The Triad handles sample representativeness by building a conceptual site model (CSM), and then using the CSM to guide additional data collection to fill remaining data gaps and further evolve the CSM. The completeness and accuracy required in the final CSM is governed by what is required to make project decisions. Building and refining the CSM and demonstrating sample representativeness is an iterative process. Dynamic work strategies combined with real-time measurement systems allow much of this iteration to be performed cost-effectively and efficiently during the course of field activities.

Sample representativeness begins in systematic planning. The history of the site, any information concerning contaminant release and distribution mechanisms, along with data from historical or current environmental sampling are used to build and refine the conceptual model. Once the CSM is understood well enough to understand contaminant populations (especially their spatial or temporal boundaries), additional sample collection can be designed so that its results are representative of the property important for decision-making. For example, the property of interest for a risk assessment might be the average concentration of cadmium in soils that a worker might be exposed to during excavation.

The second-generation data quality model used by the Triad approach expects that a number of sampling-related variables will need to be considered and addressed based on the heterogeneity and spatial/temporal distribution of contamination predicted by the CSM. These variables include sample support, sampling design, sample preservation, sample homogenization, and subsample support. These variables are important because the way samples are collected from heterogeneous matrices affects analytical results, irrespective of the analytical method. Controlling these variables is vital for solid samples, such as soils, wastes, sediments, etc. It is also critically important for groundwater collection from wells, and can be important for surface waters and ambient air monitoring as well. It is impossible to control for these variables unless one first has a clear understanding of what decisions will be made with the data, mandating that systematic planning be done. It is also impossible to predict and control these variables without a CSM for how the contamination may be distributed throughout the environmental media in question.