Sufficient consensus
Snippet generation selects and assembles the text shown under a result, usually query-dependent passages from the document. Meta descriptions are one input among several and are frequently replaced.
Length, truncation and highlighting follow the interface, not the publisher.
What the result pages leave out
These belong to the subject. People ask about them. They are missing from the agreed coverage.
- The passage you did not write for. The snippet may be assembled from a sentence buried in the middle of the page. That sentence is now your headline.
- Satisfaction without a click. A good snippet ends the search. The publisher supplied the answer and receives no visit.
- Loss of context. An extracted sentence carries none of the qualification that surrounded it. Careful writing gets flattened.
- Different snippets for different queries. One page appears in a dozen shapes. Almost no one checks more than one of them.
- The sentence that should be extractable. Writing one clean, self-contained sentence per section is the practical consequence and it appears in almost no brief.
What people actually want to know
- Which sentence of mine is currently representing my page, and did I write it to be read alone?
- What does my page look like for the five queries it actually ranks for?
- If the searcher is satisfied by the snippet, what did I gain?
- Does any single paragraph of mine survive being quoted out of context?
Every section needs one sentence that is true on its own. That sentence is the page that most people will see.
More in information retrieval
Learning to rankThe system is trained on the behaviour it caused.Query expansionYou are answering a question that was quietly edited before it reached you.Relevance feedbackAbsence produces no signal.Query performance predictionA page of ten results looks the same whether the answer exists or not.Passage retrievalYour page is being read in fragments by a system that never sees the whole.Mean average precisionA score of 0.87 tells you how well the ranking matched a list somebody wrote down.Vector space modelIf the web never described your case, the vector for it is somebody else's.Inverted indexThe web is the part of knowledge that somebody bothered to publish and a crawler managed to reach.