Linked Data Fragments brought an attention to distribution of costs associated with querying Linked Data between servers and clients. Costs associated with consumption of infrastructure resources such as CPU or RAM are often discussed. However from the business perspective the IT infrastructure is not the only factor in deciding the type of interface for publishing data. Competencies also play an important role and developing them therefore acts as a cost driver. We discuss how the necessary competencies vary between three different Linked Data querying scenarios: data dumps, SPARQL endpoint and triple pattern fragments. Results of our analysis could serve as an input into the competency gap analysis.