February 2023 Summaries

February was a significant month for feature flags, as Statsig introduced a new free feature management offering and explored its importance in product development and optimization during a virtual meetup. This event featured insights into customer use cases, the history of Statsig's SDK architecture, and expert discussions on the workings and best practices of feature flags. Statsig's product manager, MA Seger, and Engineering Lead, Tore Hanssen, highlighted the unique aspects of their Feature Management offering and shared how customers are creatively evolving the platform. Additionally, the meetup delved into CUPED, an innovation designed to enhance experiment speed and reduce bias, with insights from Ronny Kohavi and Allon Korem on fostering a robust experimentation culture. The changing landscape of A/B testing, from the early dominance of Optimizely to transformative strategies at Facebook, was discussed, along with personal experiences at Statsig that emphasize its dynamic and collaborative culture.

Statsig celebrates statistically significant increases in metrics such as daily active users and revenue with a deep green color, but recognizes that increases in metrics like app crashes or page load time are not always desirable. To address this, Statsig allows users to set the directionality of metrics, indicating whether an increase or decrease is preferred, ensuring that data is interpreted correctly. An example is given where setting the "avg_time_to_first_paint" metric to "Decrease" aligns its positive impact with a green color, correcting initial misinterpretations. The text also highlights the importance of a strong experimentation culture and infrastructure, learning from failures in A/B testing, and how Statsig's environment fosters innovative projects and team building. Additionally, it touches on the evolution of web experience platforms, the impact of A/B testing on product strategies, and insights from experts like Ronny Kohavi and Allon Korem.

Statsig is launching a suite of free feature management tools aimed at transforming how product and engineering teams work by providing modern gates, rollouts, and configurations without constraints like seat limits or monthly active users (MAUs). Known for its experimentation engine, Statsig also focuses on feature flags, offering an alternative to costly solutions or time-consuming in-house tools. Drawing inspiration from Facebook’s internal tools, which encouraged a culture of experimentation and rapid development, Statsig aims to help companies of all sizes embark on their experimentation journeys by enabling them to test and ship features quickly and safely. With over 900 companies using its platform to reach billions of users, Statsig's new offering includes advanced targeting and segmentation tools and a user-friendly experience, promoting data-driven decision-making and innovation at scale. Through this, Statsig hopes to cultivate a culture of experimentation that integrates seamlessly with existing workflows and encourages continuous learning and improvement in product strategies.

The tutorial expands on a previous Flutter app project by incorporating user engagement metrics and an experiment involving emoji icons to potentially increase interaction. Utilizing the Statsig Dart SDK, the tutorial guides users on integrating Statsig for logging user selections and setting up an experiment to display emojis as startup logos. The process involves making an asynchronous network call to fetch user data, followed by implementing an onTap event to track user interactions when selecting a startup name. The experiment divides users into test and control groups, with the test group experiencing the emoji feature, aiming to boost engagement metrics. The tutorial emphasizes the importance of experimentation in app development, referencing the CUPED method and insights from industry experts like Ronny Kohavi and Allon Korem, while also reflecting on the evolution of A/B testing and its impact on product strategy, drawing from experiences at companies like Facebook and Optimizely.

Statsig has adopted the use of remote-server side configurations, specifically Dynamic Configs, to enhance flexibility and collaboration in software development and data science projects. This approach allows for modular, shareable, and reusable hardcoded values, facilitating lightweight code review and democratizing config changes among non-technical partners. Dynamic Configs support complex production and development processes, enabling efficient concurrent development between data scientists and engineers. They simplify the management of changes across multiple files and support the rapid implementation of new features, such as timezone support, by allowing easy updates through a centralized source. This method enhances collaboration, clarity of ownership, and reduces the need for extensive code reviews while maintaining an audit log. Dynamic Configs also allow Statsig to handle complex data payloads and schemas effectively, making them a valuable tool for managing configurations and facilitating experimentation, thereby supporting the development of a strong experimentation culture.

Datadog, a cloud-scale observability service, and Statsig, a product observability platform, have enhanced their integration, allowing real-time monitoring to automatically toggle feature gates in Statsig based on Datadog's alerts. This two-way integration simplifies the process by creating a trigger in Statsig, linking it with a Datadog webhook, and configuring monitors to notify the webhook, enabling automatic feature gate adjustments in response to detected anomalies. This advancement improves the monitoring of feature rollouts, accelerates response times to metric regressions, and enhances service reliability, offering customers a more seamless experience. The integration reflects a broader trend in leveraging data analytics and experimentation to optimize product strategies, as highlighted by insights from industry figures like Ronny Kohavi and Allon Korem, and lessons from platforms like Optimizely and Facebook's A/B testing evolution.