March 2022 Summaries

PostHog 1.34.0 introduces several new features and improvements aimed at enhancing collaboration and insights sharing, including a new default homepage for projects, smart insight naming, and a comprehensive Data Management tool. This release requires the completion of all asynchronous migrations before upgrading, with specific deprecation notices for SAML users. The update enhances data organization and filtering capabilities, adds staff user controls, and supports sharding and replication for ClickHouse to scale data management. Additional improvements include a streamlined dashboard "add insight" flow, SSO login redirect, experiments with secondary metrics, and shorter insight links. The release also encourages community feedback and contributions, offering insights into open roles and the company’s remote and transparent work culture.

PostHog emphasizes the importance of collaboration, noting that organizations with multiple teammates see a 30% higher retention rate than those with a single user, and aims to enhance collaboration to increase insight discoveries by 20%. The platform has introduced features like robust permalinks and granular permissions to facilitate sharing and control over dashboards, making it easier for large teams to collaborate without conflicts. To aid discovery, PostHog offers a project homepage displaying key metrics and insights, and highlights the benefits of utilizing the Session Recording feature, which triples the number of discoveries. The tool supports two user profiles: those who frequently create insights and those who seek answers to specific questions, with features like one-click saving and searchability of saved insights. Looking ahead, PostHog plans to integrate Slack previews and email subscriptions for dashboards to further embed insights into existing workflows, while also making embeddable graph images available to ensure data context is maintained across various platforms. The company remains adaptive, continuously reprioritizing its roadmap based on user feedback and market needs.

PostHog's newly launched Experimentation suite is designed to enhance A/B testing by integrating seamlessly with its existing data analytics and Feature Flags tools, allowing users to evaluate new features' effectiveness directly within the platform. The suite supports setting target metrics, defining participant groups, and determining experiment durations while employing Bayesian analysis to assess variant performance and statistical significance. Key insights from PostHog's experience with A/B testing emphasize the importance of selecting specific local metrics to minimize external influences and capture relevant data more quickly. The suite also highlights the necessity of balancing trust in data with intuitive understanding and encourages thorough exploration of underlying causes when experiment results conflict with expectations. Additionally, it addresses the challenge of adapting to evolving environments and stresses the difference between web product experimentation and clinical trials, advocating for speed and adaptability over rigid rigor in web contexts. The suite counters the peeking problem by clearly indicating appropriate times to conclude experiments, fostering a more informed and strategic approach to decision-making.

In 2021, PostHog achieved significant milestones, including raising a $15 million Series B, rebranding in just four weeks, migrating from PostgreSQL to ClickHouse, and attaining product-market fit for PostHog Scale, which led to substantial growth in users and open-source contributors. To further enhance its capabilities, PostHog introduced new features during its Launch Week, such as a revamped Data Management UI to address common data management pain points, a scalable architecture through 'Project Sparkle' for improved performance and ease of self-hosting, and an Experimentation suite for A/B testing and multivariate tests with statistical analysis. Collaborations with partners like Altinity were announced, along with the launch of the PostHog Marketplace to offer support and integration services. New features focused on collaboration and session recording were also introduced, emphasizing PostHog's commitment to evolving as a comprehensive tool for product development.

PostHog, initially a simple Python application using Django and Celery with a PostgreSQL datastore, faced scalability challenges as user data volumes increased, prompting a shift to an architecture incorporating Kafka and ClickHouse for improved data processing and querying capabilities. This new setup, while powerful, introduced complexity with additional distributed services requiring enhanced monitoring, migrations, and management, particularly due to ClickHouse's relative novelty compared to PostgreSQL. To address these challenges and ease deployment and maintenance, PostHog focused on two primary goals: improving its test framework and enhancing built-in monitoring. The test framework now includes multiple layers like lint, unit, integration, and end-to-end tests to ensure reliability and compatibility across different cloud platforms and Kubernetes versions. Enhanced monitoring utilizes open-source tools like Grafana and Prometheus to provide critical insights, simplifying maintenance for self-hosting users. Moving forward, PostHog aims to further align its practices to Kubernetes standards, expand cloud platform support, and enhance observability with OpenTelemetry integration.

