θ-MDA is a flexible and efficient operator for complex ad-hoc multi-dimensional aggregation queries. It separates the specification of aggregation groups for which aggregate values are computed (base table b) and the specification of aggregation tuples from which aggregate values are computed. Aggregation tuples are subsets of the detail table r and are defined by a general θ-condition. The θ-MDA requires one scan of r, during which the aggregates are incrementally updated. In this paper, we propose a two-step evaluation strategy for θ-MDA to optimize the computation of ad-hoc range aggregates by reducing them to point aggregates. The first step scans r and computes point aggregates as a partial intermediate result x̃, which can be done efficiently. The second step combines the point aggregates to the final aggregates. This transformation significantly reduces the number of incremental updates to aggregates and reduces the runtime from (∣∣r∣∣⋅∣∣b∣∣) to (∣∣r∣∣) , provided that ∣∣b∣∣<∣∣r∣∣‾‾‾√ and |x̃| ≈ |b|, which is common for OLAP. An empirical evaluation confirms the analytical results and shows the effectiveness of our optimization: range queries are evaluated with almost the same efficiency as point queries.