Data, by nature, represent the information about yesterday’s world. The more the data, the better we can learn about what happened. In conventional wisdom, the bigger the data, the more profitable the analytics will be. This, however, is not always true. The profitability of big data analytics is largely determined by the questions that we can ask.

Take hotel revenue management (RM) as an example. The objective of hotel RM is to influence demand via pricing such that the total revenue will be maximized. Demand for hotel rooms comes from a number of distribution channels, which may range from call centers to brand websites to online travel agencies, and so on. The success of implementing hotel RM requires a good understanding of demand patterns of yesterday and tomorrow.

A classical question that we frequently ask is: how demand behaves across the channels? To answer this question, a dataset of inquiries and bookings over the channels are often collected and analyzed. With the advance of IT technology, the dataset might grow bigger because we are able to collect additional information that were difficult, if not impossible, to get before. For example, for an online inquiry, the history of its clicking path can also be captured if a brand website is appropriately implemented. The use of additional data, in this case, is indeed helpful for us to better answer the question. It not only allows us to analyze how demand is distributed across the channels, but also to predict how it might change. From the viewpoint of traditional business intelligence (BI), this question appears to be perfect for data analytics.

Under the context of RM, this question seems to be not “big” enough. As we know, the ultimate RM objective is to maximize the total revenue for tomorrow. This question, however, has a false belief that limits us from achieving this objective:  the maximal revenue has and will be gained from the existing channels only. This belief is particularly unrealistic in this rapidly evolving world, where tomorrow’s channels might be quite different from those of yesterday. If we continue to ask questions based on this false belief, our data analytics will fail to capture revenue opportunities for the future demand.

Therefore, we need to ask big questions while performing data analysis. For instance, in addition to the above question, we may also ask: How would demand to the other channels migrate if the call center were removed? How would demand be displaced if a new channel like mobile were added? Would the change of channel landscape help increase the total revenue? And so forth. These big questions will challenge us to identify and collect the right data, but the resulting data analytics will help us move closer to the RM objective.

Data do not grow by themselves. Their growth is driven by the big questions we can think of. The big analytics based on the big