Learn

Reranking: a second pass that reorders the results

Reranking is a second scoring pass over the passages a first retrieval step returned, reordering them by how relevant each one truly is to the query. The first search is fast and approximate; the reranker is slower and more careful, so it promotes the genuinely useful hits and demotes the near-misses.

At a glance

What it is
A second pass that reorders retrieved passages by relevance
Why bother
Nearest in vector space is not always most relevant
The trade
Slower and more accurate than the first retrieval step
What runs it
A reranker, a model that scores query-passage pairs
Flow

Two passes, fast then careful

A fast search returns many rough candidates, then a slower reranker scores each one against the query and reorders them. Green is the truly relevant passage moved to the top.

1
First retrieval (fast, approximate) returns many candidate passages, roughly ranked
2
Reranker scores each candidate judges true relevance to the actual question
3
Reordered shortlist the most relevant passages now sit on top

What does a reranker do?

The first step of a retrieval system is built for speed. It searches a large collection of embeddings and returns the passages whose vectors sit nearest the query, but nearest in that numeric space is a rough proxy for relevant. Some of what comes back is genuinely on point, and some of it is a near-miss that happened to land close.

Reranking is the cleanup pass. A reranker, a model that scores how well a passage answers a specific query, takes the candidates the first search returned and gives each one a fresh relevance score against the actual question. Then it reorders them. The truly useful passages rise to the top, the lucky near-misses fall. The first pass casts a wide, fast net; the reranker is the careful read of what the net caught.

When is it worth the extra step?

Reranking costs time. It runs a slower model over every candidate, per query, so it earns its place only when ordering matters and the first pass gets it wrong often enough to notice. If you retrieve many candidates and hand only the top few to a model with a limited context window, a reranker makes those few count.

It has a hard limit worth stating plainly: a reranker can only reorder what it was given. If the passage you needed never came back from the first search, because of bad chunking or a weak embedding model, no reranking will conjure it. It sharpens a good shortlist. It does not rescue a bad one.

Reranking helps when

  • The first search returns near-misses ranked too high
  • You retrieve many candidates and need the best few
  • Relevance is subtle and similarity alone gets it wrong

Reranking will not

  • Find a passage the first step never retrieved at all
  • Come free; it adds a slower scoring step per query
  • Fix bad chunks or weak embeddings further upstream

Related terms

← All terms Reviewed: June 2026