Facebook ‘Lookalike Audiences’ help advertisers reach users similar to current customers, others in their database
Facebook is testing a new feature called “Lookalike Audiences” that helps advertisers target users similar to those in their Custom Audience databases, a spokesperson from the company tells us. Advertisers in the beta have seen lower costs per action than with traditional targeting options.
Lookalike Audiences can be created after an advertiser has uploaded a list of first-party data, such as customer email addresses, phone numbers or user IDs to make a Custom Audience. Facebook’s algorithms analyze the Custom Audience and produce another audience segment that is likely to have a similar customer profile. The advertiser can then create any Facebook ad type and target it to the Lookalike Audience. No personally identifiable information is shared back with advertisers and Lookalike Audiences can only be used within Facebook, not exported for email marketing or other ad targeting.
The feature is in limited beta from the Power Editor tool in the U.S. Lookalike Audiences cannot yet be created through the main self-serve ad dashboard or the Ads API. Advertisers can optimize for reach, which will return Lookalike Audiences that are larger but a bit less precise, or optimize for similarity, which will return audiences that are smaller but more similar to the advertiser’s existing Custom Audience. Lookalike Audiences can be combined with other interest or demographic targeting options, so an advertiser could limit its ads to a similar audience that lives in Ohio, or a similar audience who is not already a fan of the advertiser’s page, for example.
Lookalike Audiences could be a powerful new targeting opportunity that helps advertisers achieve their goals and makes ads more relevant for users. CEO Mark Zuckerberg alluded to this type of innovation on the company’s fourth quarter earnings call last week.
“There’s a big opportunity in front of us to make every ad that we’re showing a lot better,” he said. “The biggest ways we’re going to do this are by improving targeting and relevance so we can show everyone content that they care more about and by designing better ad products that aren’t just about links and text and images. For targeting, I’m most excited about the work that we’re doing on Custom Audiences.”
Custom Audiences, which enables advertisers to apply their CRM data to social advertising, has already shown promise for advertisers and for Facebook. On the same call last week, Facebook COO Sheryl Sandberg called Custom Audiences “some of the best targeting available on or off line today.”
She pointed to how men’s fashion site JackThreads achieved a 30 percent lower cost per acquisition than other platforms by using Custom Audiences and saw a 6x return on ad spend. A few months ago, AdParlor told us that one of its clients, a travel company, saw 25x return on investment using Custom Audiences.
Lookalike Audiences would allow businesses to reach users who aren’t in their database but who are similar to those that are, based on the type of Facebook data that advertisers can currently target against. This algorithmic approach could be more effective than having advertisers select their own demographic and interest-based targeting. Facebook tells us that an online travel site saw 70 percent lower costs per action using Lookalike Audiences, and an online shopping site experienced 56 percent lower CPA and 94 percent lower costs per checkout.
Ads and analytics company Optimal offers its own version of lookalike audience targeting using its Audience Matrix and Expander tools. CEO Rob Leathern said he couldn’t comment on Facebook’s features but pointed to a case study where one of Optimal’s clients achieved 60 percent lower costs per fan by identifying the interests of a brand’s most loyal customer base and then targeting friends of fans who matched that profile.
“When a brand has a lot of fans, then their friends become very diffused,” Leathern said. “So we focused on the friends of the fans that looked most like actual customers of the company to build that segment out. It has (and continues) to work well for our large Fortune 1000 customers, and we expect a lot of brands to use interest-based targeting to hone in on specific customer subsegments that can be uniquely productive for them.”
Ads API company Ampush says it is looking to test Facebook’s Lookalike Audiences for a daily deals company and other clients, but it is too soon to discuss performance.