June 5, 2020
In an industry where there are many points of friction for influencer marketers and algorithms, artificial intelligence and machine learning solutions seem to pervade all industries and business categories. How can companies set strategies to scale ROI? How can you scale the identification of your creator partners? How can you increase trust in content?
While #SMWONE, CreatorIQHead of Customer Success, Brooke Hennon joined her colleague Bhavin Desai, VP of Product Strategy, to answer these questions and find out how the platform uses advanced data science to create intuitive influencer marketing solutions. Through a holistic, results-based recommendation engine, CreatorIQ combines several data science models that include influencer identification, content assignment prediction and targeting.
Here are the key findings and insights:
- Anyone with a social presence can exert influence
- To scale the reach and frequency of your campaigns, you should rely on similar, popular media
- Predicting content mapping is a useful mechanism for future-proof tasks
Find the best creators for your campaigns
A common reference point among marketers when it comes to finding the right influencers for a partnership. The vast majority have spent a lot of time looking for more influencers who match the desired collaborations or existing relationships and want efficiency. According to Hennon and Desai, this process is currently an art form without guidance, and data science is a key solution to becoming less subjective and more efficient in order to achieve the desired results.
"We use millions of data points for performance, creator approvals, brand affinity and industry alignment to create a referral engine that is trained to identify the best creators for each campaign." Due to the constant evaluation of the performance data, additional technicians can be added to the technician if necessary, which are based on current high-performance creators. This system can also be used by Lookalike developers, but they are specific to a certain group of people (e.g. I want 10 more influencers like this one to be based in London for this next campaign). Beyond location, gender and interests are other elements that can be used to identify the most suitable authors. All of this can be combined with the core goals of the campaign and brand to identify the most suitable creators for each campaign.
“As influencer marketing grows, it has to be scaled to go from head to toe with other marketing methods. We have to do it more efficiently, ”said Hennon. It starts with finding and collaborating with the right influencers with powerful content tailored to your brand.
Increase trust in content by predicting content attributes
Once you've refined your identification strategy, the second step focuses on scaling content performance. When marketers think about the direction of content to inform influencers, they often use little science and rely on subjectivity, much like looking for influencers at the start of a campaign. To fix this, CreatorIQ uses various approaches to identify and drive delivery of powerful content.
"What resonates on TikTok is very different from what resonates on Facebook," said Desai. To eliminate some guesswork, the company relies on visual insights to create data science models that focus on identifying and recommending powerful content. This is done in collaboration with some of the leading virtual recognition engines like Google Vision to analyze tens of millions of content. This information is used to create custom models that can correlate with certain visual and performance attributes that are recognized in the content and make recommendations about what is most likely to perform well. This is particularly informative when you create paid campaigns.
An important aspect: Predicting content allocation is not only useful to build confidence in the present, but can also be a useful mechanism for future-proof tasks in additional campaigns. Influencer marketers often lack the bandwidth to create data-driven letters, and this is a solution that gets the headache out of this equation.
"Another important result of the Visual Insights model is the ability to use data from content not only for the content of referral campaigns, but also to meet refined creator requirements to drive the creation of powerful content," said Desai.
Scaling the target group address and reach
Marketers want to maximize the reach of the best performing influencer content – beyond the fans and followers of partner influencers. In the influencer area, various variables have to be taken into account, from content to people to different target groups. Knowing which levers to pull is a recurring problem. How can there be a more elegant and effective way to expand content?
CreatorIQ addresses this issue with a model that uses influencer data such as demographics, organic and paid performance to drive the creation of lookalike audience segments that are used as input for targeting social platforms
"We’ve seen significant improvements in conversion data compared to the standalone audience approach that is available directly on the platforms, such as Facebook’s ad manager," said Desai. It is important that brands are able to take a subset of influencers that correlate with high performance within a given campaign and then identify additional similar influencers based on their good performance. These can be used to create a “seed segment” that enables a similar target group approach. The immediate result: improvements in advertising spending.
"Anyone with a social presence can exert influence," Hennon said at the end of the session. Build your army of brand ambassadors by recruiting employees or super fans of your product.
Join over 100,000 other marketers expanding their skills and knowledge by subscribing to our weekly newsletter.