This report mainly focusses on A/B testing of webpages. It is the process of comparing two websites to find out the best one among the two (Hannak et al., 2013). This process is also known as split testing. The following paragraphs will explain the concept of A/B testing in details.
A/B testing is defined as the process of testing two versions of the same webpage. It involves hypothetical testing. The process tests the effectiveness of the two webpages by comparing the two. The two webpages being compared should be of same versions however, should differ in one aspect that will affect the behavior of the user. Websites are mainly designed so that the visitors visit them and decide to buy something. This is the concept of converting the visitors from just visitors to buyers, which is known as the conversion rate. The webpage whose rate of converting visitors to buyers is more wins in the process of comparison. The following paragraphs will explain A/B testing in details. Webpage A can be the currently used one while webpage B can be the modified version. Mainly the e-commerce websites are compared with the help of A/B testing. The factors that determine the behavior of the users are the images used in the website, the organization of the webpage, the colors used to make the website attractive, layout of the webpage and so on. Multivariate testing is similar to that of A/B testing. The only difference is that in multivariate testing a number of variants (in this case webpages) can be compared that has the same versions.
The aspects of a webpage that can be tested using A/B testing are headlines along with sub-headlines, images, links, testimonials, paragraph text, awards and badges, colors and so on.
Various search engines permits A/B testing and states that there is no risk on the search rank of website if A/B testing is used. However, the search rank of websites can be reduced if A/B testing tool is abused by purposes like cloaking. Therefore, Google has adopted some processes to eliminate these problems. One such process is no cloaking. This process says that the A/B testing tool cannot be used for changing spirit of the webpage. If Google finds that scope along with design of webpage has been changed then Google can interpret it as cloaking and the page can be subjected to penalty. Experiments that involves redirects should use rel=canonical tag such that it is beneficial to return back to the original page. Redirection is allowed as long as the page is not directed to unexpected or unrelated content.
Hypotheses plays an important role in the optimization process of search engines. Search engine optimization is the process by which websites are ranked according to relevant search information (Connolly et al., 2014). Therefore, certain hypotheses is required to make by search engines to rank these webpages according to data provided. Optimization without hypotheses is like driving car on a road without any destination.
In A/B testing, two webpages of similar versions are compared according to their conversion rates. Conversion rate is the rate by which the websites converts visitors into buyers. Visitors are attracted by graphics used to design the webpage. The webpage whose conversion rate is high will appear first on search engine (Kohavi et, 2013). A/B testing tools are mostly used by e-commerce websites.
For comparison through A/B testing two customers or variables are required. When the number of variables being tested is small then the data that is produced after comparison is more reliable and accurate (Lipusch et al., 2017). This makes the process significantly fast and interpretation becomes easy.
The A/B testing team consists of Data Scientist that takes care of data that is collected by the analytics tools, and information collected by website owners, developers that develops or manipulates the page without making any changes in the working of the page, designer who are responsible for designing new components in the page, and at last is the faithful companion that checks all new changes that are made in the webpage (Cook, 2016).
The advantages of A/B testing are that the number of visitors that are initiating transaction can be easily viewed. New ideas can also be tested using this testing method. However, there are some disadvantages like the method can be used only to achieve specific goals.
There is another type of testing method that is similar to A/B testing and this is called multivariate testing. The only difference is that A/B testing considers two variables for comparison whereas multivariate testing takes into consideration a number of variants during comparison.
As the said website is that of a business school therefore, the website should consist details of the courses provided and the course fees. Most of the other business school webpage consists of the above said information. This would help the customers visiting the websites gain related information at first sight. If the students get information easily then they will show interest in applying for the courses offered by the business school.
There are several software packages that do A/B testing like Kissmetrics, Unbounce and AB tasty. This part will explain Unbounce and AB tasty (Böhmer Ganev and Krüger, 2013). Unbounce is the software that can test the webpages without any knowledge of HTML (Clow, 2013). AB tasty is another software that helps to test the websites, optimize their conversion rates and modify webpages without writing any code and having any technical knowledge.
Conversion funnel is the term that the e-commerce websites use to explain conversion of visitors to buyers. This method describes the path through which visitors navigate in websites and finally decides to convert into sales. One of the analytical tool that is used to analyze sales data by managers of e-commerce websites is sales funnel software (Jayawardane Halgamuge and Kayande, 2015). This software helps mangers to gain knowledge about total sales that occurred in a single business year. The data collected helps them to analyze the conversion rate that is number of visitors that are converted into buyers. It provides mangers to the insight of future opportunities that will result in increasing profit of e-commerce websites.
From the above discussions, it can be concluded that A/B testing is important to compare the websites and take proper initiatives to improve them. Along with this testing method, there are other methods like multivariate testing that is used to test webpages. However, for the more accurate data produced by A/B testing, it has become common and used widely.
Böhmer, M., Ganev, L. and Krüger, A., 2013, March. Appfunnel: A framework for usage-centric evaluation of recommender systems that suggest mobile applications. In Proceedings of the 2013 international conference on Intelligent user interfaces (pp. 267-276). ACM.
Clow, D., 2013, April. MOOCs and the funnel of participation. In Proceedings of the Third International Conference on Learning Analytics and Knowledge (pp. 185-189). ACM.
Connolly, S.M., Pegden, W.A., Roald, C.B., Jordan, J., Prateek, U. and Yuknat, D.A., Cimpress Schweiz GmbH, 2014. Search engine optimization assistant. U.S. Patent 8,650,191.
Cook, R., 2016. A Crash Course in Empathy-Design Thinking 101.
Hannak, A., Sapiezynski, P., Molavi Kakhki, A., Krishnamurthy, B., Lazer, D., Mislove, A. and Wilson, C., 2013, May. Measuring personalization of web search. In Proceedings of the 22nd international conference on World Wide Web (pp. 527-538). ACM.
Jayawardane, C.H.W., Halgamuge, S.K. and Kayande, U., 2015, December. Attributing conversion credit in an online environment: An analysis and classification. In Computational and Business Intelligence (ISCBI), 2015 3rd International Symposium on (pp. 68-73). IEEE.
Kohavi, R., Deng, A., Frasca, B., Walker, T., Xu, Y. and Pohlmann, N., 2013, August. Online controlled experiments at large scale. In Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining (pp. 1168-1176). ACM.
Lipusch, N., Ebel, P., Dellermann, D. and Bretschneider, U., 2017. Customer Feedback for Startups–A Process Model and Research Agenda.