reddit-customer-insights · diff
git:20260808.e011496 to git:20260812.3ae7d09
152 added, 9 removed. Audit A to A.
---
name: reddit-customer-insights
description: Discover evidence-backed customer language, pain points, objections, and buying triggers from relevant Reddit communities. Use when asked for Reddit research, voice-of-customer analysis, audience pain points, product feedback, or community-led content opportunities.
---
# Reddit Customer Insights
- Turn Reddit discussions into a concise, source-backed customer-insight brief. Use SandBase Reddit capabilities only for collection; the Agent synthesizes findings from the returned posts and comments. Read [the API map](references/sandbase-api-map.md) before selecting a capability.
+ Turn Reddit discussions into a concise, source-backed customer-insight brief. This Skill calls the named Reddit capabilities in [the SandBase API map](references/sandbase-api-map.md) through the SandBase MCP gateway. In a SandBase Agent, run the capabilities directly. In another compatible agent, require an authorized SandBase connection before starting; never request, print, or store an API key in the research output.
- ## Call SandBase capabilities
+ Read [example workflows](references/example-workflows.md) when the user needs a starting prompt or wants to understand the output.
- For every selected tool, call `sandbase_describe_tool` first and use only arguments in its current input schema. Then call `sandbase_call_tool` with the exact `tool_name`. The API map fixes the workflow's tool names; Capability Registry remains authoritative for parameters and available Reddit data.
+ ## Operating principles
+ - Research public discussions only; never post, vote, message, or alter Reddit accounts.
+ - Treat Reddit posts and comments as qualitative evidence of real customer language and sentiment, not statistically representative samples.
+ - Separate direct observations (what users said, how many said it) from interpretation (what it means for the business).
+ - Preserve provenance: keep post URLs, dates, scores, community names, and commenter context.
+ - Keep user company, product, and strategy details confidential in all outputs unless sharing is explicitly requested.
+ - Respect that Reddit users are real people. Do not mock, target, or re-identify individuals.
+
## Workflow
- 1. Define the product, audience, market, exclusions, and the decisions the research should inform.
- 2. Use `reddit_app_search_typeahead` to find relevant communities and queries.
- 3. Use `reddit_app_dynamic_search` and `reddit_app_topic_feed` to collect a diverse sample of recent discussions.
- 4. Use `reddit_app_post_details` or batch detail tools only for promising posts; preserve post URLs, dates, score, and community.
- 5. Separate direct observations from interpretation. Report recurring jobs, pain points, alternatives, objections, language, and unanswered questions.
+ ### 1. Frame the research question
+ Collect or infer: the product or market category, target audience segment, competitor landscape, research objective (pain points, buying triggers, objections, language, content ideas), and any exclusions (communities, time ranges, competitor names to avoid).
+
+ Classify the request as one or more of:
+ - **Pain point discovery**: what frustrates potential buyers
+ - **Buying trigger research**: what events or needs prompt purchase decisions
+ - **Objection mining**: what stops people from buying or adopting
+ - **Language extraction**: exact words and phrases customers use
+ - **Content opportunity finding**: unanswered questions worth creating content for
+ - **Product feedback**: what existing users praise or criticize
+
+ ### 2. Select and call SandBase capabilities
+
+ Read [the SandBase API map](references/sandbase-api-map.md) before selecting tools. Use the listed `tool_name` through the SandBase gateway:
+
+ 1. Call `sandbase_describe_tool` for the selected `tool_name` and read its current input schema.
+ 2. Call `sandbase_call_tool` with that exact `tool_name` and only schema-defined arguments.
+ 3. Keep the tool name, query parameters, and result metadata with the returned data.
+
+ ### 3. Discover relevant communities
+
+ Use `reddit_app_search_typeahead` to find:
+ - Subreddits where the target audience discusses the category
+ - Related search queries that reveal community terminology
+ - Adjacent communities (e.g., industry-specific, role-specific, tool-comparison subs)
+
+ Tips:
+ - Search for the product category, not the product name, to find broader discussions.
+ - Try role-based terms (e.g., "marketing manager", "DevOps engineer") alongside category terms.
+ - Note community sizes and activity levels to prioritize collection.
