Twitter is looking to harvest the content of tweets by giving advertisers the capability to target to people based on what they're saying and whom they follow.
Its new product, which was beta-tested by brands including Walgreens but launches today, consists of about 350 specific interests in 25 broad categories. Those categories are populated with users based on the content of their tweets (but only retweets or mentions that Twitter classifies as an "engagement"), whom they follow, and Twitter's own black-box algorithms that can make educated guesses about user interests, according to head of product marketing Guy Yalif.
Keyword targeting had previously been available for promoted accounts (where marketers pay for each new user who follows them) and for promoted tweets in search, but not for promoted tweets in user timelines, which is Twitter's core ad product. Those could only be targeted to a brand's followers and users that Twitter deemed similar to them, making them a blunt instrument.
Tapping into the vast digital exhaust of its 140 million monthly active users offers a better value proposition to advertisers. For example, the "movies and television" category includes more granular interests such as "animation," "Bollywood" and "foreign," while political advertisers will have the option to target to Democrats and Republicans, as well as conservatives, liberals and non-partisan voters.
There's also an option to create custom segments and target to users who are similar to the followers of a specific Twitter handle. However, Twitter still isn't providing reach or volume estimates to its advertisers. The ad tool also won't show the size of the audience segments for each interest, but Mr. Yalif said they're all in the millions.
Mr. Yalif also noted that Twitter has long been discerning user interests and already used the data to populate its "Discover" page, where it recommends content to read and users to follow. It would appear that Twitter is now making its value proposition more like Facebook's, since the latter often touts the impact of its reach paired with granular targeting options, but Mr. Yalif made a distinction.
"We think it's differentiated because [Twitter's interest targeting is ] based entirely on public signals people are sending," said Mr. Yalif, who compared Twitter's ad system to paid search.
Walgreens social-media strategist Eric Gottloeb said engagement rates on Twitter were up for the duration of the two-week test . While Twitter isn't yet an "always-on" platform for Walgreens in terms of paid media, he said he expected the targeting capabilities to make it a more-compelling platform for digital marketers at the company.
"I thought we could bring it to new areas of the organization who wouldn't necessarily think Twitter was the best way to get the message out," he said.
Since it moved from a first-price auction to a quality-based, second-price auction last summer, the onus is on advertisers to promote tweets that are crafted to drive high engagement. Low engagement for a promoted tweet has a negative impact on its quality score, which subsequently causes the advertiser to win fewer auctions and impels it to raise its bid if it wants to continue running the ad.
For that reason, Mr. Yalif said that advertisers will have an incentive to pick interest targets carefully, since targeting to a group of people who are unlikely to have an affinity for their brand probably means lower engagement rates and a fall-off in quality score.
Twitter today also changed the minimum bid in its auction from 50 cents to a penny, which looks like another move to drive more advertisers into the bidding.
"For advertisers used to driving efficient performance in media, it will make it interesting," Mr. Yalif said.