Social-media monitoring tells you how many people mentioned a subject. Narrative intelligence tells you what story is being told about it, who is telling it, how fast it is moving and what you should do. This guide explains the difference and sets out what a team should expect from a platform that claims to provide narrative intelligence.
Mentions versus narratives
A mention is a single post. A narrative is a story that many posts tell together: a claim, a frame, a villain, a demand. Ten thousand mentions can carry three narratives, and the smallest of the three may be the one that matters. Counting mentions hides this. Narrative intelligence begins by reading the posts, in their own language, and grouping them by what they say rather than by which words they contain.
The four questions
A useful narrative-intelligence read answers four questions. What is being said: the specific claims and frames, not a sentiment score. Who is saying it: the public accounts, outlets and communities driving each narrative, and whether the push is organic or coordinated. How fast and where: the velocity, the platforms it has crossed and the regions and languages it has reached. What should we do: a recommendation, or at least the options, with the evidence attached.
Why sampling fails
Many platforms sample a fraction of the public conversation to keep costs down. Sampling is fine for measuring a brand's overall share of voice and disastrous for early warning, because the early signal of a problem is by definition small. A thread of forty posts in one district is invisible in a one-percent sample of a national feed. On the subjects you track, insist on complete collection.
Why language depth matters
Narratives form in the language of the community they concern. An English-only view of a multilingual public sees the narrative only after it has crossed over, which is usually after it has set. Language depth means more than translation: it means reading tone, idiom and mixed-script posts as a native reader would, and including the regional outlets and feeds that carry the local conversation.
Actors and coordination
Knowing who is driving a narrative changes how you respond. Genuine public concern deserves a substantive answer. A coordinated push by a small group of accounts deserves a different response, and the evidence of coordination may itself be the story. An actor view should show the amplification path from first post to peak and the public accounts and outlets that carried it, without profiling private individuals.
From read to action
Narrative intelligence is only useful if it ends in a decision. The output should be a written read that a leader can absorb in three minutes: what changed, why it matters, what the options are. When a response is needed, the platform should help draft it from your approved facts and route it to a named approver. It should never post on your behalf.
Evidence and provenance
Every claim in a read should point to the public posts it was built from. When a narrative becomes a legal, regulatory or board matter, the evidence must be preserved as captured, with timestamps, sources and hashes, so that any later change is detectable. A read without provenance is an opinion.
What to ask a vendor
- Do you sample, or do you collect in full on the subjects I track?
- Which languages do you read natively, and can you show me a mixed-script example?
- Can you show me who drove a narrative and whether it was coordinated?
- Is the output a written read or a dashboard I must interpret?
- Can the platform ever post on my behalf without a named approver? (The answer should be no.)
- What are your responsible-use limits, and are they in the contract?
Narrative intelligence done well gives a team lead time, clarity and a defensible record. Done badly, it is a louder version of the dashboard you already ignore. The questions above tell you which you are looking at.