Our Process

Hypothesis

A hypothesis is an idea that requires evaluation or testing to determine if it is true or false. It represents what we would like to investigate. The hypothesis is the most fundamental element in our process.

Hypotheses can be straightforward and easy to validate, while others are more complex in nature. For example, identifying certain features that allow us to analyse churn is clearly not straightforward. The reason being is that first, we would need to identify a selection of variables that exhibit some form of correlation with churn.

Data Collection and Analysis

The data that we use for our field tests is actual data sourced from operators who have given us their full consent to analyse their data and publish the findings while not disclosing any personal details.

As part of our data collection process, we must ensure that we have a large enough sample size to test the hypothesis in question. We then collect the required data and embark on a preliminary data analysis process. If it is deemed that the data collection process is not sufficient, then we recollect the data and reiterate the data analysis until we have ample data.

Once we are confident that the data collection and preliminary analysis stage has been performed to the best of our ability, we discuss the findings internally. After which, we usually engage in another round of collection and analysis. Finally we identify a blog post author and a colleague who functions as a peer review. This is to ensure that we question ourselves sufficiently and adhere to scientific standards.

Publishing results

Once the analysis is complete, then we arrive at a conclusion. Of course, the conclusion does not always confirm our subjective bias, and we do not try to examine the data in a way that confirms our own subjective frame of mind.

This may lead to blog posts for which the absence of a result becomes a result in itself.