Queries
Queries: how search engines interpret what you typed
A query is rarely used as typed. Between the search box and the index there is a chain of interpretation: parsing, correction, expansion, intent classification and rewriting. This guide explains each step and why the same words can produce very different results.
The words typed into a search box are the start of a process, not the input to a lookup. By the time the index is consulted, the query has usually been changed.
Parsing and normalisation
The raw string is tokenised and normalised the same way document text was during
indexing — case folded, punctuation handled, terms possibly stemmed. Operators
such as quotation marks for phrase matching, site: and - for exclusion are
recognised and extracted here rather than treated as content.
Spelling correction
Misspellings are extremely common, and engines correct them silently in most cases. The correction is informed by what other users searched for and clicked after searching the same misspelling, and by whether the corrected form has plausible results.
Synonym and concept expansion
“Cheap flights” and “budget airfare” express the same need in different words. An engine expands a query to include terms it has learned are equivalent in this context. Modern systems do this with learned representations rather than a fixed thesaurus, which is why the expansion is context-sensitive: “apple pie” and “apple stock” expand in completely different directions.
Intent classification
Queries are classified by what the user wants to do. The classic three-way split is still a useful mental model:
- Informational — the user wants to know something. “How does crawling work”.
- Navigational — the user wants a specific site. “Search Console login”.
- Transactional — the user wants to do or buy something. “Buy SSL certificate”.
Intent determines the shape of the results page as much as the ranking. A navigational query returns one dominant result; an informational one returns a list, often with a featured snippet or an AI summary above it.
Query rewriting and fan-out
For complex questions an engine may decompose the query into several sub-queries, run them separately and combine the results. Google has described this “fan-out” behaviour in the context of AI Mode, where a single question becomes multiple related searches whose results are then synthesised.
The practical implication for anyone writing content is that the query you are targeting is not necessarily the query being run. Pages that answer the underlying question completely tend to match more of the rewritten variations than pages optimised for one exact string.
Articles in this guide
The articles for this guide are being written. The guide above covers the whole topic in outline, and the article index lists everything published so far.