Now that advertisers are starting to spend real money on Twitter, the pressure is on to prove that those investments deliver tangible results. Twitter took a step in that direction today by announcing its partnership with the data giant Datalogix to gauge the impact of tweets -- both paid and organic -- on sales for consumer packaged goods marketers.

Datalogix and Twitter find the correlation between tweets and purchases by matching email addresses that have been scrambled -- or "hashed" -- to ensure users' anonymity. Twitter users provide an email address when registering their account, and Datalogix collects emails through loyalty programs. (Datalogix has a similar deal with Facebook to connect users and their purchases.)

Twitter provides data sets consisting of users who've been exposed to a given brand's tweets and ones who haven't to Datalogix, which has a vast repository of purchase data. It can then use email matching to track how many people in those groups have recently purchased the brand's product.

Thirty-five brands in categories spanning beverages, food, wellness, household products and alcohol participated in a test Twitter and Datalogix recently wrapped up, according to Ameet Ranadive, a product manager for Twitter's revenue team. Mondelez International and its brands Oreo, Wheat Thin and Trident gum were among them. He declined to say how many users were part of the test.

The idea going forward is to offer the service -- which the marketers in the test group didn't pay for -- more widely to CPG advertisers. Twitter is currently only making it available to a limited set of clients, Mr. Ranadive said.

According to the Datalogix findings, Twitter users who've interacted with the brands' promoted tweets -- meaning they've taken an action like replying to it or retweeting it -- purchased 12% more of the brands' products in stores than the control group. Users who had merely seen a promoted tweet -- which comes at no cost to advertisers, since Twitter doesn't charge for impressions -- purchased 2% more.

Lastly, Twitter users who had seen the brands' organic tweets purchased 8% more than the control group. And those who saw five or more organic tweets in the measurement period purchased three times as much as the baseline of users who had been exposed. However, causation seems harder to pin down in this instance, since the people exposed to organic tweets from brands are voluntarily following them, which makes them likelier to be brand loyalists.

Mr. Ranadive observed that Twitter started its Datalogix measurement program with CPGs because they tend to be among the more active marketers on Twitter, but that it ultimately plans to extend it to other categories like automotive.

"[CPGs] invest pretty heavily in organic tweets," he said. "[And] for the first time we're able to show that acquiring followers and tweeting to them enhances sales."