QBST, Query-based Salient Terms, is a Google ranking signal that learns which words and word pairs should appear prominently on pages relevant to a given query. It is known from Eric Lehman's trial testimony as summarised in Judge Mehta's 2024 opinion, and a QBST feature appears in the 2024 leak; Google does not document it publicly. It is trained on about 13 months of user data. For sites, it means the terms that consistently characterise good answers to a query are a real relevance input.
What is QBST?
QBST, short for QBST, is a Google signal that identifies the words and pairs of words a relevant page for a query should contain prominently. Judge Mehta's 5 August 2024 opinion uses the example of "1600 Pennsylvania Avenue" and "White House": a page relevant to one should feature the other.
The opinion describes QBST as a "memorization system" that helps Google "understand facts about the world", trained on about 13 months of user data.
Query-based Salient Terms, or QBST, is a Google signal that helps respond to queries by identifying words and pairs of words that “should appear prominently on web pages that are relevant to that query.”
Judge Amit P. Mehta, quoting Eric Lehman, Memorandum Opinion, United States v. Google LLC (D.D.C., 5 August 2024), CourtListener
How does QBST work?
The record describes QBST as memorisation, not language modelling. Like Navboost, it learns from what users did with past results over a long window, then applies that knowledge to score documents for the same queries. Where RankBrain and DeepRank generalise to unseen queries, QBST remembers associations for queries it has seen.
The 2025 remedies opinion adds context: query interpretation data inside Glue includes "salient query terms", and RankEmbed's training data includes the salient terms Google derived from the query. Plaintiffs expressly did not ask Google to hand over "the salient terms for a given user query".
It is trained on about 13 months of user data.
Judge Amit P. Mehta, Memorandum Opinion, United States v. Google LLC (D.D.C., 5 August 2024), CourtListener
What evidence is there that Google uses QBST?
The label is Trial or leak evidence. The primary source is Lehman's testimony, cited in findings 95 and 102 of the 2024 opinion, which also records that newer large language model signals "did not replace Navboost and QBST in ranking". The leaked VideoContentSearchQbstTermsOverlapFeatures model shows QBST terms used as features in video search, measuring how much of the query's salient terms are covered by anchor text and by the top Navboost query.
The leak shows QBST as an input in one video context; it does not show how QBST is weighted in web ranking. The SparkToro write-up that broke the leak stresses the documentation does not "prove which elements are used in the ranking systems."
Fraction of salient terms of original query covered by anchor text.
Leaked Google API documentation, VideoContentSearchQbstTermsOverlapFeatures, google_api_content_warehouse v0.4.0, HexDocs
What does QBST mean for your site?
If Google memorises which terms good answers to a query contain, a page that omits the characteristic terms of its topic looks less relevant, however well written. The practical test is term coverage: the words and phrases that appear on most pages that satisfy a query, used prominently (title, headings, lead), not stuffed.
This is falsifiable. Compute the terms that appear on at least 7 of the top 10 results for a query and check your page. Pages missing several of them should, on average, rank worse for that query than pages that include them. The same logic applies to anchor text, since the leaked feature compares salient terms with anchors.
- Use the entity names and paired terms a knowledgeable answer would naturally include.
- Place core salient terms in the title, H1 and first paragraph.
- Do not repeat them mechanically; salience is about presence and prominence, not frequency.
Which Laurelin audit checks test for QBST?
QBST's learned term lists are not visible. The new check, page omits salient terms shared by top-ranking pages, uses the ranking set as a proxy for what the signal has memorised.
Related checks: thin content often explains missing terms; keyword stuffing is the over-correction to avoid; title does not match content intent and missing H1 affect prominence; vague headings hide salient terms from the page outline.
What are the key dates for QBST?
Frequently asked questions about QBST
What does QBST stand for?
Query-based Salient Terms, according to Judge Mehta's August 2024 opinion in United States v. Google.
Is QBST the same as TF-IDF?
No evidence says so. TF-IDF is a general weighting formula; QBST is described as a memorisation system trained on about 13 months of user data.
Does QBST mean I should add more keywords?
It suggests covering the terms that characterise good answers, placed prominently. Repetition beyond natural use is a separate failure.