Explore intelligence
Methodology
A concise view of how observed hiring evidence becomes published market intelligence—and where interpretation must stop.
From evidence to publication
A versioned, inspectable chain
Each stage has a distinct responsibility so public metrics never depend on unfinished processing state.
- 1
Data sources
Permitted public ATS boards and controlled imports provide source-attributed job evidence.
- 2
Deduplication
Exact and deterministic cross-source matching prevents syndicated postings from inflating demand.
- 3
Normalization
Company, title, location, dates, employment context, and provenance become canonical job records.
- 4
Taxonomy
Versioned standard and TDA-internal concepts provide stable codes, labels, aliases, and mappings.
- 5
Signal extraction
Explainable rules detect skills, technologies, tools, certifications, and work functions with evidence and confidence.
- 6
Weekly aggregation
Unique canonical jobs contribute once to ISO-week demand facts; demand share uses the matching eligible-job denominator.
- 7
Trend methodology
Support-aware historical rules classify observed movement as emerging, accelerating, sustained, stable, volatile, declining, or dormant.
- 8
Publishing snapshots
Only completed, quality-gated analytical facts are copied into an approved immutable public snapshot.
- 9
Coverage limitations
Results describe covered observed job advertisements—not every vacancy, worker supply, labour shortage, or guaranteed future demand.
What the public metrics mean
- Observed Jobs are distinct eligible canonical postings.
- Demand Share is the concept count divided by eligible jobs in the same period and scope.
- Signal confidence describes the strength of historical evidence; it is not probability.
- Published dates, calculation dates, and data periods remain distinct.
Interpretation boundary
Demand is not supply.
Job advertisements describe covered hiring signals. They do not establish worker scarcity, salary pressure, complete market size, causal relationships, or forecasts.
Trend labels summarize observed completed weekly history under the published methodology version. Emerging does not mean guaranteed future growth; declining does not mean a capability is unimportant.