> ## Documentation Index > Fetch the complete documentation index at: https://docs.peakmetrics.com/llms.txt > Use this file to discover all available pages before exploring further. # Smart Cats Workflow > Smart Categories allow you to automatically classify mentions based on custom prompts. Below are four advanced prompt examples tailored to common use cases. These can be deployed directly or adapted to suit your organization's unique goals. ## Need Help Crafting Your Prompts? The **PeakMetrics Insights Team** is here to help you build high-impact, scalable Smart Category prompts tailored to your goals. We've partnered with dozens of teams to launch workflows that power **brand perception tracking, risk detection, campaign analysis, and more**. That said, **ChatGPT is also an excellent companion** in this workflow. It’s great for **brainstorming prompt language, testing edge cases, and exploring new use cases**. We often recommend starting there to quickly iterate. Here’s a starter prompt you can use with ChatGPT: > *“I’m building a media classification system. Help me create a clear, effective prompt to categorize mentions about \[topic] with response options like \[examples]. The prompt should be scalable and work across thousands of mentions.”* Once you’ve drafted a few ideas, bring them to us—we’ll help validate, refine, and scale them together. * To jumpstart your thinking, we’ve also included a **handful of common use cases and example prompts** below—ready to deploy or adapt: *** ### 1. Favorability Toward a Brand (Smarter Sentiment) **Use Case:** Move beyond generic sentiment scoring to capture nuanced favorability based on tone, framing, and intent. **Prompt:** “Analyze this content and determine how favorably the brand \[Acme Corp] is portrayed. Consider not just explicit language, but also the framing, underlying tone, and broader context (e.g., positioning in a list, comparison to competitors, or the presence of loaded terms).” **Response Options:** * Strongly Favorable * Somewhat Favorable * Neutral * Somewhat Unfavorable * Strongly Unfavorable * Undefined **Best For:** Brand marketing, PR, comms, executive reporting *** ### 2. Product Adoption Trend Classification **Use Case:** Detect directional shifts in how a specific topic is being adopted (e.g., “AI" or "food dyes”). **Prompt:** “Based on the content, does this mention indicate a trend of increasing, decreasing, or unchanged adoption of \[Topic X]? Consider product launches, executive statements, hiring trends, or regulatory reactions.” **Response Options:** * Increasing \[Topic X] Adoption * Decreasing \[Topic X] Adoption * No Indication Available * Not About \[Topic X] Adoption **Best For:** Strategic planning, market research, analytics teams *** ### 3. Threat Detection for Risk Teams **Use Case:** Flag threats across various categories based on explicit signals. **Prompt:** “Identify the primary type of threat—if any—present in this mention toward \[Entity]. Consider direct language and implications relating to physical harm, cyber incidents, reputational damage, or misinformation. Do not infer unless the content provides clear indicators.” **Response Options:** * Physical Threat * Cybersecurity Threat * Reputational Threat * Other / Misinformation * No Threat Detected **Best For:** Physical security, cybersecurity, threat intel, crisis response *** ### 4. Ideological Lean Classification **Use Case:** Analyze ideological lean for bias, stakeholder alignment, or balance. **Prompt:** “Determine the ideological leaning expressed or implied in this mention, considering tone, policy alignment, terminology, and known stance of the source outlet. Focus on framing, not just keywords (e.g., ‘climate change hoax’ vs. ‘climate emergency’).” **Response Options:** * Left-Leaning * Center / Neutral * Right-Leaning * Mixed / Unclear **Best For:** Public affairs, advocacy, ideological analysis *** ### 5. Mention Type Categorization (e.g., Reviews, Complaints, Giveaways) **Use Case:** Segment types of mentions for better content understanding and workflow routing. **Prompt:** “Classify the type of mention this is related to \[Brand]. Consider the content and tone to determine whether it’s a product review, a giveaway/promotion, a complaint, a general question, or media coverage. Only choose one category that best fits.” **Response Options:** * Product Review * Giveaway or Promotion * Customer Complaint * General Inquiry or Question * Press/Media Coverage * Other **Best For:** Community management, PR, campaign tracking, CX triage *** **Pro tip:** You can add example posts in your prompt to give the Smart Category more context and yield a more accurate result. *** ### **What Happens Next?** Once you've completed the steps above, give the system 5–15 minutes to populate your new Smart Category within your workspace. After that: * Navigate to your **Workspace Dashboard** * Scroll down to the **Charts** section * You’ll now see your new Smart Category included From there, you can: * **Filter by your Smart Category** to generate an AI-powered summary of the narrative * Or **click directly into the chart** to explore the mentions, key amplifiers, sentiment breakdowns, and all the surrounding context you need to take action This gives you a real-time view of the conversation—without the noise. *** ### 👥 Ready to Go Deeper? The **PeakMetrics Insights Team** is world-class at crafting Smart Categories that are **usable, scalable, and grounded in your goals**. → [**Schedule a custom working session with us**](https://meetings.hubspot.com/katie-kleinknecht) to co-design your categories, test prompts, and operationalize insights. Want to integrate your Smart Category outputs into Threat Scoring or Reporting Pipelines? Contact your Account Representative or email [support@peakmetrics.com](mailto:support@peakmetrics.com) to learn how.