Quantitative characteristics of indicators are often neglected in their development. Detection of change in an indicator is at its root a quantitative and statistical problem: so, effective use of an indicator depends critically on the design of its sampling program. That design has many aspects.

	Good indicator design is often neglected:
		it is crucial to effective use of indicators
		indicators have to answer the questions asked and be cost effective.

The statistical power to detect change in the indicator depends on the:
	natural variability and sensitivity to pressure of the component(s) being measured
	amount of change one wants to detect
	sampling methods
	careful choice of placement and frequency of sampling
	number of samples collected.

The scope of the indicator and scale of its sampling program have to be considered. The time and space scales of the sampling and the indicator must be compatible. For instance, it would not make sense to sample a slow-growing seagrass species weekly if the goal were to detect changes due to chronic influences of pollution over several years. Conversely, inventories taken at 5-10 year intervals may not detect harmful invaders soon enough to allow relatively simple eradication.

Indicator data are used to monitor natural systems, which have their own scales and boundaries, and support environmental management within designated regions. Sampling must be designed to match these differing boundaries if the data are to reflect the condition of the system accurately and be of most use to managers. It is not simple to reconcile the differing requirements of, say, bio-regions and administrative areas: catchments, for example, are well defined ecological units but their boundaries do not usually coincide with government jurisdictions.

Finally, the indicator's effectiveness and efficiency in representing the state of the system must be addressed. Is measurement of the indicator feasible in technological and logistic terms? Are measurements repeatable? Will the data give a statistically representative sample from the system? Is aggregation of the data within larger regions feasible? Can the indicator be combined with similar indicators collected in other regions? Expert assistance is often needed to help answer such questions.