

Gender wage equality is something that a majority of modern companies work towards. Measuring and addressing gender pay gap in practice can however be challenging due to the variety of roles, educational paths, experiences and responsibilities that compose any given company workforce. In this paper we propose a methodology to (i) determine the level of pay inequality between men and women within a single organisation and to (ii) suggest adjustments to address it. Our approach is novel in the sense that previous studies were done on homogeneous cherry-picked populations from different enterprises at a country level. By contrast, our model aims at decision support for compensation exercises within a single organization. The methodology is applied in two steps. First we evaluate the health of the overall organisation from the gender gap point of view. We combine expert-driven rules with unsupervised learning methods to identify individuals that are comparable, and then provide metrics to assess the level of gender-based pay gap within these groups. For cases where such a gap is ascertained, we build a predictive model for the gender neutral salary of each individual based on comparison with his/her closest peers within the workforce. Once the individuals that have a deficit in terms of compensation are identified, we test how the budgetary constraints reflect on a catch up plan. We have applied this methodology to a European organisation of about 1500 employees to support the HR department in its efforts to assure competitive compensation. The results confirm the existence of differences in compensation between men and women but also that the adjustments proposed by our models do lead to narrowing the pay gap.