Why Are Brands Struggling to Track Influencer Campaigns?

With much of brand marketing moving away from Madison Avenue and onto mobile devices, influencer campaigns have moved into the forefront. The influencer marketing industry will reportedly reach $13.8 billion this year, with brands funneling more and more of their budgets toward digital startups and agencies that work closely with top social media stars.

What could go wrong?

Despite rapid growth in the industry, many questions remain over how effective influencer marketing really is. What does an influencer’s follower count actually mean? Why do some influencers with smaller audiences drive more clicks and sales than those with massive followings? What else can influencers bring to their campaigns beyond direct sales?

Without accurate ways to measure and monitor the success of influencer campaigns, most brand marketers are left sitting in the dark. And many of the metrics used to track the success of campaigns on traditional advertising channels don’t work with influencer marketing, where context is everything and the volume of brand mentions can be deceiving. 

One firm that thinks it’s found a solution is Sama, a data-labeling company best known for its data annotation platform. Sama, which has worked with the likes of Google, Walmart, NASA, and Peloton, is doubling down on its strategy to locate, extract, and tag key brand mentions in posts for influencer marketing reporting. It’s one of a number of firms in the space using artificial intelligence and other technologies to detect the differences among mentions.

Through a partnership with Tribe Dynamics, the influencer marketing analytics platform, Sama has expanded its data volume by more than 4x. In the coming months, that expansion should enable Sama’s users to increase mention reporting by over 350%.

How is data volume boosting influencer measurement?

Take the example of a beauty brand like Benefit Cosmetics. From a brand marketing perspective, there is a big difference between a tweet that says, “I love this lipstick from Benefit,” and one that says, “the benefit of using lipstick.” Generic tracking algorithms would attribute the second tweet to a campaign, despite there being no relevance to the beauty brand. However, AI-driven tracking technology uses contextual clues and machine learning to track brands’ share of online conversations.

“There’s no denying that influencer relationships are a growing and critical part of marketing today. That said, the industry’s infancy has revealed several pain points in the effective monitoring and analysis of campaign success,” says Sama CEO Wendy Gonzalez. “When we partnered with Tribe Dynamics in 2017, we quickly realized the need to build robust influencer communities for a fast-growing, global portfolio of brands. Together, we’ve made it a goal to enable brands to effectively measure the ROI of their influencer program through high-quality training data and AI.” 

The rapid acceleration in ecommerce over the past year and a half has created even more opportunities for influencer campaigns, leading Sama to double down on its strategy to help brands take full advantage of the channel.

“While influencer marketing isn’t a new concept, it’s become exponentially more popular over the past several years,” Gonzalez says. “From both a corporate and individual perspective, the potential for success in influencer campaigns has skyrocketed from an alternative form of advertising to a necessary part of every marketing plan over the past five years.”

Gonzalez says Sama’s partnership with Tribe Dynamics has allowed the company’s annotators to assume the tasks of locating, extracting, and tagging key brand mentions in the posts generated by Tribe Dynamics. Sama also handles 60 to 70% of the monthly vetting tasks to identify brand mentions and common language ways of talking about a brand that the algorithm isn’t yet trained on. The company’s partnership database now includes more than 2,500 brands and 200,000 influencers, enabling the tracking of social handles and colloquial ways of discussing different brands. 

“Typically, algorithms might miss slang or abbreviations like #ABH, referring to the popular cosmetics company Anastasia Beverly Hills. This continuous vetting also helps overcome model drift,” Gonzalez says. “In the fast-moving world of social media, training a machine learning model is not a single, finite stage in the process.” 

Sama’s Machine Learning-Assisted Annotation technology enables its algorithm to become more intelligent over time, so even after it’s deployed in a production environment, the company’s steady stream of new training data—and continuous vetting—ensures high levels of accuracy.

Gonzalez says she believes the importance of measuring campaign success is widely ignored across the influencer marketing space. Currently, picking the right influencer with whom to partner is difficult enough, and finding the right way to evaluate campaign performance is even more challenging. She’s hopeful that in time, brands will be able to rely more heavily on automated analysis and training data to accurately measure campaign success and drive more targeted spending.

“As the space continues to grow, automated analysis is necessary to find out whether a brand’s content is delivering its desired results,” she says. “Many brands don’t know that AI can process and assess vast quantities of data before evaluating the value of certain types of content. Not only is it practical, but AI can produce these assessments at scale.”

Stephanie Miles is a senior editor at Street Fight.