Presented By: Web Analytics 101 Avoid Common Mistakes and Learn Best Practices June 2012 Lubabah Bakht, CEO, Vitizo
Web Analytics 101, Lubabah Bakht, June 2012 This 12-page whitepaper provides a deep understanding of leveraging web analytics for business performance, i.e. finding actionable insight. Take the easier route by learning directly from a web analytics ninja, instead of taking the difficult, long and expensive route of learning by experience. Learn about good metrics, bad metrics, goals, segmentation, and listening to the customer s voice. Lubabah Bakht CEO, Web Analytics Ninja http://ca.linkedin.com/in/webanalyticsninja Certified by Market Motive as Web Analytics Practitioner and qualified by Google as Google Analytics Qualified Individual. She is also a student of web analytics industry guru, Avinash Kaushik, Google Brand Evangelist and author of Web Analytics 2.0. 2
1.0 WHY USE WEB ANALYTICS? Web analytics is one key component that successful businesses share in today s market. Although many companies have a web analyst working with the collected site data, best practices have still yet to be comprehended and embraced. Many businesses simply do not know where to start with their web analytics data. Should they be testing their landing pages, analyzing the customer s voice, or measuring outcomes? In this whitepaper, I will share some best practices for successful web analysis, i.e. gaining actionable insights. MOST PREFERRED WEB ANALYTICS VENDOR Given that Google Analytics (GA) has a user-friendly interface and it has the most advanced web analytics platform in the market, Vitizo recommends our clients to use GA. Being the most competitive vendor in the market, Google is providing GA as a free analytics program. Given there are not many limitations for businesses to sign up with GA, this whitepaper will solely concentrate on best practices for web analytics on GA. Keep in mind that inserting the Google Analytics Tracking Code (GATC) into your website coding is a part of implementation and not a part of measuring web analytics. There are two methods to measure your web analytics data: the amateur method is reporting clickstream data and the ninja method is analyzing for actionable insights. 3
ALIGNING KPIS WITH BUSINESS OBJECTIVES Before logging into your GA account, the first step is to define your business objectives and targets. The big question on the World Wide Web: what is the purpose of your website s existence? Are you selling products, providing services or creating awareness? Do you have an ecommerce or lead generation site? You are on the wrong path if you are the only person in your company who understands your business s key performance indicators (KPI). KPIs must be comprehended by all the relevant departments in your company including C-level, marketing, web developers, social media managers, and of course, web analysts. Only then will your business be empowered to find actionable insights to meet and recreate the KPIs. You must define your KPIs and they must be DUMB: Doable, Understandable, Manageable and Beneficial. Once the KPIs have been defined, you will be ready to move along to the critical steps for uncovering insights. This whitepaper will provide a deeper insight for successful web analysis including examples of bad and good metrics, the importance of segmentation, and the power that lies within the voice of the customer. 2.0 WHAT S THE DIFFERENCE BETWEEN METRICS AND DIMENSIONS? If you have a Google Analytics account, it is likely you already have an idea of what is a metric. In the most basic terms, a metric is a number, e.g. number of unique visitors, number of page views, number of conversions, etc. There are two types of metrics: 4
count such as visits, page views, and ratios such as conversation rate. A metric will assist measuring your KPI in order to help you understand how you are doing against your objectives. Along with metrics, you need to understand what is a dimension. Dimensions are essentially attributes of visitors, e.g. location, device, browser type. On a cheat sheet, metrics are the columns and dimensions are the rows in your GA reports. With the combination of all the available metrics and dimensions, you can come up with hundreds of reports from your web analytics data. However, 100 reports are neither useful nor powerful. You may want to start off with the magic number of 3 reports to uncover valuable insights. In order to do that, you have to define which metrics are aligned with your KPIs. BAD METRICS There are metrics and then there are bad metrics. Although there may be instances where these metrics are relevant with KPIs and targets, in most cases they are useless. Examples of two types of bad metrics are averages and percentages. WHY ARE AVERAGES BAD METRICS? Averages have a tendency to lie and hinder decision-making. Average-time-on-site is an example of a bad metric. Out of 10 visitors, one person may have spent substantially longer time on the site than the 9 other visitors. The data from this one visitor will skew the real average and give a wrong picture of your website visitors duration of visit. The same can be applied to average pageviews per visitor. The one visitor who looks through all your web pages (maybe even twice) will skew the mean of the other 9 visitors who may have viewed only 2 pages. A skilled web analyst would use other 5
