Precision measures how many retrieved search documents are relevant to the query. A low-precision system returns too much irrelevant information, making users work harder to find useful results. section 7.2.3
Use precision to assess search quality alongside complementary ranking measures:
- Compare retrieved documents with manually rated relevance judgments from a diverse set of real user queries. section 7.2.3
- Track precision as the search system changes, using improvements or regressions to guide optimization. section 7.2.3
- Read precision with recall, which measures how many relevant documents were found, and MRR, which measures how highly relevant results rank. section 7.2.3
Example: API integration case studies
For a query such as “API integration case studies,” precision indicates the percentage of returned results that contain relevant case studies. Low precision means the result set contains too much unrelated material. section 7.2.3
