Pinecone diagnosis: hybrid dense plus sparse search underperforms

TL;DR: Symptom: hybrid queries return worse results than dense-only. Check that sparse values are present on records, the alpha balance is sane, and you picked the right combination pattern. Confirm it 1. Fetch a few records and confirm sparse_values are present and nonzero.

The fix

1. Fetch a few records and confirm sparse_values are present and nonzero.
2. Run dense-only and sparse-only queries separately; if sparse-only returns junk, the sparse vectors (or the sparse index config) are the problem.
3. Check how you combine: server-side combination needs both vectors on each record in one index; client-side fusion needs two queries plus RRF.
4. Missing sparse values: re-ingest with sparse vectors (e.g. from a sparse embedding model like pinecone-sparse-english-v0 or BM25 weights).
5. Balance: scale query vectors to adjust the dense/sparse weight (alpha) and evaluate; do not leave it at a default nobody chose.
6. Wrong pattern: pick one documented pattern (filter-then-rank, client-side RRF fusion, or server-side combination) and implement exactly that.

When to use this

• This covers exactly what the title says: Pinecone diagnosis.
• You are setting this up for the first time, or auditing an existing setup.
• You want the key gotchas in one place before you start.

When not to use this

• You are doing a different workflow with Pinecone; these steps are specific to the title above.
• You need the full reference docs; this is the short path, not the manual.

Compatibility

• Not pinned to a specific version; follows current Pinecone behavior.

Pinecone's hybrid-search guide defines the three combination patterns and the alpha weighting for server-side dense/sparse combination, plus the sparse embedding model option. Verified against live docs 2026-09-26.