How international organizations (IOs) work and what exactly they do depends largely on their personnel. In their personnel policies, IOs are expected by their member states to respect specific rules identifying the desirable geographic distribution of their staff. Across a number of organizations, these rules assign the staffing quotas primarily based on countries’ economic power, allocating most of the quota to a small share of the richest and most powerful countries, providing them with the ability to control the IOs’ work. However, in many cases the IOs’ primary client countries are not the powerful ones, but instead the poorest and thus the least represented ones. From the perspective of the IOs’ functioning (not to mention legitimacy), this is a major problem. If the IOs follow the rules on desirable representation, they are bound to lack what is referred to as soft information about their client countries, the thick information about the local social, cultural, and institutional conditions, conditions they nevertheless need to respect in their work. Based on this tension between the powerful states’ effort to control the IOs and the IOs’ functional need for soft information and hence for more equitable representation, I expect that IOs will use their partial autonomy in personnel policies and counter-act the formal desirable geographic distribution rules. To assess the argument, I study quantitatively the staffing policies of IMF, UNDP, UNESCO, and WHO. In accordance with the formal rules I find that the developed countries are massively over-represented in absolute terms. Yet, I also find that the poorest countries, the IOs primary clients, are over-represented as well, if only in relative terms, in comparison to their assigned quota. This pattern of relative over-representation is strongest in WHO and weakest in IMF. Further analysis suggests that it indeed follows the functional information logic outlined above.