The Pattern
Analytics is not a commodity skill to be automated or assembly-lined. It is a craft -- an irreducibly complex discipline that requires multiple interrelated skills functioning as a "distinguishable whole." Like woodworking or medicine, proficiency comes from years of practice, mentorship, and imitation-then-innovation. The field has been held back by treating it as either pure engineering (just build the pipeline) or pure business (just answer the question), when it is actually a distinct discipline with its own methods, standards, and professional identity.
Where It Appears
- [[01-projects/newsletter/sc-e05-analytics-crafting|SC E05: Analytics Crafting a Way Forward]] -- The founder's clearest articulation: analytics is irreducibly hard and should follow the professionalization path. The woodworking analogy -- "all those interrelated skills to the point where they function as a distinguishable whole" -- captures why no single skill (SQL, Python, communication) is sufficient. The community needs shared galleries of real work, not toy datasets.
- [[06-reference/2026-04-03-analytics-is-a-profession|Analytics Is a Profession]] -- Tristan Handy's argument for formal professionalization. The zoom-level problem is unique to analytics: constantly shifting between business-level framing and record-level investigation. The five-level spectrum from datasets to business recommendations shows the range of judgment required.
- [[06-reference/2026-04-03-data-cleaning-is-analysis|Data Cleaning Is Analysis]] -- Randy Au reframes "cleaning" as craft. Every analysis requires data shaped in a unique way. The reason cleaning cannot be taught generically is that it requires analytical judgment about the specific problem at hand. Calling it "grunt work" devalues the most judgment-intensive part of the process.
- [[06-reference/2026-04-03-datas-big-whiff|Data's Big Whiff]] -- Benn Stancil on the failure to dignify ad hoc analysis. The most important analytical work -- the stuff behind one-way-door decisions -- gets scattered across laptops and Slack threads while dashboards get all the infrastructure investment. The craft is undervalued precisely where it matters most.
- [[06-reference/2026-04-03-analytics-is-a-mess|Analytics Is a Mess]] -- The messy, exploratory phase is not a bug to be fixed. It is the craft in action. There is no capital-T truth in metrics -- every definition is a judgment call. Maturity means planning for the mess, not trying to eliminate it.
- [[06-reference/2026-04-03-recipe-for-data-intuition|Recipe for Data Intuition]] -- Data intuition is a practiced skill, not an innate talent. It develops through repeated exposure to data patterns, anomalies, and contexts. This is the apprenticeship model applied to analytics.
- [[06-reference/2026-04-03-good-data-scientist-bad-data-scientist|Good Data Scientist, Bad Data Scientist]] -- The craft distinction between good and bad practice. Good practitioners communicate proactively, scope rigorously, and tie work to business outcomes. Bad practitioners optimize for technical impressiveness.
- [[06-reference/2026-04-03-reforge-why-analytics-efforts-fail|Why Most Analytics Efforts Fail]] -- Analytics fails when treated as a project rather than a craft. The five root causes -- tracking vs. analyzing, developer vs. business mindset, wrong abstraction level, poor communication, and project vs. ongoing initiative -- are all craft failures, not technology failures.
- [[06-reference/2026-04-03-missing-analytics-executive|The Missing Analytics Executive]] -- The profession lacks senior leadership representation. Without craft leaders at the executive level, analytics remains a service function rather than a strategic discipline.
- [[01-projects/newsletter/sc-017-analytics-arcade|SC 017: Analytics Arcade]] -- The founder's strongest proof-of-craft story: a single unifying metric (Net New MRR) aligned an entire organization and contributed to a unicorn exit. The craft is not just technical skill -- it is the ability to translate data into organizational transformation.
- [[01-projects/newsletter/sc-013-analyzing-in-public|SC 013: Analyzing in Public]] -- The case for open practice. Crafts advance when practitioners share real work, not just polished tutorials. The secretive nature of internal analysis is a barrier to the profession growing up.
- [[06-reference/2026-04-03-embrace-the-grind|Embrace the Grind]] -- The grind is part of the craft. The willingness to do tedious, unglamorous data work -- the kind that looks like magic to outsiders -- is a competitive moat. There is no shortcut.
Tensions
- Craft vs. scale: Craft implies artisanal attention. Organizations need analytics at scale. The resolution is professionalization -- establishing standards, training, and shared methods that allow the craft to scale without losing its judgment-intensive core. This is the same tension medicine resolved with residency programs and board certifications.
- Open practice vs. proprietary advantage: The founder argues for analyzing in public, but companies treat their analytical methods as competitive advantages. The tension is real -- sharing too much erodes moat, sharing too little stunts the profession. The current resolution leans toward sharing methods and frameworks while keeping specific business data private.
- Identity crisis: Is analytics engineering? Science? Business? The "craft" framing is an attempt to sidestep this question entirely -- a craft is its own thing, not a subset of something else.