High-Value Service Sales Proxy Data

Analysing an A/B test conducted to evaluate the effectiveness of two distinct web page presentations for a company selling a high-value service. Due to the long sales cycle and infrequent nature of purchases, collecting sufficient direct sales data would be too time-intensive. Consequently, the dataset employs a proxy variable—the duration users spent on a detailed interior page describing the service—to gauge which presentation is superior at engaging potential customers and driving sales success. The ultimate aim is to determine if users spend more time engaging with Page B compared to Page A.

Page The specific web page presentation (Page A or Page B) that was viewed by the user during their session.

Time The total amount of time, measured in hundredths of seconds, that the user spent on the web page. For instance, a recorded value of 1.22 is equivalent to 122 seconds.

The dataset originates from the book "Practical Statistics for Data Scientists." It includes 36 valid sessions across the two web presentations. Page A accounts for 58% of the sessions (21 records), and Page B accounts for 42% (15 records). The data is tabular, has two columns, and is distributed in a file named web_page_data.csv. The mean engagement time recorded is 1.41 (hundredths of seconds), with a standard deviation of 0.93. The expected update frequency for this specific sample is never.

This data product is perfectly suited for performing statistical analysis, specifically hypothesis testing, to determine if the mean engagement time differs significantly between the two presentation groups. It serves as an excellent resource for learning about and practising A/B testing methodologies and the strategic use of proxy variables when the true variable of interest is difficult to measure directly.

The dataset covers 36 user sessions viewing the two specified web page presentations (A and B). The scope is limited to these experimental sessions and does not include any geographical, temporal, or demographic data notes regarding the users.

CC0: Public Domain

The dataset is valuable for data science students, beginners, and statistical learners interested in real-world applications of A/B testing. It is also suitable for marketing analysts looking to validate web engagement time as a leading indicator of success in a high-value service context.

Original Data Source: High-Value Service Sales Proxy Data