Intent signal capture from unstructured surfaces

Description

The pattern that extracts intent from text, voice, and other unstructured customer interactions: search queries, support tickets, chat transcripts, product reviews. Increasingly important as customer behavior moves into conversational interfaces and away from clickstream events. Requires NLP or LLM processing to convert unstructured input into structured signals downstream recipes can use.

Common failure modes

• 01Sentiment analysis treated as intent, when sentiment and intent are correlated but not equivalent and the conflation produces misdirected campaigns
• 02Privacy implications of processing chat or call transcripts under-disclosed, especially when LLM processing moves the data outside the original collection context
• 03LLM-based extraction inconsistent across sessions because no canonical prompt, so the same logical input produces different structured signals depending on when it was processed
• 04Intent signals captured but not joined to identity, leaving them unusable for personalization and visible only in aggregate

Tools for this pattern

The capability a stack needs to run this pattern, and the vendors that provide it, on Martech Stack Builder.

• NeedsConversational AI / Virtual Agents
• RelatedNatural Language Processing

Recipes composing this pattern