statistical measures and alternative default to reports to understand user activities hidden by averages. IN GENERAL, PERCENTAGES ARE BAD METRICS Percentages are a large part of the web analytics data available within GA. Some are very powerful in giving actionable insights while most are meant to distract amateurs. Excellent examples of good percentages are bounce rates and conversion rates, but these metrics can be improved as well. To address percentages, some solutions include looking at raw data and/or identifying statistically significant data in order to contextualize the data. Without understanding where the percentages stand in relation to each other and the raw numbers, it is difficult to understand whether they are worth your time. SOLUTION: RAW DATA For example, looking at the conversion rate of your referrals will not give an accurate picture. What if some referrals are sending in more or less traffic thus skewing the conversion rate? Solution would be to look at the number of visits from each referrals to avoid misinformation about the conversion rate from the referrals. SOLUTION: STATISTICAL SIGNIFICANCE Analyzing the differences between percentages can be useless as they are often skewed. The numbers do not tell whether the differences are significant or not. By 6
applying basic statistics, you can calculate the standard deviation to identify whether the numbers are significantly or insignificantly spread apart. GOOD METRICS Now that the bad metrics have been addressed, let s look at two good metrics. Keep in mind that there isn t any one-size-fits-all recommendation for good metrics, but these two are pretty solid: bounce rates and abandonment rates. A bounce rate is a good metric to measure the quality of your traffic and measure the quality of your landing pages. An abandonment rate reflects the percentage of people who decide to make a purchase but change their mind before finishing their payment process. BOUNCE RATE High bounce rates indicate that people are coming to your website but not interested to explore any deeper. Low bounce rates indicate people are browsing your web pages one after the other. You can use bounce rate to measure your acquisition tactics. Are you getting high bounce rates from your search traffic, referral traffic or advertising campaigns? The bounce rate will instantly tell you which traffic source is bringing the most relevant traffic to your website. Bounce rates are also good to assess the relevancy of your landing page. If certain pages are receiving high bounce rates, it will be wise to visit those pages and apply conversion optimization strategies to increase user activity. Blogs are the one primary exception when it comes to actionable insights from bounce rates. People s behaviour with blogs differ substantially to websites. Visitors will find a blog, read the blog post, and then leave the blog without reading any other blog posts. Other than blogs, bounce rates tend to be a very insightful metric. 7
ABANDONMENT RATE Abandonment rates are explicitly applicable for sequentially directing visitors from page 1-4, e.g. purchasing an item, signing up for credit card, etc. These activities require visitors to follow a structured path in order to complete their task. Abandonment rate is great for understanding which point customers are dropping off in the structured process. You can use the site abandonment rate, cart abandonment rate and/or checkout abandonment rate to see how you can optimize the payment process for your customers. 3.0 SEGMENTATION & AUDIENCE BEHAVIOUR Simply looking at aggregated data will not provide any actionable insight. It does not allow you to place the data in context with certain factors at hand. Segmenting the data and looking at audience behaviour will provide actionable insights by placing the data in context and analyzing visitor behaviour. SEGMENTATION Segmentation allows you to place data into defined groups and measure metrics in comparison across groups. For example, Traffic Sources categorizes all the visitors into large groups such as search, referral, direct, and campaigns. Without looking at each category, you will be able to differentiate search audience activities from campaign audience activities. The buying behaviours of audience acquired through email may 8
significantly differ from audience acquired from the Google search engine. Audience referred from social networks may engage more with your rich media than audience who entered your site through search engines. Segmentation allows to gain insights on puzzles, such as personas of your website traffic. Aggregate data can rarely, if not never, give such actionable insights. Keep in mind that segmenting traffic sources is not a one-size-fits-all solution for all businesses using web analytics. Segmentation is abstract and requires you to identify which factors have higher probability of providing your business actionable insights. AUDIENCE BEHAVIOUR Another example of segmentation is looking at distribution. As mentioned earlier with averages, a good solution to taking apart average-time-on-site metric is to look at the visit duration report. Each time a new visitor arrives on your website, there is a small package of data downloaded onto the visitor s device to identify the visitor in the next 30 days. This non-intrusive cookie allows GA to track whether a visitor is new or returning to your site. Having the distribution data of number of visits and days to visits, you can identify the threshold number of days that your returning visitors tend to increase or decrease their engagement. You can dig further and segment whether visitors from search engines, paid campaigns or referral tend to visit your website more often. Without segmentation, it is difficult to understand the primary purpose of your visitors and find opportunities for optimization. The creativity lies in your hands as long as it is aligned with understanding your KPIs and outcomes. 9
