Product Sense
A common refrain heard from employers regarding hiring data scientists is that they lack “product sense.” Digging deeper, the cause of this failure is not because data scientists are naive about how products work or how to analyze them, but because data scientists don’t analyze a product in the same way that a product manager does. Specifically, they fail to link technical ideas to the concepts that less technical roles are familiar with.
The natural link that a product manager sees (the chain between a feature and profit) is not something that data scientists have been trained to focus on. Further, the language that is used when analyzing a product has nuances that most data scientists have not been exposed to.
While teaching in USF’s MSDS program, I created a class which would be a miniature version of a traditional MBA competitive/business strategy course. The purpose of this course is to (hopefully) expose data scientists to the core principals of profitability and (even more hopefully) better prepare them for working with product managers and other non-technical co-workers.
Approach
This course attempts to answer one question: “Why are some companies profitable and others not?” We specifically focus on tech companies and start-ups, as that is where most data science students end up working. Understanding this question, and how non-technical people approach this, allows us to link data-driven analysis to a business. Failing to understand this linkage is the root cause of “not having product sense”.
Along the way we also discuss the basics of company formation, role definition (specifically it relates to data science) and the basics of equity and equity compensation.
Final thoughts
This page contains the notes for that course as well as links to the reading. Note that these are lecture notes from which I teach from. They are not perfect representations of the lectures as in-between material and Q&A material from the course is not contained.
Topics
| No. | Topic | Outline | Readings |
|---|---|---|---|
| 1 | Company Formation (Notes) | Basics of (startup) company formation (rounds, equity, etc.). Public vs. Private companies. Valuations and public company equity compensation. | |
| 2 | Private Equity compensation and the story of Eero (Notes) | Public Company Assignment discussion. Mechanisms of private company equity compensation (grants, vesting, cliff, etc.). Discussion of Eero, its formation and acquisition. | |
| 3 | Porters Forces and analyzing TNC companies. (Notes) | Porter’s Five Forces as a means of analyzing industry profitability. Value Chains. Applying Porter’s system to TNC companies. | |
| 4 | Porter criticisms, Information Rules and Data as the new oil. (Notes) | Post-porter notions of long-term profitability. The information Rules Era conceiving data moats (and their failure). | |
| 5 | The “New” and algorithmic moats (Notes) | The failure of data and algorithmic moats. AI as consulting, not technology. a16z and recent investments in “AI”. Looking at Asserts.ai and Ambient.ai. | |
| 6 | Unit Economics (Notes) | Final moat conversation. Unit Economics two ways (CPV vs. LTV/CAC). Blue Apron and moving between unit economic models. | |
| 7 | Data Science Team Organization (Notes) | Models of Data Science organizations within a company. Traits and characteristics for successful data science teams. Understanding data science roles. | |
| Odds and Ends | Tidbits and leftovers (Notes) | More details on LTV calculations, attribution and case study on Warby Parker Clones. |
