Data Doesn’t Have to be Hard: Three Data Myths and How to Bust Them with Mailchimp with John Humphrey, former Head of Data Platform Product at Mailchimp
John Humphrey, former head of data platform product at MailChimp and current principal at mfact, joined SaaStr live at Workshop Wednesday to discuss three data myths and how to debunk them.
- The first myth is that working with data is hard, but John emphasizes that anyone can start working with data by leveraging their business knowledge.
- The second myth is the belief that everything needs to be done perfectly, but John argues that improvement, not perfection, is key.
- The third myth is the notion of scarcity, and John highlights the resources and solutions available for data analysis.
Every Wednesday we bring you the best speakers in SaaS. 100% LIVE. Sign up for the series: https://bit.ly/3U0IeN0
- 00:00:00 In this section, the host introduces John Humphrey, former head of data platform product at MailChimp and current principal at mfact. They discuss his role in unlocking and activating data to drive better outcomes for customers. John shares that he is currently working on a project called mfact, which aims to help startups unlock the value of their data. He also talks about the hardest thing he worked on, which was building a customer data platform at MailChimp.
- 00:05:00 Humphrey discusses the challenges of reimagining how to use data in a product and the importance of understanding product management. They emphasize the value of learning from different roles and perspectives to gain a better understanding of the complexities involved. The speaker also highlights the misconception that data has to be hard and emphasizes the need to demystify the topic and make it more approachable. They acknowledge the abundance of buzzwords and hype surrounding data and aim to provide practical guidance on how to leverage data effectively to benefit the business and customers.
- 00:10:00 In this section, the speaker discusses three myths about data. The first myth is that working with data is hard, especially because it involves technical aspects and math. However, the speaker emphasizes that anyone can start working with data by leveraging their knowledge of their business and their customer. They explain that business expertise is actually more important than data expertise, and if needed, there are resources available to acquire data skills. The second myth is the belief that everything needs to be done perfectly or not at all. The speaker emphasizes that improvement, not perfection, is key, and even small improvements in metrics can have a significant impact. The third myth is the notion of scarcity, that there are not enough resources or expertise to extract value from data. The speaker argues that there are resources and solutions available, and it’s important to prioritize data investments just like any other business investment.
- 00:15:00 In this section, the speaker discusses the concept of “good enough” when it comes to using data. He compares it to the idea of a minimum viable product (MVP) in product development, stating that sometimes having a simple solution, like copying and pasting data into an Excel file, can be effective instead of aiming for perfection. He uses the example of the fitness app Strava, where users don’t always need to click on the analysis button to understand their metrics, but can contextualize and analyze the data themselves. The moral of the story is that sometimes just having a number to start with is enough, and progress should not be blocked by the notion of needing comprehensive insights. The speaker also addresses the myth of scarcity, emphasizing that as founders of startups, dealing with limited resources is familiar territory. He suggests leveraging existing strengths and resources, such as the data already present in a product’s database, and not being afraid to try things and make mistakes since the stakes are not life or death. Overall, the key message is to focus on making progress and continuously improving, rather than getting caught up in perfectionism or perceived resource limitations.
- 00:20:00 In this section, the speaker shares their experience of starting at Goodreads and being asked to measure user engagement with limited resources. They initially planned to use a data warehouse for analysis but quickly realized they didn’t have one, so they improvised by using existing tools like SQL and Python. They also highlight the importance of using cost-effective solutions where possible and resorting to more expensive options when necessary. Despite the scrappy approach, they were able to build impactful segmentation and provide valuable insights for the company. The speaker emphasizes the need to focus on the right metrics that align with the desired business outcomes, rather than solely relying on basic email metrics. They advise starting with a solid foundation for measuring success and then gradually progressing to measure email performance.
- 00:25:00 In this section, the speaker addresses the question of how to share data learnings with teams and emphasizes the importance of considering the size of the team and their existing methods of sharing information. They caution against creating new, unfamiliar workflows and instead suggest leveraging existing norms for sharing data within the organization. The speaker also mentions that for engineering-heavy organizations, internal wikis or platforms like Confluence may work well, while for other teams like marketing or operations, different approaches may be necessary. They stress the need to understand the organization’s readiness for new tools and processes. Additionally, when explaining data and metrics to non-technical teams, the speaker advises against using jargon and buzzwords and instead suggests speaking in the language of the business, focusing on solving their specific problems and filling in gaps in their understanding.
- 00:30:00 In this section, the speaker emphasizes the importance of speaking in tangible business terms when discussing data. By translating data insights into language that stakeholders can relate to, such as backlog and efficiency gains, data professionals can better leverage data for decision-making. The speaker also mentions the need to understand the context and constraints faced by stakeholders, as this insight can inform the next steps in data analysis. Additionally, the speaker tackles a question about metrics to track for an early-stage B2B SaaS startup with self-service onboarding capabilities. They suggest drawing inspiration from both B2B and B2C metrics, while acknowledging the challenge of limited data volume in a B2B context. They advise not letting perfection be the enemy of good enough and to focus on metrics that provide meaningful insights, even if the data resolution is not as high as desired.
Hype Outweighes Reality in Many AI Data Tools — For Now
- 00:35:00 Humphrey discusses the current state of AI and data tools, acknowledging that while there have been advancements, much of the hype still outweighs the actual substance and practicality of these technologies. However, the speaker emphasizes the importance of focusing on the fundamentals of data structuring and data quality, as these play a crucial role in enabling AI tools to make sense of the data. The speaker also addresses the question of favorite data tools, stating that there are many great tools available in the market and the key is to pick one and start using it rather than getting caught up in the pros and cons of each tool. Additionally, the speaker mentions that reporting platforms, such as Tableau and Looker, can still be frustrating to work with, but the good news is that at the end of the data pipeline, users have the flexibility to use multiple reporting platforms without causing conflicts or confusion.
Early-Stage Startups May Not Be Able to Hire an Entire Data Team at First
- 00:40:00 Humphrey discusses the possibility of working with people on a project basis and emphasizes that for early-stage startups, it may not be practical to hire an entire data function. Instead, he suggests fractionalizing the data needs and providing only the necessary components. He mentions that he can be contacted directly for further discussion.
How to Structure Your Data Warehouse
- 00:45:00 In this section, the speaker discusses the importance of getting the data warehouse right and how it forces businesses to think about what data is important and how to capture and transform it. They emphasize that starting with a strong data lake mitigates the risk of vendor lock and allows for more flexibility. The difference between a CRM and a CDP is also explained, with a CRM being more focused on one-to-one sales motions and a CDP being more automated and one-to-many. The speaker suggests looking at existing S1 filings for inspiration on what data points to focus on when going public, as well as engaging in iterative conversations with advisors to refine metrics and tell a compelling story. Additionally, they mention that certain metrics in the VC world may matter more now due to the changing landscape.
- 00:50:00 Humphrey emphasizes the importance of focusing on fundamentals and solving real business problems instead of getting caught up in the hype of generative AI. They encourage data professionals and startup founders to stay focused on the essential aspects of their customers’ needs and metrics that demonstrate progress. The speaker also advises approaching problem-solving without immediately considering data as the solution but instead understanding the pain points and possibilities for the customers. Only after clarifying these aspects should data be incorporated into the solution. The speaker also mentions the possibility of training AI models with limited data by leveraging existing models and publicly available datasets. They suggest considering prompt engineering as a workaround for dealing with limited data.
- 00:55:00 In this section, the host expresses gratitude to John for his insightful answersa. John also mentions his enjoyment of connecting with the “SaaStr Community” and hopes to see many of them at an upcoming event.