How Twitch Uses Element Human to Predict What Advertising Audiences Will Love

Understanding and predicting what resonates with audiences is more critical than ever. Traditional advertising metrics often fall short in capturing the nuances of consumer engagement in this dynamic environment. Element Human, a human data company, has partnered with Twitch to provide a cutting-edge solution that leverages human behavioral data and machine learning to predict advertising success and help brands create content that audiences genuinely "love".

‍

The Twitch Benchmark Study: Machine Learned Performance

The study analyzed 74 videos across 37 brands with 17,544 participants to understand what makes ads successful on Twitch. It examined memory, emotion, and self-reported responses using methods like facial coding, eye-tracking, and implicit association tests.

• Trustworthiness and Creativity Matter: Trustworthiness drives purchase intent, while creativity drives brand positivity.
• Better Ads = Better Results: Higher video ratings correlate with increased purchase intent and other positive outcomes.
• Relevance Drives Engagement: Ads that are relevant, interesting, and upbeat are more likely to be engaged with.
• Happiness "Rubs Off": Happy ads positively impact brand perception and association.
• Avoid Neutrality: Neutral ads lead to less engagement, clarity, and uplift.
• Embrace Some Confusion: A little bit of confusion can engage people's rational minds.
• Tailor to the Platform: Gamer-custom content and ads that "fit" well on Twitch perform better.
• Longer Ads Engage: 30-second ads tend to be more engaging than 15-second ads.
• Data-Targeted Campaigns: Psychological profiles can be used to create effective, targeted campaigns.

‍

‍

‍

Above & Beyond Traditional Metrics

For decades, traditional advertising relied on simpler metrics like TV ratings and print circulation, serving mostly homogeneous audiences with a limited number of content producers. The advent of "modern marketing" brought an explosion of advertising formats and non-linear platforms, leading to greater fragmentation and a reliance on focus groups and surveys to understand audience reactions.

However, the emergence of the creator economy represents another profound shift. With hundreds of millions of creators producing content for billions of viewers daily, the complexity is immense. At this scale, traditional market research practices are no longer detailed enough to derive actionable insights. This necessitates a move towards machine learning and the capture of human behavior to understand audience reaction.

Element Human's mission is to fill this gap by building the world's richest human dataset. Our multi-method measurement system captures Attention, Emotion, Memory, and Brand Uplift as people engage with content. Since 2020, this dataset has grown to over 30 billion data points, covering more than 115,000 people, 1,500 creatives, 650 campaigns, 200 brands, and 20 countries. For measurement in this new landscape, it needs to be:

• Timely: allowing for fast feedback to creators.
• Relevant: mapping human behavior data to all stakeholders (creators, audience, brand).
• Impactful: providing tools for creators to measure performance.
• Contextual: enabling testing across different platforms.
• Accessible: utilizing machine learning for cost-effective insight processing.

‍

‍

A Unique and Engaged Audience

Twitch, a leading live streaming service, sees approximately 7 million creators streaming diverse content monthly, from gaming to fashion and cooking. What makes Twitch particularly compelling for advertisers is its highly interactive and immersive live experience, fostering strong communities.

Twitch's audience is predominantly young, with over 70% being between 18 and 34 years old. Crucially, Twitch research reveals that these users are ad-receptive and willing to support brands that benefit their favorite streamers. This highlights a distinct set of values and cultural behaviors that brands must adapt to for effective connection.

Two guiding insights define Twitch's approach to advertising:

• High User Intention: Twitch audiences exhibit very high levels of attention due to the live, immersive nature of the content. The challenge for brands isn't to gain attention, but to make their creatives more impactful within this already engaged space.
• Positive Emotion: There is a significant amount of positive emotion generated on the Twitch platform. This creates an opportunity for brands to cultivate strong positive feelings around their products.

Given these insights, Twitch sought a measurement solution to showcase its unique metrics for advertisers, focusing on emotion, attention, and key performance indicators (KPIs) like Authenticity and Creativity. The solution needed to be highly scalable and learnable to optimize planning, creative selection, and strategy for its Gen Z audience.

‍

‍

What Makes Twitch Audiences Engage?

