This article originally appeared in LinkedIn.

Remember the initial euphoria of the customer relationship management (CRM) revolution? Starting in the early 1990s, many companies bought the hype and the technology, only to wind up with unusable databases, rebellious sales teams and depleted capital budgets.

There’s no doubt that CRM solutions can deliver real value. Indeed, CRM was the sixth most popular business tool in Bain & Company’s 2015 Management Tools & Trends survey. Yet CRM failure rates remain high: More than 30% of all CRM implementations fail, estimates C5 Insight, a CRM consultancy.

Big Data seems to be going down a similar path as companies try to uncover useful insights about their customers. The arms merchants make big promises: Plug your data into our solution, and we’ll deliver a stream of insights that enable dramatic improvements in marketing productivity, customer experience quality and service operations efficiency. It’ll be a snap for your team, as our technology and your data scientists will do the heavy lifting.

Why is history repeating itself? One reason involves the tendency to mine data without keeping the customer’s best interest in mind. Think of airline yield management systems, or pricing online goods differently based on whether the consumer uses a high-end Apple computer or a cheap Acer. That amounts to plundering, just in a more scientific manner. Likewise, a Big Data approach should avoid plundering and solve customer problems with a big heart.

There are other challenges with Big Data. As my colleagues Eric Almquist, John Senior and Tom Springer explain in a recent report, the situation reflects the difficulty of generating value from existing data, let alone the reams of unstructured data now being generated from social media and other online activity.

Your company can profit from Big Data, though, if you see through the hype to the real work that needs to be done.

Hype: The technology will identify business opportunities all by itself.

That didn’t happen with CRM, and it won’t happen now. Companies that successfully harness the power of Big Data solutions, like Progressive and Uber, tend to start by applying advanced analytics to solve a small number of high-value business problems with their in-house data before they invest in technology.

In the process they learn how to implement solutions organizationally. They also gain insight into operational challenges and come to understand the limitations of their data and technology. They can then define the requirements for their Big Data solution.

Hype: Harvesting more data will automatically generate more value.

Sure, it’s tempting to acquire and mine new data sets produced by social media and mobile devices. Yet large organizations are already drowning in data, much of it held in silos where it cannot easily be accessed, organized, linked or interrogated. It’s generally easier to work with data that has some history rather than attacking new data sets.

Hype: Data scientists will find value for you.

Good data scientists can generate insights. But to profit consistently from those insights, you need an entire operating model that harnesses the data and analytics in a repeatable manner. One telecom service provider did this by partnering its data and analytics teams, the IT division and frontline functions, such as sales, marketing and product development. The business units inject experience and knowledge into the insights from the data scientists, raising the odds that their solutions will be pragmatic and can be implemented at scale. The IT division, which owns the data architecture, figures out how to incorporate new technology and defines the policies and rules that govern them.

This partnership model consolidated and merged two years of customer data from various databases to identify the root causes of value-destroying behaviors. The three teams then defined targeted strategies that could be implemented to turn these value-destroying customers into profitable customers. The result: millions of dollars in incremental revenue.

While the Big Data revolution is real, technology cannot spin straw into gold. The leading companies have assembled deep benches of analytical talent, deployed “test and learn” methodologies and involved their business units from the start in order to turn insights into practical tactics.