March 2021 Summaries

Jamie Quint, an expert in analytics stacks for founders, emphasizes that the core functionality of tools will remain unchanged despite future advancements. He highlights seven best-in-class tools: Amplitude, Mode, Segment, Fivetran, Snowflake, dbt, and Census. However, a significant gap remains in addressing data quality monitoring and meta analytics on data. Additionally, data teams face new responsibilities, including managing the flow of data between stakeholders and ensuring data quality and compliance. Emerging roles like the analytics engineer can help bridge this communication gap. To start building an ideal analytics stack, focus on fundamental functionalities and communication flows, using Jamie's guidelines as a starting point.

At Census, we've seen companies activate various data to make their business teams more productive. We've seen marketers use product usage data to tailor marketing campaigns to customers with specific characteristics. Customer support agents also use data to proactively reach out to customers based on data in Zendesk. Data engineers have found event data to be a powerful tool, often revealing what users and customers value in services. However, individual events are only as valuable as the insights that can be gained from correlating them with signals that business teams can act upon. To enable this, Census has partnered with Mixpanel to release an integration that allows major cloud warehouses to serve as data sources for Mixpanel, regardless of the warehouse used. This partnership enables marketers and product teams to analyze trusted data from their source of truth within Mixpanel's interactive reports, making it easier to perform custom funnel, flow, cohort, and retention analysis.

When Census started in 2018, the goal was to bring software engineering practices to the world of ops and data integration. The evolution of the data ecosystem is crucial for growth and adoption. Boris Jabes will host a conversation about data on Clubhouse, featuring CEOs from companies that make up the modern data stack. The first episode discussed why Census was built, the past, present, and future of data warehouses, and how data teams will change in the future. George Fraser mentioned that Fivetran's connectors are built around change data capture, enabling customers to update their data more frequently. Arjun Narayan emphasized the importance of understanding the limitations of materialized views and the need for a better coalescing strategy. The conversation also touched on the challenges of data provenance and the potential for automated systems like Fivetran and dbt to solve this problem.

The modern world of marketing relies heavily on data science and attribution modeling to better understand customers and optimize sales funnels. Attribution modeling allows marketers to determine how credit is assigned to each marketing channel or touchpoint for a given conversion, enabling them to make informed decisions about which tactics to pursue and how well a campaign is working. There are several types of attribution models, including first-click, last-click, linear, last non-direct click, time decay, and position-based models, each with their own pros and cons. To implement attribution modeling, marketers can use various methods such as doing it themselves in Python, using Google Analytics, or leveraging third-party tools like Bizible. By assigning credit to marketing channels or touchpoints, attribution modeling helps marketers increase return-on-investment (ROI) and customer acquisition costs.