Learning how to improve effort estimation in small software development companies

Small software development companies have special effort estimation requirements. In particular, small companies often find that they must tender for contracts for which they have no expertise in the project problem or solution domains. In addition, they have little spare capacity for project management and improvement. They have therefore no formal way of learning how to improve future effort estimates. A discussion of why small software company effort estimation needs are likely to be different to larger companies is provided; and a proposal for specific support for effort estimation in smaller companies is given. The support takes the form of a simple Bayesian statistical modelling approach. The approach can be used by estimators to learn how to improve the consistency of their estimation in existing and in novel problem and solution domains