A fundamental gap persists in Agile software implementation. At a micro level, we are awash in data, inundated with stories and story points from a life cycle management tool like Jira, Redmine, or VersionOne. At a macro level, we struggle to adequately define functional requirements sufficient to support consistent sizing via function points (FP). Even if we do manage to functionally size planned future work, we often have not accrued a historical database of actual effort and cost tied directly to epics and features – the very objects we need for an apples-to-apples comparison with our program baseline. The #NoEstimates advocates throw up their hands and say that a macro-level planning estimate – five years' worth of annual budgets, for example – is futile. However, whenever we are spending "other people's money," especially the American taxpayer's, we are obliged to apply best practices in quantifying that longer-term commitment up front.

Building on previous research, this paper presents a framework for macro agile estimation based on fully analogized sizing scales that enable the application of expert judgment to produce an accurate characterization of early-stage uncertainty. It also provides a blueprint for building a database of analogies to populate such scales and presents empirical results from applying them.