YinYin is an economist by training who transitioned into the tech industry after starting her career in financial services. She has worked at LinkedIn for over four years, beginning in the area of networks and algorithmic fairness. She currently leads a team that focuses on experimentation methodologies and tools for LinkedIn's Ads organization, enabling advertisers to measure ad effectiveness and optimize their spending.
Some Key Takeaways:
Causal inference testing helps determine the impact of different actions by creating controlled, scientific comparisons. This method is especially useful in experiments with small sample sizes, as it allows for more accurate measurement of effects even when data is limited.
YinYin’s experimentation team support multiple business verticals by developing methodologies and tools that can be applied across different products. The challenge lies in understanding each vertical's unique context, which often requires embedding team members to create tailored, effective statistical solutions.
Balancing user experience with ad exposure is a constant challenge, as too many ads can lead to "ad blindness" and reduced engagement.
Variance reduction techniques are crucial for increasing the statistical power of experiments, especially with small sample sizes. By closely matching treatment and control groups or using pre-experiment data, these methods help produce more reliable results in experiments where data is limited. (See Links in Reference Section Below)
Generative AI (GAI) presents new opportunities for content creation in advertising, but systematic vetting through experimentation is needed to ensure accuracy and brand alignment. This emerging field offers potential for more efficient and innovative ad generation methods.
In this episode, we cover:
(00:15) Introduction to YinYin Yu and Her Background
(01:33) Causal Inference Explained
(05:37) How the Experimentation Team Supports Different Verticals at LinkedIn
(07:32) Challenges of Supporting Multiple Verticals Within a Large Organization
(13:28) How YinYin’s Team Supports the Different Advertising Divisions
(15:58) Challenges and Tradeoffs Between Advertising on LinkedIn and the User Experience
(28:11) Variance Reduction Techniques to Help Increase Statistical Power When Running Experiments on Small Sample Sizes (See Links in Reference Section Below)
(31:12) Leveraging General Artificial Intelligence (GAI) for Content Generation and the Role Experimentation Plays as These Technologies Continue to Develop
(37:35) Experimentation Resources That YinYin Finds Valuable and Worth Sharing
Referenced:
Stitch Fix on how they’re using GAI to automate personal style
Meta AI blog and Meta Research Blog
Deep Dive Into Variance Reduction (the most common method for variance reduction)
Towards Optimal Variance Reduction in Online Controlled Experiments (a more advanced method for variance reduction developed by YinYin’s team at LinkedIn in 2022)
🎙️ Interested in elevating your personal brand and thought leadership?
Join one of three available formats: 1:1 interviews with top leaders in experimentation, micro-interviews designed for those just starting their thought leadership journey, and intimate roundtable discussions for 2-3 established in-house brands seeking peer-to-peer insights. Ready to take the next step? Send me a message on LinkedIn to explore how you can get involved.
Whenever you're ready, there are 3 ways I can help you:
1. Full-Service Experimentation: Looking for an all-in-one solution? Our experimentation service covers everything from crafting the strategy to executing the tests. We’re tool-agnostic, so there’s no need to change your setup. With all team members with 10+ years of experience working with companies like Best Buy and Verizon, we guarantee fast and effective results.
2. Conversion Research: Stop guessing and start converting. Our CRO service uses a mix of quantitative and qualitative UX research, including digital analytics and usability tests, to find what works best for your audience. I’ve helped brands like Capella University and Verizon Wireless achieve more with their existing traffic.
3. Brand Booster Roadmap: Future proof your brand experience. Work directly with Brian, Chief Optimization Officer, and Jennie, Head of Brand and former Head of Brand Strategy & Management at Wayfair. You’ll receive a comprehensive, tailored audit and brand strategy guide with data-driven recommendations. This includes brand positioning, online experience analysis, omni-channel content analysis, and a detailed brand experience roadmap. Contact us for a free consultation and to learn more.
About the Host
I’m Brian Poe, Founder and Chief Optimization Officer of Mammoth Insights, where we specialize in helping companies enhance their digital strategies through Conversion Rate Optimization (CRO) and experimentation. With over 15 years of personal experience in the marketing industry, we have a deep passion for uncovering hidden opportunities in data to drive significant improvements in overall business performance. At Mammoth Insights, we offer affordable CRO services, including quant/qual research, market analysis, UX and copy research, A/B testing, and more, tailored specifically for mid-sized to enterprise companies. Our mission is to make complex CRO concepts easy to understand and implement at an affordable price without sacrificing expertise or quality.
