Web Analytics

Web analytics is the analysis of the behaviour of internet users. It serves the following key objectives:

  • Monitor the health of a website — Track and measure web traffic to assess performance vis-à-vis benchmarks and metrics.
  • Improve effectiveness of the website in terms of conversion rates and other performance parameters by means of controlled website tests.
  • Improve effectiveness of elements of the marketing mix. For example testing/evaluating digital marketing campaigns.

Web Traffic Data 

Exhibit 13.21  Unless a correction factor is applied, cookie based estimates can exaggerate reach by 2.5 to 3 time. (Source: Lessons learned in Digital Advertising, comScore).

When a visitor accesses a web page, the site’s web server receives relevant information about the visitor’s request and visitor’s computer. This information which is saved and maintained in web server logs, usually contains the following data:

  • IP address of the client (visitor’s) computer
  • Time, date
  • URI (unique resource indicator) of the requested resource
  • URL (unique resource locator) of the webpage that the request emanated from
  • Information about the client’s browser, operating system etc.
  • Information about the requested resource — file size etc.

The data divulges a host of useful parameters such as the users’ geographical location (based on IP address), the time of arrival, the site where they arrived from and, in case they came via a search engine, the search query strings used to enter the site. It also reveals information on pages visited, and the duration of time spent on each page.

In addition to maintaining logs, developers can also use cookies (small text files) and web storage to store information on client computers’ local hard drives. There are a few different types of cookies that meet different requirements. A transient cookie is one that is created at the start of every session, and deleted at the end of the session. On the other hand a persistent cookie outlasts the user session. It is created when the user first enters the server’s website, and is updated each time the user re-enters the site using the same computer.

Web storage, which is essentially an improvement on cookies, allows for greater storage capacity and improved security. It provides two objects for storing data—local and session storage—which differ in scope and lifetime. Local storage persists after the browser is closed, whereas session storage lasts only for the session. Session storage also permits separate instances of the same web application to run in different windows without interfering with each other.

Web storage and cookies are primarily intended to improve the user’s on-site experience. For instance cookies can store information to allow users to re-enter sites without having to log in. The stored data usually includes unique user IDs and user preferences.

The use of web storage and cookies provides for tracking of users via their devices, and determining how many user computers actually visited the site. The IP address cannot be relied to track this information because of ambiguity arising due to proxy servers, caching and so on.

Besides logs, web storage and cookies, an alternative method called page tagging  allows outsourced services to perform web analytics. This process which requires the insertion of a few lines of code on the relevant webpages, has gained wide acceptance since SaaS (software as a service) vendors like Google Analytics and Statcounter  started providing web traffic data analysis for free.

Note that it is not advisable to use methods that solely rely on cookies for audience measurement purposes because cookies measure computers, not people, and because cookie deletion rate is quite high (see Exhibit 13.21). According to comScore, in Asia Pacific, depending on whether they are first or third-party cookies, 30 to 40% of net users delete cookies, as often as 4 to 5 times per month. Based on these rates, comScore estimates that the size of a site’s visitor base will typically overstate the true number of monthly unique visitors by a factor of up to 2.74 times in Asia Pacific. Errors also creep in because people own multiple devices, because they share devices and because cookies can become outdated.

Considering, however, that measurement service providers like comScore and Nielsen can accurately measure deletion rates and other related variables, across geographies and demographics, and thereby estimate the correction factors, a hybrid approach that combines with panel-based methods could yield reliable estimates.

Note too that mega social media platforms like Facebook possess large databases of device IDs that can be used to source relevant information about net users.


A tag (aka pixel or beacon) is a tool for capturing information on a website. The process requires the insertion of a transparent pixel or few lines of JavaScript code onto the webpage.

When a user’s browser loads the webpage containing the tag, the browser executes the tag code and directs the collected information to the third-party’s server. The third-party, usually a marketing or analytics firm, uses the data for performing their services.

Tags are extensively used by online marketers for running ad campaigns, and by analytics firms to perform web analytics. They accomplish a range of tasks including collecting data about users, integrating third-party content such as advertisements, and setting up of cookies for tracking audiences over time. The information on users can be quite extensive ranging from context such as IP address, mobile phone, how they arrived at the site to behavioural. Information on users’ profile may also be obtained from cookies.

Website Intelligence

Web analytics tools use information from logs, cookies and/or page tagging processes to segment site visitors (i.e., devices), and track their progress down the prospecting funnel, from leads to enquiries, enquiries to prospects, and prospects to customers. The software tools keep track of conversion rates at each stage of the prospecting funnel.

It is possible also to keep track of incoming traffic from advertisements placed on external sites. To isolate these visitors appropriate parameters are added at the end of the linking URL.

Here is a list of metrics that are useful for assessing the health of a website and benchmarking performance:

  • Unique Visitors: The number of users who visit the site over the reporting period.
  • New Visitors: The number of first-ever unique visitors (Assumes cookies have not been deleted).
  • Repeat Visitors: The number of unique visitors who visit the site more than once during reporting period.
  • Conversions: Number of visitors who complete a target action such as purchasing something, subscribing to newsletter etc.
  • Conversion Rate: Proportion of visitors who perform a target action.
  • Bounce Rate: Proportion of visitors who leave the site without visiting any page other than their landing page.
  • Abandonment Rate: Proportion of visitors who start a target action, but do not complete it.
  • Cost per Conversion (CPCon): Cost of an advertising campaign divided by the total number of conversions resulting from the campaign. (This requires that the inbound URL from the ad has been tagged so that analytic tools can distinguish the traffic resulting from the campaign).

