- Published on
Two dashboards disagree on active customers, and an agent asked to settle it invents a third number. The fix for both is declaring the model once.
Data-modeling
All Posts
- data-engineering (15)
data-modeling (15)
- declarative-architecture (10)
- ai (9)
- methodology (7)
- data-architecture (4)
- architecture (3)
- automation (3)
- semantic-layer (3)
- analytics (2)
- data-warehouse (2)
- data-contracts (2)
- semantic-clarity (2)
- team (2)
- daana (2)
- case-study (2)
- unified-star-schema (1)
- analytics-engineering (1)
- daana-cli (1)
- release (1)
- data-warehousing (1)
- integration (1)
- ensemble-modeling (1)
- ux-design (1)
- identity-resolution (1)
- focal-framework (1)
- platform (1)
- Published on
Churn looked wrong on the office TV. An agent walked the platform's declarations to the root cause before standup ended, because there was nothing to guess.- Published on
Skill files and context documents drift because each one is a second copy of the truth. "Talk to your data" holds up when meaning is declared once, in a model, and every layer is generated from it, including the one the agent reads.- Published on
Stop hand-building a galaxy of star schemas. If the business layer is modeled, the consumption layer - Unified Star Schema, mart, OBT, star - is downstream of it and should generate itself.- Published on
Analytical systems are built for questions that do not exist yet. A layered DAS/DAB/DAR architecture with ensemble modeling makes continuous change survivable.