In this article, you’ll learn the following: |
Overview
Empirical data is raw, unprocessed information collected directly from a website through observation and experimentation. In your VWO reports, this data is represented as the Unique Conversions per Visitor. These are typically the metric conversions that are observed on your website.
It forms the foundation for all statistical analyses and decision-making processes. Unlike statistical predictions, which can be influenced by various models and assumptions, empirical data presents the factual reality as observed.
Understanding Metric Types in VWO
In VWO, metrics are categorized into two types: Binary and Continuous.
- Binary Metric: A metric that calculates the number of Unique visitors who completed an event is considered binary.
- Continuous Metric: A metric that calculates either the Event totals or the Value of an event property is considered continuous.
Significance of Empirical Data
Empirical data is essential for two reasons:
- Highlights the Sample Size: Empirical data highlights the number of samples that back the statistical conclusions of the study. Conclusions based on low sample sizes are usually not reliable.
- Foundation for Statistics: While statistical estimates provide an expectation on what can happen in the future, empirical data gives an insight into the ground truth that these estimates are based on.
Types of Empirical Data
Empirical data can be broadly categorized into two types: Binary Data and Non-binary data.
Binary Data
Binary data represents outcomes that can have one of two possible values, such as conversions that involve a binary outcome (yes/no), such as:
- Page Visits: Did a visitor view a particular page?
- Clicks on a CTA: Did a visitor click on a call-to-action button?
- Form Submissions: Did a visitor submit a form?
- Purchases: Did a visitor complete a purchase?
For example:
If a website had 1,000 visitors in a month and 50 of them made a purchase, the empirical conversion rate is calculated as
Conversion Rate = (50 conversions / 1,000 visitors) * 100 = 5%
This straightforward calculation provides a direct view of the website's performance without the influence of external assumptions.
Non-binary Data
Non-binary data encompasses variables that can take any value within a range, such as revenue.
For instance:
If a website generated $5,000 from 100 converting visitors out of 2,000 total visitors, the average revenue per visitor is calculated as
Average Revenue per Visitor = $5,000 / 2,000 = $2.5
This figure helps you to understand the direct financial outcome per visitor.
In some cases, non-binary data can also be related to conversions. For example, the amount of revenue generated per conversion is a continuous measure. In such scenarios, there are three key measures to consider:
- Conversion Rate: The ratio of conversions to total visitors.
- Average per Conversion: The average revenue or metric per conversion.
- Average per Visitor: The average revenue or metric per visitor.
Among these, the average per visitor is the most comprehensive measure as it combines the conversion rate and the average per conversion.
How VWO Uses Empirical Data?
At VWO, empirical data is integral to generating accurate and insightful reports. The focus is on the average per visitor for all statistical analyses, as this metric provides a holistic view of website performance. By leveraging empirical data, VWO ensures that the reports reflect the true performance of the website, free from assumptions and statistical biases.
Conclusion
Empirical data is an invaluable asset in website optimization, offering a clear and unambiguous picture of past occurrences and current realities. By effectively leveraging empirical data alongside statistical analyses, professionals can gain a comprehensive understanding of their strategies' effectiveness and make informed decisions to optimize future outcomes.
By understanding and utilizing empirical data, you can ensure that your website's performance is accurately measured and optimized, leading to better insights and more effective decision-making.