PostHog has announced a new partnership with Altinity, marking the launch of the PostHog Marketplace, where users can access third-party services that enhance PostHog, such as support, integration, and development services. Altinity, known for its expertise in ClickHouse and data infrastructure, is the first partner and will provide various services to help large organizations using PostHog, including 24/7 technical support for ClickHouse, bespoke training, and Altinity Stable builds for ClickHouse. Additionally, Altinity offers Altinity.Cloud as a backend for users preferring a private cloud environment, effectively managing ClickHouse deployments and turning PostHog into a fully managed service. This partnership aims to facilitate self-hosted deployments, which, while offering benefits like better data control, can be challenging to manage. More partners will join the marketplace in the future, and PostHog invites feedback on this collaboration through its community page.

PostHog aims to enhance its product analytics platform by implementing significant performance improvements to ensure efficient data querying, particularly for large-scale users. Recent upgrades have focused on speeding up property filtering by materializing frequently accessed properties into new columns, resulting in a 25x improvement in query performance. Additionally, PostHog has optimized JOIN operations for large user bases by altering its schema to reduce memory usage and increase speed by up to 10x. To further enhance performance, the company has reorganized data sorting strategies using ClickHouse's MergeTree table engine, achieving a 23% speed increase in query execution by updating the ORDER BY clause to incorporate event types. PostHog has also introduced an asynchronous migration system to manage extensive data restructuring processes efficiently. Looking ahead, the platform plans to eliminate JOINs for person and group data, implement smart caching for time-series queries, and leverage better JSON support in ClickHouse to maintain high performance at any scale.

PostHog has introduced a new Data Management experience to address challenges in managing the rapidly growing volume of data within its analytics platform. This update is part of a broader launch of new features called the "Universe of New Features." The Data Management page, now available on PostHog Cloud and soon for self-hosted users, serves as a centralized hub for managing, organizing, and editing event data. Key enhancements include the ability to categorize events with tags, mark important events as "Verified" to prioritize them in analyses, and use freshness indicators to select relevant events when creating insights. The platform now allows users to easily define and edit events while building insights, without disrupting existing data. The new Activity tab provides near real-time debugging of incoming event data. Best practices for data management include creating clear event definitions, verifying organization-wide events, and leveraging tags for categorization. PostHog plans to continue expanding these capabilities with features like role-based ownership and anomaly detection, and invites user feedback to refine the experience further.

In the realm of database management, schema evolution requires migrations, which can vary in complexity and time consumption, posing significant challenges, especially for platforms like PostHog using ClickHouse. Traditional synchronous migrations can lead to downtime and issues for self-hosted users, prompting PostHog to devise an innovative system for asynchronous migrations. This system allows extensive data migrations to occur in the background, ensuring no service disruption and offering tools for easy management and debugging. The asynchronous model, built on Celery, enables flexibility by allowing migrations to be run within specified version ranges, giving users control over timing while maintaining completion guarantees. Additionally, it incorporates safety measures such as prechecks and health checks to ensure smooth operations in diverse environments, along with a user-friendly UI for monitoring and control. This approach not only addresses the migration challenges for self-hosted instances but also enhances the overall user experience by making the process straightforward and reliable.

Open Graph images, which appear when links are shared on social media platforms, serve as a valuable opportunity for free advertising, yet many companies underutilize this space by displaying only their logos or branding. In a bid to capitalize on this opportunity, PostHog revamped its approach by creating custom Open Graph images tailored for individual web pages, dynamically generating images to reflect current content. This initiative, inspired by GitHub's recent improvements to their link previews, aims to enhance the visual appeal and context of shared links, thereby increasing engagement and click-through rates. PostHog's strategy involves a mix of static and dynamic images, with the latter automatically generated to reduce the manual workload for designers, resulting in more informative and visually enticing previews. This project not only highlights the importance of Open Graph images in digital marketing but also underscores the potential impact of thoughtful design in capturing user attention and conveying key information quickly.