Hybrid retrieval in practice: what keyword search finds that embeddings miss, how to fuse two ranked lists, when a reranker earns its latency, and how to tell which signal failed.https://context-augmented-llm.com/https://context-augmented-llm.com/blogs/how-many-candidates-to-retrieve/https://context-augmented-llm.com/blogs/how-many-candidates-to-retrieve/Retrieval depth, fusion depth, rerank depth, final depth. Where each cut-off silently loses the right document, and how to find yours.Tue, 28 Jul 2026 00:00:00 GMTtradeoffsfusionhttps://context-augmented-llm.com/blogs/filtering-before-or-after-retrieval/https://context-augmented-llm.com/blogs/filtering-before-or-after-retrieval/A metadata predicate applied after the search returns an empty list; applied before, it changes what the search means. Neither is free.Mon, 27 Jul 2026 00:00:00 GMTtradeoffsdensehttps://context-augmented-llm.com/blogs/which-signal-failed/https://context-augmented-llm.com/blogs/which-signal-failed/A query returned nothing useful. A procedure for finding whether lexical missed, dense missed, fusion buried it, or the document was never indexed.Sun, 26 Jul 2026 00:00:00 GMTdiagnosisfusionhttps://context-augmented-llm.com/blogs/when-a-reranker-earns-its-latency/https://context-augmented-llm.com/blogs/when-a-reranker-earns-its-latency/A cross-encoder reads the query and passage together, which is why it ranks better and costs more. Where it belongs, what to feed it, and when to skip it.Sat, 25 Jul 2026 00:00:00 GMTrerankingtradeoffshttps://context-augmented-llm.com/blogs/fusing-two-ranked-lists/https://context-augmented-llm.com/blogs/fusing-two-ranked-lists/BM25 scores and cosine similarities aren't on the same scale. Reciprocal rank fusion, score normalisation, and what each method throws away.Fri, 24 Jul 2026 00:00:00 GMTfusiontradeoffshttps://context-augmented-llm.com/blogs/what-keyword-search-misses/https://context-augmented-llm.com/blogs/what-keyword-search-misses/Vocabulary mismatch: when the user's words and the document's words don't overlap, term scoring has nothing to score. Where stemming stops helping.Thu, 23 Jul 2026 00:00:00 GMTlexicaltradeoffshttps://context-augmented-llm.com/blogs/what-vector-search-misses/https://context-augmented-llm.com/blogs/what-vector-search-misses/Identifiers, rare terms, negation and exact phrases. Five query classes where embeddings return plausible neighbours instead of the document you asked for.Wed, 22 Jul 2026 00:00:00 GMTdensetradeoffshttps://context-augmented-llm.com/blogs/rewriting-the-query-before-you-search/https://context-augmented-llm.com/blogs/rewriting-the-query-before-you-search/The string a user types is rarely the best search key. What to strip, what to resolve, what to pull out as a filter, and what to leave alone.Tue, 21 Jul 2026 00:00:00 GMTquery-transformhttps://context-augmented-llm.com/blogs/query-expansion-and-how-it-drifts/https://context-augmented-llm.com/blogs/query-expansion-and-how-it-drifts/Adding terms to a query fixes vocabulary mismatch and dilutes precision at the same time. Where synonyms belong, and how expansion goes wrong.Mon, 20 Jul 2026 00:00:00 GMTquery-transformlexicalhttps://context-augmented-llm.com/blogs/searching-with-a-hypothetical-answer/https://context-augmented-llm.com/blogs/searching-with-a-hypothetical-answer/Generate a fake answer, embed that instead of the question, and search with it. Why the trick works, and the three ways it misfires.Sun, 19 Jul 2026 00:00:00 GMTquery-transformdensehttps://context-augmented-llm.com/blogs/one-question-three-searches/https://context-augmented-llm.com/blogs/one-question-three-searches/A compound question retrieves the average of its parts. How to split it, how to merge the result lists, and when splitting makes things worse.Sat, 18 Jul 2026 00:00:00 GMTquery-transformfusionhttps://context-augmented-llm.com/blogs/learned-sparse-retrieval/https://context-augmented-llm.com/blogs/learned-sparse-retrieval/A model that predicts term weights and adds terms the document never contained. What it inherits from lexical search, and what it costs.Fri, 17 Jul 2026 00:00:00 GMTlexicaltradeoffshttps://context-augmented-llm.com/blogs/weighting-one-signal-over-the-other/https://context-augmented-llm.com/blogs/weighting-one-signal-over-the-other/A