Web Analytics

Web Analytics

Web analytics is the use of web data with the purpose of helping businesses goals such as increased revenue, sharing, or usability. The measurement, collecting, and reporting is often done by a third party web analytics company which provides the information for the company to analyze and make decisions.

Uses

In its most basic form it serves as a collection tool which measures web traffic, but can be much a more detailed tool that can help with market research, asses effectiveness of a website, and facilitate A/B testing. In many instances, companies can use web analytics to track their competitor’s websites to analyze their effectiveness.
Web analytics can be used both qualitatively and quantitatively since most companies can analyze the data to inform qualitative decisions. Additionally, web analytics is effective in helping the business offline as well as online because the path a customer takes to the company website is tracked, unlike the physical movements into a store.


History of Web Analytics

  • 1993: Founding of Webtrends

Webtrends is widely considered to be the first commercialized web analytics company, by selling software that could be installed by website owners to perform analysis of their sites. This company was revolutionary in terms of getting web trends on the map as an industry. Other influential companies that emerged around the same time include: NetGenesis, Accrue, Omniture, Urchin and WebSideStory. As these respective companies developed, two different models of web analytics developed: software retailers and application service providers. With the first approach, web analytics companies would install software onto clients servers, while the second approach used browser tag based analysis hosted on an external server controlled by the company.

  • 1993: Creation of Log Files

Log files record information about viewer activity on a given web page. This information usually includes things like a visitor's IP address, the time stamp, length of visit, whether or not someone is a new or returning visitor, etc. With the creation of log files, came the coining of the term "hit", or an instance of a viewing of a particular web page, measured by page refreshing. In the early stages of web analytics, log analysis required that website owners have access to their server log files and know how to interpret this data.

  • 1996: Hit Counters

Hit counters, or web counters, are a specific web analytics software that displays the number of "hits" or visitors on a particular page since the installation of the software. Although many web analytics companies had been counting page visitors, in 1996 this software was released to the public. With this software, the data was automatically analyzed and presented to the website owner. Often, hit counters would be installed so they could be seen by the viewers, incrementing in real time so owners and page visitors could access the information.

  • 1997: JavaScript Tags

In 1997, JavaScript tags were created as a new method of data collection and an alternative way to track internet activity. They are still seen to be the most popular method of data collection because they have the ability to capture a larger variety of information. Unique codes are attached to each page on a user’s website. Each time the specific page is visited, the JavaScript code executes and generates a call to the server to record statistics. Some of the benefits of this type of data collection are that it can track individual pages accurately, it uses cookies to identify individual visitors, and the data is more transparent. With JavaScript tagging, website owners have more flexibility and control over what is tracked.

  • 2005: Google Analytics and Urchin

Urchin Analytics was one of the early web analytics software packages developed in 1998. Its initial goal was to find ways to understand customer engagement with a website or web service. It was one of the first software packages that had the ability to quickly parse information from server logs, while simultaneously intending to cater to people who were not programmers. In 2005, Urchin was bought by Google, creating the beginning of Google Analytics. This purchase revolutionized the field of web analytics because it made first-class web analysis tools available to anyone for free. Web analytic began to shift from a statistical focus to a more holistic analysis of visitors and how website owners could potentially increase the efficiency of their sites.

  • 2012: Google Discontinues Urchin Analytics

Aiming to streamline where they were prioritizing product efforts, Google began discontinuing multiple different softwares they were utilizing in order to concentrate their focus to fewer packages. This included discontinuing use of Google Labs as well as Urchin Analytics to focus more on other platforms such as Google+ and further developing Google Analytics.

  • 2013: Angelfish is Created

Angelfish was created in response to Google’s discontinuation of Urchin analytics. It was essentially an updated, or modernized software that was more flexible in terms of the types of devices that it could adapt to and operate on, because it was developed as a self-contained application.


