RAG Retriever Evaluation — Cheat Sheet

Quick-reference · contextual recall · contextual precision · golden data · LLM-as-a-judge

1-PAGE REFERENCE

1Retriever: the job

Retriever: given a query, fetch relevant context from the vector database for the generator to use.

A strong RAG answer begins with high-quality retrieval.

2How retrieval works

QueryEmbeddingVector searchTop-k chunksContext

3Build before evaluating

Load documentsChunkEmbedIndexRetrieve

Evaluate each component as you build—not only at the end.

4Failure mode 1: missing context

Information needed to answer the question exists in the knowledge base but was not retrieved.

Measure with recall.

5Failure mode 2: noisy context

Too many irrelevant chunks are retrieved, distracting the generator and using context-window space.

Measure with precision.

6The ideal retriever

High recallFind all facts required for the answer.
High precisionReturn mostly useful chunks, early.

The target is high performance on both metrics.

7Recall–precision trade-off

Increasing k returns more chunks. This can improve recall but often lowers precision because extra chunks may be noise.

Example: k = 5 → 3 relevant; k = 10 may find the missing 2 but add 5 irrelevant chunks.

8Reference-based evaluation

Recall and precision need a reference for what the correct information should be.

Reference: a golden dataset with user question + ideal answer.

9Why not label chunks?

Exact relevant chunk IDs change if you alter chunk size or chunking strategy.

Better reference: an ideal answer remains valid even when chunks are reorganized.

10Golden dataset row

realistic questionideal answerexpected facts / claims

Use student-like questions for a student doubt-solving RAG app.

11Contextual recall

Supported ideal-answer claims ÷ total ideal-answer claims

Did the retrieved context contain all information needed for the ideal answer?

12Contextual recall workflow

Ideal answerBreak into claimsCheck retrieved contextCoverage score

13Recall example

Ideal answer has 3 atomic claims. Retrieved chunks support all 3.

Contextual recall = 3 / 3 = 1.0

If only one of two claims is found: 1 / 2 = 0.5.

14Contextual precision

Rank-aware relevance of the retrieved chunks

Are useful chunks near the top of the context, with minimal noise?

15Precision intuition

  • Relevant chunk at rank 1 → good
  • Relevant chunk buried after noise → worse
  • Many unrelated chunks → lower precision
  • Top-ranked context should help answer the query

16Contextual vs Recall@K

Recall@KContextual recall
ReferenceRelevant chunk IDsIdeal answer claims
Chunking changeCan break labelsStill valid
JudgeExact matchLLM-as-a-judge

17LLM as judge

The judge decomposes the ideal answer into claims, then determines whether each claim is supported in the retrieved context.

Benefit: semantic evaluation without fragile exact chunk labels.

18Golden data sources

  • Manual authoring by a domain expert
  • LLM-assisted generation + human review
  • Synthetic-data tooling
  • Successful production interactions
Warning: synthetic questions can be unnatural or off-topic.

19How to improve recall

  • Increase k carefully
  • Improve chunking strategy / chunk size
  • Use stronger embedding models
  • Improve document coverage
  • Add metadata filtering where appropriate

20How to improve precision

  • Lower k when safe
  • Improve metadata filters
  • Use reranking
  • Improve chunk boundaries
  • Remove noisy or duplicate source content

21Evaluation loop

Golden questionRetrieve top-kLLM judgeRecall + precisionInspect failures

22What to inspect

Do not rely only on average scores. Read low-scoring questions and discover whether the problem is embedding, chunking, top-k, filtering, source data, or relevance ranking.

23Mental model

A retriever is good when it brings all the facts needed to answer the question and avoids irrelevant noise.

Rule: high contextual recall + high contextual precision = useful RAG context.