fusion weight is a claim about your query mix, not your corpus. How to derive one, why a global weight is wrong, and when to skip it.Thu, 16 Jul 2026 00:00:00 GMTfusiontradeoffshttps://context-augmented-llm.com/blogs/why-a-similarity-threshold-doesnt-travel/https://context-augmented-llm.com/blogs/why-a-similarity-threshold-doesnt-travel/The cut-off that worked on your test queries fails on the next batch. What a cosine score is comparable to, and what to threshold instead.Wed, 15 Jul 2026 00:00:00 GMTdensediagnosishttps://context-augmented-llm.com/blogs/typos-and-which-retriever-forgives-them/https://context-augmented-llm.com/blogs/typos-and-which-retriever-forgives-them/A misspelling deletes a lexical match and blurs a dense one. Fuzzy matching, correction, and why the fix depends on which word was misspelled.Tue, 14 Jul 2026 00:00:00 GMTlexicaldensehttps://context-augmented-llm.com/blogs/when-five-results-say-the-same-thing/https://context-augmented-llm.com/blogs/when-five-results-say-the-same-thing/Retrieval returned five passages and one fact. Diversity at query time: per-document caps, marginal relevance, and what each one costs you.Mon, 13 Jul 2026 00:00:00 GMTtradeoffsdiagnosishttps://context-augmented-llm.com/blogs/which-field-the-match-landed-in/https://context-augmented-llm.com/blogs/which-field-the-match-landed-in/A title match and a body match are not worth the same. Field weighting, why concatenating fields loses information, and how boosts go wrong.Sun, 12 Jul 2026 00:00:00 GMTlexicalhttps://context-augmented-llm.com/blogs/the-analyser-decides-what-you-can-match/https://context-augmented-llm.com/blogs/the-analyser-decides-what-you-can-match/Tokenisation, stemming and stop words run before scoring. Get them wrong and lexical retrieval cannot find terms that are plainly in the text.Sat, 11 Jul 2026 00:00:00 GMTlexicalhttps://context-augmented-llm.com/blogs/a-reranker-cannot-find-what-retrieval-missed/https://context-augmented-llm.com/blogs/a-reranker-cannot-find-what-retrieval-missed/Reranking reorders a candidate set and never adds to it. Why a good reranker on a shallow pipeline is money spent on the wrong stage.Fri, 10 Jul 2026 00:00:00 GMTrerankingdiagnosishttps://context-augmented-llm.com/blogs/using-an-llm-as-a-reranker/https://context-augmented-llm.com/blogs/using-an-llm-as-a-reranker/Asking a general model to order candidates needs no training data and brings non-determinism, position bias and per-query cost. When that trade works.Thu, 09 Jul 2026 00:00:00 GMTrerankingtradeoffshttps://context-augmented-llm.com/blogs/when-query-and-document-differ-in-language/https://context-augmented-llm.com/blogs/when-query-and-document-differ-in-language/Term matching across languages scores zero. Multilingual embeddings, translating the query, and why a mixed-language corpus fails per language.Wed, 08 Jul 2026 00:00:00 GMTlexicaldensehttps://context-augmented-llm.com/blogs/routing-a-query-before-you-retrieve/https://context-augmented-llm.com/blogs/routing-a-query-before-you-retrieve/Classify the query, send it to the retriever equipped for it. Cheaper than fusion, and it fails completely when the classifier is wrong.Tue, 07 Jul 2026 00:00:00 GMTquery-transformtradeoffshttps://context-augmented-llm.com/blogs/making-a-term-mandatory/https://context-augmented-llm.com/blogs/making-a-term-mandatory/Term scoring is additive, so a document missing the decisive word can still win. How required terms work, and what they do to a hybrid pipeline.Mon, 06 Jul 2026 00:00:00 GMTlexicalfusionhttps://context-augmented-llm.com/blogs/boosting-recency-without-breaking-relevance/https://context-augmented-llm.com/blogs/boosting-recency-without-breaking-relevance/Superseded documents outrank current ones because relevance ignores dates. Where a time signal belongs in a ranking, and how it goes too far.Sun, 05 Jul 2026 00:00:00 GMTtradeoffsfusionhttps://context-augmented-llm.com/blogs/when-every-document-looks-the-same/https://context-augmented-llm.com/blogs/when-every-document-looks-the-same/On a homogeneous corpus, dense scores compress and ranking becomes arbitrary. How to spot a low-contrast corpus and what still discriminates.Sat, 04 Jul 2026 00:00:00 GMTdensediagnosis