Module 4: Query Understanding

20 minutes

From Keywords to Intent Models

Google's query understanding system doesn't process queries as lists of words — it models the intent behind queries. This involves:

1. Query classification: What type of information is the user seeking?
2. Query reformulation: How might the user have phrased this differently?
3. Query context: What does the user's location, history, and device add to the intent?
4. Query disambiguation: When multiple intents are possible, which is most likely?

Query Reformulation

One of the most important patent concepts for content strategy: Google rewrites every query internally through multiple reformulation paths before matching to documents.

Reformulation types:

Specialization: Making the query more specific

• "coffee" → "arabica coffee beans" → "arabica coffee beans from Ethiopia"

Generalization: Making the query broader

• "broken wrist cast removal" → "cast removal" → "orthopedic care"

Paraphrase reformulation: Different words, same meaning

• "how to lose weight" → "weight loss strategies" → "reducing body fat"

Stemming and variants: Word form changes

• "optimize" → "optimizing" → "optimization" → "optimized"

The implication: Your content must align to the meaning behind queries, not just their surface form. A page optimized for "SEO optimization" but missing "search engine optimization," "website ranking," and "organic search" is under-optimized for the full reformulation space of that intent.

Query Classification in Practice

The Query Classification Audit implements the full intent decision tree. The core principle: Google's SERP format reveals the classification.

When Google shows:

• A local 3-pack → it classified the query as LOCAL
• A featured snippet → it classified as INFORMATIONAL with a direct answer
• Shopping results + product boxes → TRANSACTIONAL
• News carousel → TIME-SENSITIVE
• Brand Knowledge Panel → NAVIGATIONAL

Reading the SERP as a classification signal: Before writing content, look at what Google is actually showing for your target queries. The SERP format is Google telling you: "This is the type of content that satisfies this intent." Match that format.

Query Context Signals

Beyond the text of the query, Google uses:

Device context: Mobile queries have stronger local and immediate intent. "Plumber" on mobile + midday on a weekday = likely emergency. "Plumber" on desktop on Sunday = likely research.

Location context: Same query means different things in different locations. "Weather" in Miami vs. weather in London — different responses needed.

Session context: Queries within the same search session provide context for each other. If a user searched "how to bake bread" then "sourdough starter," the second query is in the context of bread baking.

User history: Personalized results adjust for what the user has previously engaged with.

SEO implication: You cannot optimize for context signals — but you should understand that the same page may rank differently for different users searching the same query. What you CAN control is matching the dominant intent interpretation for your target audience.