Considering the vast amount of information as also the large number of metrics, one can easily get inundated while examining the data. It is important therefore to begin with objectives and key issues, and drill down to relevant information that addresses those issues.

Online is one of a number of ways that marketers engage with their customers. In order to improve their understanding of the overall relationship with customers and how each platform contributes to the development of the relationship, it is important for marketers to take a holistic view. They need to interpret website intelligence in the context of other sources of market information.

Conversion Tracking

Metrics that track conversion (e.g. conversion rate, cost per conversion) require the inclusion of snippets of code in the relevant pages on the website. For instance, to track a transaction, the code would be placed on the checkout confirmation page that appears after the user successfully completes the transaction. The code relays the information that the conversion occurred, and this is matched against the visitors who viewed the content (e.g. advertisement) that was placed to trigger the conversion.

Conversion tracking helps marketers measure the return on their investment in advertising, and it helps them gauge how effective their owned media, such as a blog or a content page, is in generating conversions. The data is also used to optimize and personalize advertisements and other forms of communication, so that content is directed to people who are more likely to take the intended actions.

Controlled Website Tests

Analysis of web traffic reveals a great deal about the behaviour of users, i.e. what people do on websites. The data however does not explain why they do what they do, which is crucially important if marketers wish to act on the information.

The understanding of “why people do what they do” typically involves a two stage process — constructing hypothesis or scenarios, and testing those hypothesis. Marketers may need to conduct qualitative research to explore issues, and draw conclusions and hypothesis. For instance, if an advertisement or some other stimuli is not having the desired impact, a focus group discussion can unearth the objective, subjective and emotive reasons why this is so.

As regards testing hypothesis and alternatives, the website itself is the ideal platform. A technique known as multivariate testing allows marketers and web designers to test different combinations of elements (e.g. headlines, paragraph text, images, call-to-action buttons, testimonials etc.) on a web page and assess the impact of those changes on the site’s performance.

Multivariate testing is conducted in a live, controlled environment. The web page is dynamically generated and rotated among incoming visitors so that each gets to see one of the different combinations of elements. The visitors are tracked and their behaviours recorded as they navigate the site, to determine which elements (i.e. independent variables) on a web page make the biggest impact on the site’s performance, where site performance is measured in terms of a specific objective metric (i.e. dependent variable) such as conversion rate.

A/B Testing (or split testing) is a relatively simpler way to test changes on a web page. In A/B testing (or A/B/C or A/B/C/D testing) different versions of the web page are tested to gauge and compare the performance of each version.

If a marketer or web designer wants to choose between one of two or more versions of a web page, in that case A/B testing is the appropriate methodology. On the other hand if she is seeking to optimize the elements of a web page, multivariate testing will reveal which elements she needs to focus her attention on.

The net affords great flexibility for evaluating digital marketing campaigns. It is possible to pre-test campaigns (A/B testing) before they are launched, and continue to evaluate them in real time after they are launched. It is easier to manage, control and execute test programmes on the net. It is also much less expensive. For these reasons, consumer marketing companies may evaluate campaigns on the net before airing them on TV.

Testing can be outsourced easily to analytics service providers by inserting code in the pages that are required to be tested.

SEO and controlled website testing are of great interest as they help to maximize the ROI of a website. By boosting the ranking of a site on SERPs, SEO helps internet marketers to bring more visitors to their site. By improving the design and content of web pages, multivariate and A/B tests help to increase the probability that visitors will take the desired action once they arrive at the website.

As we conclude this section on Digital Marketing, it is important to appreciate that while web analytics provides comprehensive details of behavioural engagement, it provides very limited understanding of attitudinal and mental engagement, and is weak on ad diagnostics. This is crucially important because as far as advertising goes, quality matters a great deal.

The mechanisms and the key themes of advertising are covered in the chapter, How Advertising Works. In the context of diverse marketing initiatives, varied target audience and best practices in the development and execution of advertising, this chapter imparts an understanding of the ingredients that make advertising effective and impactful.

The chapter Advertising Analytics imparts a comprehensive understanding of audience measurement and engagement measurement.

In a multi-media world, it is important that audiences are measured on an apple-to-apple basis, so that advertisers have an understanding of their total audience across media — digital, TV, press and radio. At present however, the norms of measurement, are divergent.

Average minute audience (AMA), which is the norm for measuring viewership on TV is vastly different from the impressions we measure for online media.

The Advertising Analytics chapter discusses these issues and relates the developments towards solutions to measure viewers across all platforms.

The chapter also devotes considerable attention to the measurement of attitudinal engagement, which is key to our understanding of the quality of advertising.

Previous     Next

Note: To find content on MarketingMind type the acronym ‘MM’ followed by your query into the search bar. For example, if you enter ‘mm consumer analytics’ into Chrome’s search bar, relevant pages from MarketingMind will appear in Google’s result pages.

Digital Marketing Workshop

Digital Marketing Workshop

Two-day hands-on coaching on Digital Marketing and Advertising, to train participants in developing and executing effective digital marketing strategies.