In a world increasingly driven by or focusing on measuring quality, productivity or other factors it's easy to drown in a sea of numbers. Here are some things you should ask when implementing metrics or measurements. First of all, a note on the difference: measuring is the assignment of a numbered value consistently to something, so that it can be compared with similar data. The number of defects outstanding in a software product is a measure – something you can count. A metric, on the other hand, is a comparison – the metric tells you about what's happening to the thing being measured. To take the example of software defects again, typical metrics might be – the number of defects per function point – the ratio of defects to lines of code – the increase or decrease of the number of defects between two suitable points in time. Journalists and other writers are often taught the trick of asking “the five W's” – what, why, when, where, who – to help guide their writing so that the finished article is meaningful and relevant. These same questions can be used to evaluate both measurements and metrics to determine their relevance and suitability to the situation in hand. WHAT is being measured? A clear definition of what is being measured, and how the measurement or metric is calculated, is absolutely essential to give context and meaning to the number. Data only becomes information when it has meaning. WHY is it being measured? The audience needs to know the purpose of the measurement in the first place. Sometimes it's obvious, but sometimes an explanation is required to help the audience understand what the numbers are telling them. WHEN is it being measured, or should it be measured? Measuring and reporting has a cost. Gathering, presenting and interpreting the data impacts directly or indirectly on productivity. If it's a meaningful output, then it's worthwhile and therefore productive. If it's not, then it should be changed or abandoned. Measuring the output of a programming team may make sense on a daily basis, but it's unlikely to make sense to try to report on hourly. WHERE is it being measured? Choose meaningful points in a process to take measurements, and ensure they make sense (and add value!). WHO? is the complex question. Who is doing the measuring, and how are they doing it? Is the measuring manual or automated? Is the data captured or calculated in a meaningful way, and is the person capturing it or calculating it able to do so accurately? Who is the measurement being interpreted by? Is it presented in a format that the audience can understand, and do they have sufficient supporting information to make sense of it? How frequently do the metrics or measures need to be reviewed, and is there a quality process in place to ensure the review is carried out and acted upon? One final thought: if a data gathering process can't answer the “five W's”, then it's very purpose and existence should be questioned.