Intent is not stored in words. It is stored in situations. And situations are not visible from vocabulary alone.
Keyword detection treats language as a signal in itself — the presence of certain words triggers a score or a classification, their absence produces a different outcome. The result is a system that responds to surface features rather than underlying meaning. Sellers do not write enquiries to communicate their intent. They write to ask a question or start a conversation. Their intent is present in what they write, but it is rarely explicit and almost never captured in a single phrase.
Same words, different situations
Consider the word “valuation.” It appears in almost every seller enquiry in some form. A keyword system treats it as a constant. In practice, it means almost nothing on its own.
A seller who writes “I’d like a valuation as we’re looking to sell before Christmas” is in a fundamentally different position from one who writes “I was just wondering if you could do a valuation, no rush at all.”
Both messages contain the word valuation. A keyword system scores them identically on that dimension. A human reading them would not treat them the same way, because a human understands that the word is not the signal. The situation is the signal.
The same applies to urgency markers. “As soon as possible” from a seller who has explained they are relocating for work in six weeks is a strong indicator of instruction likelihood. The same phrase from a seller who has given no context and no reason carries almost no weight. The words are identical. The situations are not.
What context actually does
Context does not modify the meaning of words. It determines whether the words mean anything useful at all.
When a seller mentions that they previously had an agent round, that is not a keyword. It is a piece of situational information. It tells you that this seller has been through the early stages of the selling process before, which means they are not exploring the idea of selling for the first time. That prior engagement is a signal of a different kind entirely, one that no vocabulary list will surface.
When a seller mentions a constraint that has recently changed, a job offer accepted, children settled in a new school, a family situation resolved, the significance of that information is not in the words used to describe it. It is in what the change represents: a barrier to instruction that no longer exists.
Context converts information into evidence. Without it, the same information is just text.
Why combinations matter
Individual signals are weak. Combinations of signals are diagnostic.
A seller who mentions a specific timeline is one data point. A seller who mentions a specific timeline, references a prior valuation, and uses language that suggests they are comparing agencies rather than exploring the idea of selling, is a different picture entirely. No single element of that enquiry is conclusive. Together, they point clearly in one direction.
This is why keyword systems fail at the task of intent classification. They treat each signal in isolation. A word scores points or it does not. The total determines the outcome. But intent does not accumulate additively. A seller with three weak signals is not necessarily more motivated than one with a single strong signal embedded in the right context.
What matters is how the signals relate to each other. A timeline combined with a stated reason carries more weight than a timeline alone. A prior valuation combined with a resolved constraint is more significant than either piece of information in isolation. The combination creates meaning that neither element contains on its own.
The relationships between signals
Intent classification requires understanding not just what signals are present, but how they interact.
Some signals reinforce each other. A seller with an active reason to move, a defined timeframe, and evidence of prior market engagement is exhibiting a pattern of signals that compound. Each one makes the others more significant.
Some signals contradict each other. A seller who expresses urgency but then lists multiple conditions that would need to be met before they could proceed is not exhibiting high intent. The urgency language and the conditional language sit in tension. Reading either in isolation produces the wrong conclusion.
Some signals are neutralised by their context. “We need to sell” from a seller who subsequently explains they are not in a position to proceed for at least twelve months is not a high-intent signal. The phrase sounds decisive. The situation it describes is not.
Understanding these relationships requires reading the enquiry as a whole, not as a collection of scored terms. The meaning is in the structure of what has been written, not in its vocabulary.
Situations over vocabulary
A seller can exhibit strong instruction likelihood without using a single word a keyword system would flag. A seller can use every urgency marker available while describing a situation with almost no near-term instruction probability. The words are not the signal. What the seller is dealing with is the signal.
Vocabulary is what people say. Situation is what they mean. That gap is where keyword classification fails and where genuine intent reading begins.