Data pipelines now mediate dashboards, experimentation, reporting, and machine learning, yet their trustworthiness cannot be inferred from uptime or point estimates of data quality alone. Pipelines may continue to run while semantics drift, derivation becomes opaque, change history weakens, or access rights detach from responsibility. Relevant control domains are addressed across adjacent literatures on data quality, provenance, governance, observability, reliability, reproducibility, metadata, security, and DataOps, but those literatures typically stop at their own boundaries. What remains underdeveloped is a pipeline-specific conceptual framework that integrates these domains around a defensible answer to a single question: when is reliance on pipeline outputs justified? This article develops such a framework as a conceptual maturity model. Following a theory-synthesis approach, purposive literature assembly, and explicit maturity-model design guidance, the paper defines pipeline trustworthiness as the extent to which a pipeline can be justifiably relied upon because its outputs, behavior, and control arrangements remain intelligible, dependable, governable, reproducible, and auditable over time. On that basis, it retains eight dimensions of trustworthy pipeline maturity: data quality assurance, observability and monitoring, reliability and fault tolerance, lineage and traceability, governance and ownership, reproducibility and change management, metadata and documentation, and security and access control. It further specifies five heuristic maturity levels—Ad Hoc, Repeatable, Managed, Controlled, and Trustworthy–Optimized—and a profile-first interpretation in which overall maturity is constrained by the weakest dimension rather than estimated through a compensatory average. The contribution does not lie in claiming that the individual dimensions are unprecedented in isolation. It lies in integrating them into a pipeline-specific architecture centered on justified reliance and in specifying a non-compensatory, bottleneck-aware maturity logic. The result is a conceptual framework for distinguishing trustworthy pipeline maturity from governance-only, DataOps-only, provenance-centered, data-mesh, and generic enterprise capability approaches, while offering a more operationally inspectable basis for future empirical work. Its present payoff is conceptual as much as preparatory: it explains why looser, governance-centric, or compensatory architectures misclassify unevenly developed pipelines even before empirical calibration begins.

Data pipelines; trustworthiness; maturity model; justified reliance; observability; governance; data lineage; reproducibility; metadata management; accountability