Understand the evidence
Creator intelligence glossary
Plain definitions for the comparisons and decisions you make in Kolliq. These describe our workflows; they are not official YouTube scores or guaranteed growth formulas.
Creator-relative baseline
A reference built from a creator's comparable uploads. Its format, age window, sample size and exclusions matter as much as the resulting median.
See the method or workflowCreator outlier
An upload that is unusual relative to its reference group. A ratio describes the comparison; it does not establish why the video performed that way.
See the method or workflowCreator health
Kolliq's contextual summary of available channel evidence. Read the individual dimensions and missing data, rather than treating a label as a universal grade.
See the method or workflowChannel change detection
A comparison that identifies a movement worth investigating. Detecting a change is separate from attributing its cause.
See the method or workflowTopic momentum
A description of movement in the observed topic sample. Changes in channel mix, upload age or format can change the result without a broader demand shift.
See the method or workflowContent experiment
A recorded hypothesis with chosen variables, a baseline, review criteria and a later outcome. Kolliq records observational tests; it does not run randomized packaging experiments.
See the method or workflowContent decision loop
Observe a change, inspect the evidence, choose one test, publish, review and retain the lesson. An inconclusive result can justify collecting more evidence.
See the method or workflowAlgorithm Weather
Kolliq's name for context from observed creator cohorts. It is not an official YouTube signal or proof that the recommendation system changed.
See the method or workflowEvidence confidence
An indication of how much support the available observations provide. Missing history, mixed results and weak comparability should limit the conclusion.
See the method or workflowUse the definitions in a real decision
Reviewed September 14, 2026 by Kolliq Editorial. Read the data methodology before interpreting a result.