Web Analytics Process

The basic process of web analytics is collection of data, processing of data to useful information, understanding key business strategies, and developing online strategy. This process is not linear but each step drives the step before and after it for full efficiency.[http://jimjansen.blogspot.com/2011/06/basic-web-analytics-process.html]

The collection of data is often done through time stamps, query terms, and . Next the data becomes useful metrics through understanding the context such as the time spent on certain pages or the amount of unique visitors. Once the raw data is understood as information, it can be assimilated into business strategies to them make conscious decisions about key performance indicators, also known as KPI, such as the average order value that is purchased or conversion rates on a website. Lastly, web analytics is essential in formulating online strategies which are in line with the companies overarching mission and goals such as market share or awareness about a cause.


Logfiles, Cookies, and Pagetags

Logfiles

Logfiles refer to the data collected by the web server that is independent of the visitor's web browser. All the data is logged by the web server and stored in a text file on the local hard drive. This technique is called server-side data collection and it is used by mostly standalone software vendors to record all the requests made to the server including pages viewed, pictures, and PDFs.

Cookies

Cookies are locally stored text files that the web server sends to the web browser so that it can keep track of the visitor's activity on a particular website. Cookies can be used to determine how many first-time and repeat customers visit a website, how many visitors return each cycle, and how much time has passed between each visit for a returning customer.

Pagetags

Pagetags collect data via the web browser and send the information to remote data collection servers. This technique is called client-side data collection and is used mostly by outsourced, Software as a Service vendors.


Strengths of Web Analytics

All methods of data collection provide real time feedback to the client which is crucial for efficient and responsive customer service. Second, historical data can be reprocessed easily so returning visitors have no issues with "forgotten" information. Lastly, these methods of data collection have become very smart. For example, they can differentiate between partial and complete downloads which is important for clients because it provides an accurate statistic of what they have sold which they can use for later marketing strategies.

Weaknesses of Web Analytics

For as smart as web analytic tools have become, there still remain a few issues, especially in terms of interpretation of the presented data. Many clients take the data at face value which raises the issue of accuracy. Logfiles cannot distinguish robots from actual visitors thus it counts robots as visitors and overcounts the actual amount of visitors to the website. Also, incorrectly setting up pagetags can cause data loss and lead to an underestimated number of visitors. With cookies, they can be manually deleted by the visitor and lead to returning customers being counted as new customers which inaccurately represents the actual visitor pool1


Case Studies

Snapchat Metrics

Snapchat is an effective platform to advertise on since millions of users are active on it daily. In conjunction with formal analytics, marketers can use Snapchat to its advantage and advertise their products or services. With this, there are four key areas that marketers have to keep in mind when it comes to the effectiveness of their marketing strategies that analytics will be useful for. This includes total unique views, story completions, completion rate, and screenshots.2

Total Unique Views

Total unique views is the number of people who have viewed the first frame of a Snapchat story for more than one second. This allows the marketer to view how many different people have viewed their ad which is essential to understanding two things: How large their customer base could potentially be and whether Snapchat is the proper platform to continuing their advertising efforts. This concept is similar to the functions of cookies, where the difference between returning and new visitors is greatly emphasized to see the impact of their marketing efforts.

Story Completions

To determine whether someone has seen the entirety of a Snapchat story marketers look at how many people have viewed the last frame of the story. This allows marketers to determine whether their advertisement was eye-catching enough to retain the attention of their viewers who may become buyers of their product or service. This technique illustrates the strength of current web analytic tools and how they have become smarter in that they can distinguish between partial and complete downloads, and, in this case, partial and complete viewing of the Snapchat stories.

Completion Rate

Completion rate is the percentage of people that started viewing the story compared to how many of them saw the last part of the story. This is another metric of engagement to determine the effectiveness of their advertisement. This method is similar to the pagetag technique as the data is sent only to the marketer.

Screenshots

Another tool to check for engagement, marketers look to see how many people have screenshot their stories. Some advertisers even use the screenshot feature as a polling method to see which designs appeal to their customers more or which direction they should go in their services. Recording the number of screenshots of a story may be equivalent to likes on Facebook or Youtube. This is another example of a method similar to pagetags.

