Every company has many existing measurements that lead to data. In the last chapter, we discussed which data and how much of it we should use. But not all of this data might help to answer the question we defined in Chapter 8. Therefore, we need new metrics to bundle and aggregate different data sets to form new insights. This is the fundamental task of any metric.
New measurements or new metrics have become important in a world that now generates more and more unstructured data. As explained in Unstructured, unstructured data is hard to correlate and difficult to implement into mathematical models, except when a metric or measurement aggregates this data down to a few data points. In this chapter, we will look at the common pitfalls in creating metrics and measurements. As Figure 10-1 depicts, it is easy to create metrics, but that does not mean they make sense.
Figure 10-1. Discussion at a funeral parlor: inventing new metrics
Lets start first with a shining example of a good measurement meeting the right business question. Google was far from being the first search engine. In fact, at the time of its 1998 launch, it had many competitors, and the boom in Internet search usage was well underway. So what single factor helped propel this company, started by two Stanford students in a garage, to become a company with a marketcap of close to $350 ...