4.0 VOICE OF THE CUSTOMER Good metrics and reports may provide you with many actionable insights but they do not capture the intent of the customer. Without providing a voice to your customers, it is difficult to gain insight into intent and find true opportunity. Two tools for understanding voice of the customer will be discussed: surveys and landing page experimentations. SURVEYS A survey allows you to directly make a request to the visitor to share their experience and opinion. There are three great survey questions to ask your visitors: 1. Why are visitors on your website?; 2. Were they able to complete their task?; 3. If not, why? These three survey questions guarantee actionable insights as to how to improve customer experience. The first and second questions can be a multiple choice answer whereas the third question must provide open text response. There are two types of surveys: site-level and page-level. Site-level surveys are for the purpose of gathering customer feedback about the entire website, e.g. navigation, loading speed. Page-level surveys are for the purpose of gathering customer feedback about specific web pages, e.g. an article, a product. You need to identify where and what to gather feedback about before choosing which type of survey to insert. 10
EXPERIMENTATION Experimentation is needed to analyze customer behaviour and test presumptions about your website. The customer experience on the web is now much more complex with flash and rich media. Are they helping or nullifying website conversions? Now the search engines predominantly decide where your customers will land on your website. There is no longer one homepage but rather any page that a customer lands on becomes the homepage. Therefore, it is important to understand why people come to your website. Often employees and executives are too close to the company to make unbiased judgments about customer experience. Experimentation eliminates the guesswork. Experimentation takes your website into a lab setting to make a hypothesis, run tests, study results and then implement the findings. Experimentation provides customers with a real voice by measuring their customer experience on your tests. It also encourages creativity and nurtures ideas by not having to solely make changes based on one or two people s presumptions about the customers experience. Experimentation allows to test landing pages at scale, try various modifications, use the statistics to make the final change and grow with the findings. AB & MVT TESTING Although you have some options to choose among vendors for landing page experimentation and testing, the most recommended is Google Website Optimizer. 1 There are two types of testing that can be done with your landing pages: AB split testing and multivariate test. AB split testing allows you to work on a control landing 1 Note that Google Website Optimizer will be phased out over the next few months. This is a recent announcement by Google. The program has been added to Google Analytics as Content Experiment. Visit Vitizo s Press Release to learn more. 11
page and test landing page in order to assess performance. AB testing is not only a great introduction to testing but it also allows you to leverage existing resources and start right away. However, AB testing is limiting because it only allows testing the simple stuff. The other option of testing is multivariate testing. Vendors for MVT include Offermatica, Optimost, Sitespect, and Google Web Optimizer. The first 3 vendors are paid and the last one is free. Unlike AB testing, MTV does not have a control page and test page. Only one page is required for MVT where testing is conducted on only a few elements within that one page. Benefits of MVT include doing lots of testing quickly, requires minimal IT expertise and allows for continuous learning. However keep in mind that to allow for true results there must be radical changes and not just changing the colours of buttons. CONCLUSION 12 pages are not enough to know everything about web analytics. However, it is sufficient for a valuable step into the world of actionable insights with web analytics. If you are interested to gain even deeper knowledge of web analytics, I suggest subscribing to Avinash Kaushik s blog Occam s Razor. Avinash provides monthly detailed blog posts about specific challenges and practices in web analytics. Vitizo offers audits for new clients web analytics data in order to understand what has happened and find areas of opportunities. We will hold your hand from the very first step in defining KPIs, selecting relevant metrics, conducting landing page experiments, and much more. With Vitizo s expertise, your business is guaranteed to grow. 12