Consumer behaviour signal

What it is

Consumer behaviour signals are observable patterns in how people phrase queries, rephrase them after seeing results, dwell on or abandon answers, and choose which source to act on. Ranking systems and generative models treat these patterns as evidence of intent and answer quality.

Why it matters

In AI search the model has no fixed ranking to lean on, so aggregated behaviour becomes a primary proxy for whether an answer satisfied the user, shaping which content gets surfaced and cited next time.

How it works

Signals are captured across the session: the sequence of queries, follow-up prompts, clicks, copy actions, time before refinement, and whether the user stops searching, then aggregated to infer satisfaction at scale.

When it applies

It applies wherever a system observes interaction data, from a search results page to a multi-turn chat with a generative assistant.

Examples

• A user rephrases broad and narrow versions of the same question, revealing the system underdelivered on the first attempt.
• A searcher copies a snippet from one answer and stops searching, signalling that answer resolved the task.
• Rapid back-and-forth between two results before settling suggests low confidence in either source.

How it is measured

• Query reformulation rate within a session
• Answer abandonment or follow-up rate
• Dwell time and copy or save actions on a result
• Task completion or search exit after a given answer