A financial-services assistant may answer a mortgage-rate question correctly yet fail on a policy question. The policy might be absent, buried beneath outdated documents, removed from the supplied context, or ignored in the answer. Each is a different failure mode: a specific way the system fails to meet its goal. Naming the mode turns a vague “accuracy problem” into something that can be tested and fixed. section 1.4
RAG failures can occur during retrieval, context augmentation, or generation. Seven documented failure points distinguish missing content, poor ranking, factual contradiction, relevant information excluded from context, information present but unused, incorrect specificity, and incomplete answers. The distinction matters: improving ranking will not make a generator use evidence it already has. section 1.4 section 15.5.1
Table 1.1 RAG failure points
|
Failure point
|
Name
|
Description
|
Impact on financial services example
|
|
FP1
|
Missing content
|
Answer doesn’t exist in the knowledge base
|
Policy questions about procedures not yet documented
|
|
FP2
|
Missed the top rank
|
Correct document retrieved, but ranked too low
|
Current guidelines are buried beneath outdated
policies
|
|
FP3
|
Factually inconsistent hallucination
|
Model contradicts the retrieved information
|
Fabricated compliance procedures despite the correct
source documents
|
|
FP4
|
Not in context
|
Relevant chunk retrieved but filtered out
|
Important policy details were removed during context
optimization
|
|
FP5
|
Not extracted
|
Information in context, but not used in response
|
Model ignores the retrieved compliance requirements
|
|
FP6
|
Incorrect specificity
|
Wrong level of detail for the user’s needs
|
Strategic overview of when operational steps are
needed
|
|
FP7
|
Incomplete
|
Partial answer when a complete response is needed
|
Missing critical steps in multi-part procedures
|
Compare the named failure points with their effects on financial-services questions before choosing a fix. section 1.4
Diagnose before adding components
If the current guideline is retrieved but ranked beneath old policies, investigate ranking (FP2); RAG-Fusion is one possible remedy. If correct compliance documents are retrieved but the answer invents a procedure (FP3), a retrieval audit such as CRAG targets a different problem. First establish an evaluation baseline, then add a component only when a measured failure warrants its latency and cost. section 1.4 section 1.0 section 1.1 section 15.5.1
Some failures require more than a better single retrieval pass. Multi-hop questions need facts chained across steps; temporal questions need facts aligned in time. Errors also occur across chunking, retrieval, reranking, and generation, so an error breakdown from another system is not a substitute for measuring your own. section 4.5.5
Table 15.1 RAG techniques by complexity and performance impact
|
Technique
|
Complexity
|
Primary failure points addressed
|
When to use
|
|
Naive RAG (ch. 4)
|
Low
|
Baseline
|
Simple factual queries over clean corpora
|
|
Hybrid search (ch. 2, 8)
|
Low
|
FP1, FP2
|
Queries mixing keywords and concepts
|
|
Reranking (ch. 8)
|
Low-Medium
|
FP2, FP4
|
Retrieval noise degrades generation
|
|
HyDE (ch. 7)
|
Medium
|
FP1, FP2
|
Abstract or underspecified queries
|
|
RAG-Fusion (ch. 8)
|
Medium
|
FP1, FP2, FP7
|
Diverse query interpretations needed
|
|
RAPTOR (ch. 10)
|
Medium
|
FP5, FP6
|
Large documents; queries at varying abstraction levels
|
|
Context compression (ch. 13)
|
Medium
|
FP4, FP5
|
Long or noisy retrieved contexts
|
|
Self-RAG (ch. 9)
|
High
|
FP3, FP5, FP7
|
Hallucination-sensitive applications
|
|
FLARE (not implemented; see the ch. 11 planner)
|
High
|
FP1, FP7
|
Long-form answers; information needs emerge mid-generation
|
|
CRAG (ch. 11)
|
High
|
FP1, FP3
|
Retrieval quality varies by query
|
|
LevelRAG (ch. 11)
|
High
|
FP1, FP7
|
Multi-part queries needing planned decomposition
|
|
Graph RAG (ch. 12)
|
High
|
FP1, FP7
|
Multi-hop questions over connected entities
|
|
Adaptive-RAG (ch. 15)
|
High
|
All
|
Mixed query complexity
|
|
Agentic RAG (ch. 15)
|
Very High
|
All
|
Complex multi-step queries, fragmented knowledge
|
Use the technique-to-failure-point comparison to weigh a proposed fix against implementation and runtime complexity. section 15.5.1