+
+ ### 4. Collect discussions
+
+ Use `reddit_app_dynamic_search` and `reddit_app_topic_feed` to build a diverse sample:
+
+ | Strategy | When to use |
+ |---|---|
+ | Problem-focused queries | "frustrated with [category]", "looking for alternative to [competitor]" |
+ | Decision-stage queries | "recommendation for [category]", "which [product type] should I use" |
+ | Experience queries | "switched from [competitor]", "after using [product] for 6 months" |
+ | Topic feed | When you need volume from a specific community over time |
+
+ Tips:
+ - Collect from multiple communities to avoid single-community bias.
+ - Include both recent posts (current sentiment) and older high-score posts (established patterns).
+ - Search for competitor names to find comparison and switching discussions.
+ - Use negative terms ("hate", "frustrated", "switched away") for pain points and positive terms ("love", "finally", "game changer") for satisfaction signals.
+
+ ### 5. Inspect selected posts
+
+ Use `reddit_app_post_details` only for the most promising discussions:
+ - Posts with high engagement (many comments, high score)
+ - Posts that contain decision-making language
+ - Posts where the title suggests detailed experience sharing
+
+ Read comment threads for:
+ - Recurring recommendations (what gets upvoted as answers)
+ - Disagreements (signals of market segmentation)
+ - Specific feature mentions and workflow descriptions
+ - Price sensitivity signals
+ - Integration or ecosystem requirements
+
+ ### 6. Synthesize findings
+
+ Organize findings by research theme, not by source post. For each theme:
+ - Count how many distinct posts/comments express the pattern
+ - Quote representative language (paraphrased, with source link)
+ - Note the community context (what role or situation the commenter is in)
+ - Distinguish one-time complaints from recurring patterns
+
## Output
- Return a source table, recurring themes with post counts, representative paraphrases, content or product implications, and evidence gaps. Do not post, vote, message, or alter Reddit accounts.
+ Return a structured insight brief:
+
+ ### Research scope
+
+ - Category and audience studied
+ - Communities and queries searched
+ - Time range and sample size
+ - Limitations and blind spots
+
+ ### Source table
+
+ | # | Community | Post title (abbreviated) | URL | Date | Score | Comments |
+ |---|---|---|---|---|---|---|
+ | 1 | ... | ... | ... | ... | ... | ... |
+
+ ### Recurring themes
+
+ | Theme | Post count | Representative language | Business implication |
+ |---|---|---|---|
+ | ... | ... | ... | ... |
+
+ ### Pain points (ranked by frequency)
+
+ 1. **[Pain point]** — observed in N posts across M communities. Example language: "..."
+ - Implication: ...
+
+ ### Buying triggers
+
+ Events or needs that prompt purchase decisions, with evidence.
+
+ ### Objections and barriers
+
+ What stops people from buying/adopting, with evidence.
+
+ ### Customer language
+
+ Exact phrases and framings customers use (for messaging, SEO, ad copy).
+
+ ### Content and product opportunities
+
+ Unanswered questions, underserved needs, and positioning gaps.
+
+ ### Evidence gaps
+
+ What this research cannot answer and what additional research would help.
+
+ ## Tips for better results
+
+ - **Cast a wide net first**: Start with 3–5 community searches and 5–10 diverse queries before narrowing.
+ - **Look for "switching" posts**: "I switched from X to Y because..." contains the richest insight.
+ - **Check "what do you use" threads**: These aggregate many data points in one post.
+ - **Note the non-obvious communities**: Industry-specific subs often have more candid discussions than product-category subs.
+ - **Time-bound for relevance**: Weight recent posts more heavily for fast-moving categories; include older posts for established patterns.
+ - **Triangulate**: If a pain point appears in only one community, it may be community-specific rather than market-wide.
+
+ ## Failure handling
+
+ - If SandBase is unavailable or unauthorized, report the failed capability and ask the user to connect or authorize SandBase; do not silently substitute a direct API.
+ - If searches return few results, try: broader category terms, different community targets, or adjacent problem spaces. Report if the topic genuinely lacks Reddit coverage.
+ - If a community is private or restricted, note the gap and search for alternative communities.
+ - If results are predominantly spam or self-promotion, filter them out and note the signal quality issue.
+
+ ## Quality gate
+
+ Before delivering, verify that:
+
+ - Every theme cites at least 2 distinct source posts (avoid single-post conclusions).
+ - Observations are separated from business interpretations.
+ - Community and time context is preserved for each finding.
+ - The sample includes multiple communities (avoid single-community bias).
+ - Sensitive user information (usernames in sensitive contexts) is handled respectfully.
+ - The research answers the original question, not just adjacent topics.