A brief is the instruction you give STORY: what to analyse, over which period, for whom and in what shape. A good brief gets you a verdict-first answer you can forward as it is. This guide covers how STORY turns a brief into a report, what a strong brief contains, and six ready-made briefs you can adapt.
How STORY works through a brief
Ask. You write your brief. Use the composer filters to pick your search, date range and reply language, or describe them in the brief.
Define. STORY confirms the search and clarifies anything missing, such as the timeframe or comparison set. No data is pulled yet. When the scope is clear, a Go to plan button appears.
Plan. STORY sets out the queries and outputs it will produce. Review it, switch tasks on or off and confirm it. Your goal, searches and timeframe are then carried into every turn of the chat.
Execute. STORY runs the confirmed plan and delivers the report.
If STORY skips straight to a plan you didn't expect, or misreads the question, reply with what you meant before confirming. Changing course is cheap at the Define and Plan stages.
Three habits that make every brief better
Name the search. Use the search's name as it's saved in Pulsar, or select it in the composer. STORY only sees the searches you can see.
Give a timeframe. "Last week", "Q3" or a date range. Every claim STORY makes is compared against a baseline, so a clear window gives you a clear comparison.
Ask why, not what. "Why did sentiment drop on Wednesday?" gets you causes tested against the evidence. "Show me sentiment" gets you a chart.
What a strong brief contains
Element | Example |
Goal | "Brief the comms team on what changed in UK supermarket conversation and why." |
Search and filters | Search "UK Supermarkets", UK only, online news plus social channels |
Timeframe | "The 24 hours ending 07:30 London time" |
Audience and tone | "Plain English for press officers and senior stakeholders. No jargon." |
Structure | "Three sections: top stories, stories in detail, emerging risks and opportunities." |
Length | "No more than 700 words." |
Thin-day rule | "If fewer than three strong stories qualify, include only those and say so. Don't pad." |
What to leave out | "Exclude adverts, non-UK stories and syndicated duplicates." |
Put filters such as country and source in the composer or the plan. STORY applies them at query level, which is more reliable than asking it to filter results afterwards.
Briefs you'll schedule
A scheduled run doesn't see your chat history, only the confirmed plan and its instruction. So:
Write everything the brief needs into the brief itself. Don't rely on "as discussed above".
Leave calendar logic to the schedule. Set the window when you schedule it, not with rules like "if it's Monday, include the weekend".
Add recipients when you create the schedule, not in the brief.
See STORY: Scheduling recurring briefings for the full set-up steps.
Six briefs to start from
Replace anything in [brackets] before you run them. Default lookbacks are shown and can be changed.
1. Crisis signal detection
"Scan [BRAND] conversation over the last 72 hours in [MARKETS]. Cluster negative and ambiguous posts into narratives, measure their velocity against the 30-day baseline, trace the seed post and early amplifiers, and give me a risk verdict (Low, Watch, Elevated or Act now), the 48-hour trajectory and the intervention window."
2. Weekly brand health check-in
"Compare the last 7 days with the previous 7 for [BRAND] in [MARKETS]. Give me a scorecard of volume, sentiment and share of voice against [COMPETITORS], the three stories driving the movement, and one line on what to watch next week."
3. Competitive narrative gap
"For [BRAND] against [COMPETITORS] over the last 30 days in [CATEGORY], show share of voice, the narrative territories each competitor owns, and the gaps where [BRAND] is absent but the conversation matters."
4. Audience deep-dive
"Profile who drives the [TOPIC] conversation over the last 90 days. Cluster authors by behaviour, interests and sentiment, and give me segment cards with size, stance and two or three representative voices. Tell me which segments matter most for [OBJECTIVE]."
5. Trend and inflection detection
"Find emerging themes in [CATEGORY] over the last 30 days. Separate real momentum from noise, trace the moment each theme tipped, and rank them act, watch or ignore for [BRAND]."
6. Campaign debrief
"Debrief [CAMPAIGN], live [START] to [END], against the 7 days before launch. Score reach, resonance and earned amplification, show the peak moments and what caused them, and give me three to five takeaways we can reuse."
Reading and checking the output
Verdict first. Reports open with a one-line verdict, a strength score from 0 to 100 and the change against the previous period, followed by the evidence and up to four charts.
Every quote can be checked. Quotes are the exact text of posts in your results, credited to the account that posted them, including replies, quote posts and reposts.
Read the caveats. STORY flags thin samples, coverage gaps and data issues. Treat them as part of the answer.