Big data isn't always best. There’s a practice among scientists of figuring out how small a sample group will accurately represent a population. They use it for finding things like the minimum amount of animal test subjects needed to inform a study. It's also reliable, which is why we base FDA guidelines on such studies.
Small data, a population as little as one, remain relevant even as companies are toppling over one another to derive meaning from the bigger version -- usually millions of data points or more. When it comes to customer success, it's best to use the two together; small data can help reduce customer churn, while insights from big data can lead to upsells.