Last time you dialed a corporate call center, did the agent seem nicer than usual? That could be because companies now pull data from sites like Facebook and LinkedIn in order to select an operator who is most likely to charm you.
According to the Wall Street Journal, phone giant Sprint (S) and casino operator Caesars are among the firms using the new matchmaker tools, which works by running phone numbers through a variety of databases prior to when a call center agent picks up the phone.
The new system is based on the idea that it’s possible, using thousands of call records, to determine which agents perform best with certain type of customers.
Get Data Sheet, Fortune’s technology newsletter.
For instance, a certain agent’s call history might show she is exceptionally good at serving middle class young women who frequently dine out. Now, if one of those restaurant-loving women are on the line, the data tools route the call to that agent.
How can a company know so much about who is calling? The answer lies in data-gobbling firms such as Afiniti Holdings, which mash together not only social media profiles, but a wealth of other information as well. As the Journal notes:
Here’s Why Big Data Could Be Dangerous
When it comes to social media, the amount of data Afiniti can scrape is limited to what users display publicly on sites like Facebook (fb) and LinkedIn (msft). Nonetheless, such data can offer a wealth of clues about a person’s location, profession, and interests — information that is even more powerful when combined with records like credit and shopping histories that Afiniti buys from third party brokers like Acxiom.
For firms, using all this information to pair call center agents with consumers may translate into more sales and higher levels of customer satisfaction. And for consumers, it might make the experience of phoning a call center more pleasant. But as with any other big data breakthrough, the situation raises privacy concerns and questions of whether consumers should be able to opt out, or just call in anonymously instead.