The Leading Hotels of the World

The Leading Hotels of the World is a group of more than 375 hotels and resorts spanning 75 countries. The company does not actually own any hotels, but serves as a reservation service as well as advertising, public relations, and quality control. Since it only profits through being effective at conversion rates for hotel reservation or other amenity bookings, it is critical to effectively leveraging web analytics in its favor. LHW bought Google Analytics 360, which is the paid version of the free Google Analytics, to figure out which interactions between the site and customer leads to hotel bookings.
Lunametrics, a consulting group and partner of Google Analytics, stepped in to maximize the benefit of Google Analytics 360. Using statistical models, Lunametrics saw how users move through the page to certain pages. With the information collected, business decisions were able to be made that placed priority the most used path to bookings.
Understanding user preferences would not have been possible without web analytics. By understanding key performance indicators, LHW was able to shift their resources to better targeting and messaging based on the specific user. [http://www.lunametrics.com/wp-content/uploads/2017/05/LunaMetrics-LHW-Google-BigQuery-Case-Study.pdf]

Predictive Analytics

Definition

Predictive Analytics is the use of online data, statistical algorithms, and machine learning techniques to predict future outcomes based on past web data. It allows businesses, both non-profit and for-profit, to go beyond simply knowing the past to utilizing data from the past to anticipate the future.

Uses

Predictive analytics is incredibly useful for both for-profit and non-profit businesses because the relationships between for-profit businesses and their customers and non-profits and their donors are very similar, which means they both benefit from some of the same things, in this case predictive analytics. Both want to spend their money as efficiently as possible to get people to do what they want, whether it is to buy their products or to donate to their cause. These are just a few examples of how predictive analytics is used in both sectors.

  • For-Profit
    • Has the ability to analyze massive amounts of data to provide predictive insights that can lead to effective business strategies3
      • It takes information gathered via technology like the cookies and logfiles discussed above and uses it to show who your customers are, what they are viewing, how often they visit, and so on.
      • This is then used to provide businesses information about who to market to, which products to emphasize, which to stop producing, consumer lifetime value measures, "next best offer" capabilities. It can even forecast future sales!4
  • Non-Profit
    • Drives fundraising effectively
      • Able to pinpoint who is likely to give money and who isn’t so organizations are able to use their limited funds as efficiently as possible by only targeting those that are actually likely to donate
        • Helps non-profits decide who to contact, how often to contact them, how much to ask for from each individual and how best to reach their fundraising goals.5

Growth

Over the past few years, the use of predictive analytics has become more and more popular because of:

  1. Better, cheaper, faster technology
    1. As the has technology gets better, cheaper, and faster, more companies have found its effects of its usage worth the cost of paying for it.
  2. An increase in the amount of available, useful data
    1. Increasingly effective technology, along with a better idea of what information is truly useful, as led to the amount of available data skyrocketing. With more data, more accurate predictions are possible on a larger variety of interest points.
  3. Increased online competition
    1. a. As more and more businesses sell their wares online, competition continues to increase, meaning every tool a business has at their disposal is useful to staying afloat and profiting. As described above, predictive analytics can help businesses direct their resources in the best way possible, maximizing gain while monetary cost.

Process

As the following image demonstrates, there are five main steps in the predictive analytics process.

  1. First, the data that is thought to contain the wanted date must be pulled from wherever it is being held, whether it be in a data bank, floating around on the web, or in technology such as cookies. Think of these first few steps as the process of finding a vein of gold in a mine. First, a larger chunk of rock that contains veins of gold is removed from a mine.
  2. Then, that data must be cleaned, reduced to only what data is actually needed, and then prepared so that it is the way it needs to be to be easily used. This next step is the equivalent of the extraction of the small amount of gold in the larger chunk of rock taken from the mine.
  3. Next, out of the refined data, pinpoint what you truly want down to
  4. With the refined data prepared and the target to predict picked, the next step is to create the actual prediction with the data and statistical algorithms.
  5. Finally, once the prediction has been created, a plan of action based on said predictions can be created and implemented.
?u=https%3A%2F%2Ftse1.mm.bing.net%2Fth%3Fid%3DOIP.ykIJRSen3h7AIOqszIk5pgEgDY%26pid%3D15.1&f=1

Some Potential Pitfalls

  1. Predictive analytics can only predict outcomes of events that have occurred before.
    1. If, for example there is a senatorial race between two candidates that have never run against each other before, the prediction predictive analytics would give would more than likely be completely wrong. This is because predictive analytics uses information from the past to predict the future, and in this instance, there is little to no information from the past to predict with.
  2. The data must be completely correct and correctly refined.
    1. If the information is in anyway faulty, the prediction will be wrong because there will be untrue or misleading data skewing the entire prediction.

