There is little detail in this query. It is hard, for example, to say whether something is good or a ‘good measure’ without context. But in general yes. It should really be labelled ‘effect’ size since of itself it does not portray a causal model. We could work out the difference in heights between two groups of people and express it as an effect size without there being a causal link between height and the way the people are split into the groups. Think of it as a descriptive statistic. Like an average. Perhaps a better term would be proportionate difference.
You probably already use effect sizes in real life when you express how much one thing differs from another in a standard format. A percentage difference is one kind of effect size. In social science there are many ways of expressing an effect size – including odds ratios, indices of inequality, R-squared, Cohen’s d, Glass’ delta and so on. Each would be more or less appropriate depending on the data you have. And underying them is a common theme. Cohen’s d really requires two datasets both in normal distribution. Odds ratios are ideal for two binary categorical values. R squared requires two linearly related sets of real numbers. And so on.
Other than above, effect sizes are largely assumption free, and can be be used with any types of cases. Despite some publicised scales, there is no standard interpretation. Whether a specific amount of effect size is worthy of pursuit depends on costs and benefits.