Methodology
Ask a question → map it to an approved metric → retrieve authoritative observations → validate → transform only when needed → visualize → explain.
Observed data
Numeric facts shown as observations come from the source payload. DataEnvision does not ask a language model to invent or estimate source values.
Transformations
Time controls can aggregate higher-frequency observations into monthly, quarterly or annual averages. Relationship analysis aligns overlapping calendar-year observations and calculates Pearson correlation only when enough matched observations exist. Correlation is not presented as causation.
Tech Universe
The Tech Universe uses public BLS and Census datasets. BLS JOLTS measures the broad U.S. labor market. Census QWI and BDS industry views use NAICS 51 (Information) as a transparent tech-adjacent proxy. That sector includes information businesses and does not equal the full technology economy, so DataEnvision labels it accordingly.
Projections
DataEnvision-generated projections are labeled separately from observed source data. They are scenarios based on documented model assumptions, not forecasts issued by the underlying provider.
Cross-source analysis
When one provider supports all requested metrics, Auto prefers that common provider. When a question genuinely requires multiple datasets, each source is identified separately.