01
Where this one gets misread
Reranking puts a second model in the path, with its own cost, its own latency and its own failure mode. It can also reorder good results into worse ones, which is only visible if somebody compared.
Glossary
What is reranking?
Reranking is a second pass that reorders an initial set of candidate results using an extra model, rule or scoring method before any of them are shown or used.
Updated 21 Aug 2026
02
Questions to ask
Ask for a worked example on your own material, and the evidence needed to reproduce it.
- What model does the reranking?
- What did it improve, measured how?
- What does it add to latency?
- Can it be switched off and compared directly?
03
How Marella uses the term
We use “Reranking” only where a product mechanism or an evaluation method backs it up, and we say when the behaviour depends on how a deployment is configured.
- Backed by a product mechanism or an evaluation method
- Deployment differences flagged
What this page does not prove
- B1A definition is not a claim about how the product performs.
- B2Vendor implementations vary.
- B3Test the term against a representative workflow.
Related pages
Test the claim on your documents
Pick a real piece of work, agree what a good answer looks like, then go through the results together.