Element Human's initial phase involved deploying its measurement system on Twitch, testing 74 creatives across the US and UK. This involved a test-control methodology, capturing biometrics during ad exposure, conducting implicit tests, and concluding with a survey. The initial findings provided crucial insights into what makes Twitch audiences tick:

• Enjoyment Drives Purchase Intent: A direct correlation was found: Twitch users who enjoyed an ad were more likely to express a desire to purchase the product.
• Emotion is Paramount: Ads that do not provoke a strong emotion result in less engagement, less clarity of the brand message, and less uplift in subconscious brand associations. This underscores that Twitch users seek content that captures attention, challenges them, and evokes feelings.
• Happiness Cultivates Memory: Specifically, happiness emotion was a standout. Ads that made the audience smile or laugh delivered better subconscious associations for the brand, meaning the brand will be remembered for longer.
• Creativity and Trustworthiness are Key Drivers: Brands that foster an alignment with creativity drive positivity for the brand. Simultaneously, trustworthiness is crucial for driving purchase intent. Therefore, effective brand advertising on Twitch combines creative execution with the establishment of trust.
• Relevance is the "Cheat Code": Ads that are relevant to the audience's interests significantly outperform those that are not. For instance, on Twitch, ads with a gaming theme consistently outperformed non-gaming themed ads across various metrics.
• Platform Fit Matters for Brand Values: Ads that 'fit' better on Twitch demonstrated a greater ability to deliver subconscious brand alignment with values important to the Twitch audience, such as being collaborative, authentic, creative, inclusive, trustworthy, and interactive.

‍

‍

The Twitch Scorecard: Predicting Audience Love at Scale

To scale these insights into an always-on tool, Element Human developed the Twitch Scorecard. This innovative system combines primary research with AI to predict campaign performance before an ad even goes live.

The process involves:

1. Testing Ad Creative: Element Human captures a full range of data from survey responses, implicit associations, and human behavior metrics, including attention, emotion, and brand uplift.
2. Defining Performance Goals: Twitch works to establish specific performance goals for advertising success, such as conversion, view-through rates, shareability, and engagement.
3. Machine Learning: Element Human takes tested creative data and machine-learns it against Twitch's actual campaign performance metrics. This creates a predictive model.
4. Predictive Performance: For every new ad tested, the system predicts how engaged the Twitch audience will be with it. As more ads are tested, the database grows, and the predictive model becomes even stronger.

‍

The Twitch Scorecard is a resource offered to advertisers, providing a single "Twitchscore" for each ad. A higher score indicates greater potential for audience engagement. Advertisers can also see how their ad's engagement ranks against a growing database of tests and across key stages of the marketing funnel. This system transforms research into a practical currency, allowing insights to extend beyond research teams to strategists, planners, and creatives.

The Twitch Scorecard, launched earlier this year, delivers fast results, with most scorecards available within 24 hours of testing. It is used for several strategic purposes:

• Fit With Twitch: Helps new advertisers understand how well their existing content will resonate with the Twitch audience, guiding them on whether to adapt or create new material.
• Creative Selection: Enables brands with multiple creative assets to identify which one is most likely to perform best with the Twitch audience.
• Performance Benchmarking: Assesses how well a creative performs against established Twitch benchmarks and various funnel metrics.
• Diagnostics: Provides detailed data on which elements of a campaign are most effective in driving objectives, helping Twitch build best practices for using emotion and attention in advertising.

‍

‍

The Future is Human-Centric Advertising

As the database continues to grow, the Twitch Scorecard will become even more sophisticated, potentially offering segmented scores by categories, regions, and formats. This ongoing collaboration between Element Human and Twitch is not just about measuring advertising effectiveness; it's about continuously learning and adapting to the "human" element at the core of digital communication. By understanding and predicting the complex interplay of attention, emotion, and behavior, Twitch and Element Human are empowering brands to create advertising that audiences will truly connect with and love.

‍

The Dynamics of Consumer Engagement in Social Media Influencer Marketing

This study explores consumer engagement in social media influencer marketing using the self-determination theory and a stimulus-organism-response framework. It analyzes the impact of influencer, advertisement, and social factors on consumer engagement levels (content consumption, contribution, and creation) through psychological motivators like self-disclosure willingness, innovativeness, and information trust.

• Main Findings:
  • Innovativeness is key for high-level engagement (content creation).
  • Influencers are effective at attracting initial online traffic but have limited influence on deeper engagement.
  • Advertisement information (content) is crucial for higher-level engagement.
  • Subjective social norms significantly influence consumer engagement.
  • Information trust drives content consumption, while innovativeness drives content creation.

‍