This paper proposes a business-demographics adjusted shift-share analysis. This can be used when data availability does not allow direct association of employment changes to business demographics at the regional level. The method may be also used as an exploratory step before any explanatory econometric work is undertaken as a means of identifying classes of potential control variables. Applying the method to Greek data suggests that firm-size heterogeneity should not be ignored, that local conditions matter more than regional economic structure and that the latter are not symmetrical across sectors when it comes to the effects of business demographics on regional employment or output growth.