Twitter has been beta-testing Gnip’s Audience API since last October, and the social network announced Tuesday that the application-programming interface is now available to all advertisers.
Data product manager John Heywood wrote in a blog post that Twitter spent the nine months since Gnip’s Audience API was introduced at its Flight Developer Conference ensuring that its audience insights product meets the following requirements:
- Deliver information not available via analysis of public content.
- Return insights in real-time.
- Seamlessly integrate alongside existing Gnip APIs.
- Protect user privacy.
Heywood offered more details on the general rollout of Gnip’s Audience API:
The Audience API allows customers to retrieve demographic data across 10 user models, as well as at the intersection of a combination of any two. Realizing that these models also power Twitter’s suite of advertising products, it has never been easier for brands to understand target audiences and create content that resonates with each group.
At Twitter, everything we do is oriented around our users, and the Audience API is no different. Of the four product requirements shared above, the most important has been to protect our users and preserve the trust they have in Twitter that their data will be used in an appropriate manner. The approach we have pursued is grounded in probabilistic data aggregation and differential privacy, and our continued work in this area has allowed us to maintain the same levels of protection while making important product improvements throughout the Audience API Beta—namely, reducing the minimum audience size a customer can query from 10,000 users to 500 and greatly reducing the reporting thresholds on the output of the API.
He also outlined the three new ways that audiences can be defined using the API:
- Followers of any handle: Instantly create audience segments based on the followers of any public Twitter handle, without needing to download IDs from our public API.
- Brands’ owned organic audiences: Create audience segments made up of users that have seen a brand’s owned, organic tweets over the past 90 days. Then dive a layer deeper by constructing a segment made up of users that have engaged (clicked, liked, retweeted, etc.) with the brand’s organic content over the same 90 day period.
- First-party data (via tailored audiences): Create audience segments based on lists of email addresses, phone numbers or mobile advertising IDs, or based on Twitter users who have visited a brand’s website, that have been matched through tailored audiences. For the first time, your brand customers can programmatically harness Twitter’s demographic models to explore the characteristics of groups they know off-platform.
Readers: What are your thoughts on Twitter’s general release of Gnip’s Audience API?