the number of users logged in, if applicable. User sessions are great for understanding the behaviour of viewers, as data about which links the viewer visited and how much time the visitor has spent on each link.. This raises a few problems, for example the activity of the users who might have remained inactive for 30 minutes and then started clicking would be neglected. Like Hits, this metric should not be taken at face value either.
Unique visitors: This metric determines the number of persons who has visited the Website within a definite interval of time. Unique visitors are determined by the combination of user agent (usually browsers, though other programs like search engine spiders and screen readers are included) and IP address. The time interval can be a day or a week or a month- it depends on the type of Website. So if the time interval is one week and a visitor visits it twice in two consecutive days, it would be counted as one visit.
Bounce rate: This is a metric which indicates the percentage of visitors who have left immediately after visiting the site—usually five seconds. For a site with multiple pages, bounce rate can also be measured as the percentage of people who visit only one page, or do not click on any links; this definition depends on the tool you are using. If the number is high, it means that something is seriously wrong with your site, and you need to fix it. For most Web sites, bounce rates between 20 and 35 per cent are acceptable, and rates higher than 50 per cent are worrisome. However, in case of blogs, high bounce rates may not indicate that something is rotten—most people visit a blog to read a single post, or check if there are updates.
Conversion: This is a complex metric which indicates how many visitors to your Web site did what you wanted them to do. Consider a Web site that has a Donate button, Register link or shopping cart. You want your readers to click on these types of links, which are called Goals in anlyticspeak. When a reader clicks on such goals, that’s considered a conversion. High conversion rates (the percentage of visitors that actually click on the goals) indicate that all’s well. If this percentage disappoints, then it’s back to the drawing board. However conversion rate is entirely user determined, depending on the goals.
Traffic sources: Traffic sources are the places from which visitors come to a site. Tools usually divide traffic sources for a Web site into four types—direct traffic (by keying in URLs), referring sites (other sites which have link to your Web site), search engines (visits coming from search engines) and Other (online campaigns, newsletters, etc).
Using This Data
Now that the definitions are out of the way, how do you use that data harvested from your tool for your advantage? From the data about page views, visits, pages per visit, bounce rate, average time on site and percentage new visits over a period of time, you can get a picture of whether your site is delivering the goods. For example, if your site has a total of 20 pages, and the pages per visit is something 5.6, this means that less than half of your site gets visited. In relation to this, you can make use of another statistic which indicates the number of visits per page. This combination will show you which sections of your site aren’t giving you the desired results. With this information, you can now decide what corrective steps need to be taken. It might also mean that you need to tweak the site’s design to make sure that the areas you want people to go to are easily accessible. Google Website Optimizer, a free tool which predicts the change of user data depending on design elements is very useful in this scenario.
Answer the question why your Web site exists, then identify the two or three metrics that are critical to goal conversion.”
Avinash Kaushik, Google Analytics Evangelist
This brings us to keywords and traffic sources. Face it: unless your site is very well known, most of your visits are going to come either through search engines or referrals from other sites. Web analytics tools show which keywords have generated the maximum number of visits—this is a very good indicator of what people expect from your site. Add this information to the traffic from other sites, and you have a fair idea of what your visitors come looking for. Accordingly, you have a number of strategies for improving the visibility of your site, starting from SEO and SEM (check out Get Noticed, Get Clicked, November 2007) as well as patronising your referral sites. Conversely, if you have invested considerably in a referral site and you find that very few visitors have come through that door, you can terminate the agreement before you spend too much. You might have also launched other Internet publicity campaigns like paid advertising and email newsletters -traffic source statistics are a great way to measure the effectiveness of your campaigns. No matter how insignificant these numbers are, they help provide an insight into customer behaviour.
Tools Of The Trade
Google Analytics, the analytics tool from Google is one of the most well known, and is widely used. It is suitable for small sites and is integrated with AdWords. However, the free account has certain limitations: it can be used only if your site has less than five million page views per month: if page views exceed five million then it has to be linked to an active AdWords account that has a minimum budget of $1 (or equivalent). Another commonly used tool to track user statistics—especially on blogs—is StatCounter. This provides rudimentary data like number of visitors over a certain interval: nothing as detailed, or as intimidating as Google Analytics. Microsoft is launching its own analytics package called Gatineau (still in private beta), which will incorporate several new metrics like gender patterns as well. How exactly they plan on finding out visitors’ genders, they aren’t telling. If you don’t mind shelling out the cash, you can use ClickTracks, WebTrends, Omniture, Mint or WebSitStory. All these come in different versions and have a wide range of uses. These tools are more suitable if you host your own site, or have a need of getting customised reports. Mint, for example, provides an installer which has to be downloaded and installed on the server.
At a basic level, the data generated by these tools is self explanatory, but to make sense of that data and make appropriate changes to your site, it takes a professional. Companies who maintain big Web sites have dedicated teams of analysts to monitor the latest trends of visitor behaviour, and help businesses target audiences better. Suvomoy Sarkar, Head of Analytics at Info-edge (which is in charge of Naukri.com’s traffic analysis) feels that analysing the search histories of users can tell companies what customers are looking for, while also identifying the most important keywords you should optimise your site for, or the Ad-words you need to buy. “Web Analytics has impacted a many-fold increase in profit from our online marketing efforts, and nearly two-fold increment in the user experience for our search,” says Sarkar.
The Bottom Line
Though Web analytics seems to be a forbidden domain with an extremely high learning curve, it’s not very difficult to apply some small insights to optimise a Web site. Avinash Kaushik, Google Analytics Evangelist and author of Web Analytics: An Hour A Day, opines that even without going deep into technicalities or bellyaching about low unique visitors or hits, there are plenty of things that can be fixed by a cursory look at an analytics report. “Look at the bounce rate for your [promotional] campaigns…and understand which of these campaigns stink. Or look at the bounce rates for your top entry pages…. you’ll immediately know which pages are not even engaging enough to get your visitors to make one click.” Fix the obvious, is the message, and we couldn’t agree more with that.