There tends to be two ways of thinking when it comes to big data, which can make for a double-edged sword when considering investment in the space. One way assumes it will sweep the entire landscape, and if you don’t participate, you’re going to be left in the dust. The other is based on the idea that it’s all purely hogwash, says Wesley Chan, director of stock selection research with Acadian Asset Management.

The reality is that plan sponsors need to be educated and truly understand their investment style before treading into the big data ocean, because making a wrong choice can lead to poor investments and disappointment — and applications aren’t going to be one size fits all.

“Managers are going to say that big data is going to generate some sort of alpha,” Chan says, so ask questions. How much should I expect it to generate and where would the value-add be? Or how costly is it and how long will it take to capitalize on this data source?

Big data is basically made of large, unstructured data sets, often from consumer or internal to company data sources. It often requires new techniques and is paired with machine learning, a branch of artificial intelligence based on pattern recognition that can cut through noise in large samples.

The strategy opens up a lot of new possibilities for investors to learn about specific stocks and rms or other assets; idiosyncratic details give more insight into the behaviours of the company — “Information like event-driven strategies or company sales for the next quarter or if there’s a trend in sentiment about how people perceive a firm,” Chan says.

But caution is needed. Investors need to think hard about whether they are targeting a small group of assets or if they need actionable intelligence immediately. Big data can help in both cases, but it may be harder to find an edge in broader diversified or longer-term styles.

For example, photographic data from satellite providers is being used to track parking lot numbers, new infrastructure builds and other pieces of information that could be used for investment decisions. The trouble, says Chan, is that in most cases, the resolution is low and there are strict regulations on how often those photos can be taken. Further, the accuracy and timeliness of such data may be matched and beaten by other, more conventional sources. As a result, this data actually shrinks in terms of applications the longer you look at it.

“You think you’re covering the whole world, controlling for time and frequency of picture-taking, but you’re not,” he says.

Mobile data can also be valuable. It can provide detailed information on exactly where users are and what they’re doing. This data can predict what consumers are going to do in terms of sales going forward, or even where they work and what products they buy.

Ultimately, it’s important to define your investment style and match it with your big-data investment, Chan says. Be clear on your investment thesis and the data it requires because a lot of vendors may not be around in five years. Find the key data yourself and be prepared for a dead end. If big data were easy to find, it wouldn’t be worth anything, and Chan says investors need to be prepared for some kind of investment in this area, as the sources and applications will only continue to mature.