Search Engine Optimization and Web Analytics

Within Google

What is Search Engine Optimization?

  • Search Engine Optimization (SEO) - tactics or strategies that increase the success of a website in appearing in "free" or "natural" search results within a given search engine
    • In order to maximize success, website owners utilize different types of analysis to better cater to clients and optimize the popularity of a website.

Landing Page Analysis

  • Landing page analysis relates website visits and actions taken within a particular visit to a page.
  • Information presented may include page statistics (i.e. total visits, new visits, percent of new visits,etc.), ecommerce measurements (i.e. number of transactions, revenue, and conversion rate), and "goal" statistics (i.e. the percent of visitors who took the "goal" action, number of "goals" achieved, and the value of those respective "goals").

Segment Analysis

  • Segment: an isolated subset of data that is analyzed separately from other segments and from the website as a whole.
  • This type of analysis provides more detailed information about a specific portion or function of a website.
  • Allows website owners to compare the performance of different sections of a site.

Structure and Content Optimization

  • Creating a URL that is transparent in relating your website to its content
    • I.E. use of relevant or commonly searched words, being concise
  • Crawl – the exploration of websites by search engine software in order to index their content
  • Session ID – data provided for the identification of a user who is accessing a system currently
  • Directory structure – streamlining the format of a website increases accessibility and allows users to retrieve information faster with easier navigation
  • Keep website content up to date and interesting
  • Focusing on client-focused content and structure can add to the success of a website

SEO and Mobile Devices

  • Mobile websites are formatted differently to accommodate different hardware as well as a smaller screen
  • Although many website designers keep formatting in mind when creating a mobile version of a website, search engine optimization is not always a priority
    • With large portion of visitors to the Google search engine utilize Google’s mobile search page, failing to cater to this type of search disregards a large portion of potential website visitors

Google Analytics and SEO Reports

Visual Explanation


Future of Web Analytics

It does not appear that web analytics will be going anywhere anytime soon. If anything, it will only continue to grow. As new information is gathered, technology advances, new and expanding markets, and online competition rises, all signs point to web analytics becoming faster, more widely used, and a critical component of any business's continued existence in an unforgiving market. More small and medium-sized businesses in the US are starting to use it, and it's use is also expanding rapidly in Asia.6 It is unlikely that either of these trends will stop. For the businesses that already have it, it is just too useful and successful of a tool to get rid of, no matter if they are for-profit or non-profit, to stop using it.

Bibliography
: Clifton, Brian. "Understanding Web Analytics Accuracy." : https://brianclifton.com/pro-lounge-files/accuracy-whitepaper.pdf : Cicero, Nick. "4 Important Snapchat Metrics Your Brand Should be Measuring." : http://www.convinceandconvert.com/social-media-measurement/snapchat-measuring/ : Penn, Christopher S. "The Predictive Analytics Process: Introduction." : http://www.christopherspenn.com/2017/10/the-predictive-analytics-process-introduction/ : Penn, Christopher S. "The Predictive Analytics Process: Picking Variables." : http://www.christopherspenn.com/2017/10/the-predictive-analytics-process-picking-variables/ : IBM. "Predictive Analytics." : https://www.ibm.com/analytics/data-science/predictive-analytics : Davenport, Thomas H. "A Predictive Analytics Primer." Harvard Business Review. : https://hbr.org/2014/09/a-predictive-analytics-primer : Laracy, Mark. "Why Nonprofits Should Be Building Predictive Models." Rapid Insight, Inc. : http://www.rapidinsightinc.com/why-nonprofits-should-be-building-predictive-models/ : Analytics Magazine. "Report Explores Future of Web Analytics". : http://analytics-magazine.org/report-explores-future-of-web-analytics/
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License