In the task of predictive analysis, predicting the unknown itself is a challenging problem. Moreover, the entry of an unknown variable in the equation makes the task all the more troublesome. Summary-level data are generally inaccurate and lack deep insights, because of which sometimes such unknown variables manage to creep in. Buyer life cycles generally vary in length in spite of which analysts generally tend to work with smaller cycles, which is dangerous because sometimes important marketing decisions are taken based on flawed information. B2Bs are also depending on real-time insights and are scrapping linear prediction models. It is noticed that, combining Big Data with traditional CRM information is also not sufficient because data science involves lot of research and experimentation. Hence we can conclude that predictive analysis derives its success from data governance and collection. Read more at: http://www.marketingprofs.com/opinions/2016/30118/predictive-analytics-has-a-scaling-problem-and-bad-data-is-to-blame