How to Analyze Reports

VOC Insight MCP doesn’t just help you collect and analyze consumer data — more importantly, it helps you turn that data into business decisions. Once collection is complete, you can use the results in two complementary ways:

• Read the brand social analysis report to quickly understand what happened in the market.
• Use AI deep follow-up analysis to keep asking questions around the report, understand the causes, and plan your actions.

​

Type 1: Brand Social Analysis Report

​

When to use

When you want to quickly understand your brand, competitors, and the overall market.

Typical scenarios include:

• Regularly monitoring overall brand performance
• Tracking consumer discussion and sentiment shifts
• Comparing your brand with competitors
• Reporting analysis results
• Providing a data foundation for later strategy

The brand social analysis report helps you quickly answer:

• What have consumers been discussing lately?
• Which platforms perform best?
• Where are the gaps between your brand and competitors?
• Which product issues do users care about most?
• What new trends in the current market are worth watching?

It helps you answer: what happened in the market (What Happened).

See The Five Modules in Detail below for each module.

​

Tools

The report reflects the analysis from the most recent completed data collection.

If your brand has scheduled monitoring, confirm whether the data has been updated before viewing the report; if you need the latest data, refresh first, then generate the report.

Recommended workflow:

1. refresh_brand — Refresh brand data (billed by data collection volume). Triggers a fresh collection; use get_refresh_progress to check collection progress.
2. get_social_voc_report_kit — Generate the brand social analysis report (free). Once the data is ready, it returns the complete report in one call — analysis modules, AI interpretation, and report styling — with the AI assembling the full report automatically.

​

Type 2: AI Deep Follow-up

​

When to use

The report is not the end of analysis, but the starting point for further decisions.

After viewing the social analysis report, you can keep asking the AI questions in natural language about any finding in the report.

AI deep follow-up does not regenerate a report. Instead, on top of the current report, it draws on the complete dataset to keep analyzing — helping you understand causes, find opportunities, and plan actions.

You can ask a single follow-up, or ask several questions in a row around the same report. The entire process stays based on the current report’s data and never re-collects data.

​

Why use AI deep follow-up instead of asking your local AI assistant directly?

Local AI models have excellent general reasoning ability, but for analysis based on a VOC report they lack two key capabilities.

Complete data context

The brand social analysis report shows curated analysis results, while the data that actually supports those conclusions — including:

• Original posts
• Comment content
• Platform metadata
• Sentiment labels
• Intermediate analysis results

is all stored in the complete dataset behind the current report.

AI deep follow-up can access this complete data directly, not just what the report displays, so it can analyze more accurately and comprehensively.

Industry analysis frameworks

Beyond the complete data, AI deep follow-up also combines Pangolinfo’s built-in industry analysis frameworks, including:

• Consumer insight methods
• Brand competition analysis frameworks
• Social marketing best practices
• Sentiment attribution analysis
• Growth-opportunity identification methods

As a result, it not only explains the data but also produces strategy recommendations with real business value.

​

How it differs from the brand social analysis report

​

Recommended follow-up scenarios

Diagnostic (why)

• “Why has this brand’s negative sentiment risen recently?”
• “What content is mainly driving competitor X’s volume growth on YouTube?”
• “What are users discussing behind this platform’s risk alert?”

Strategic (how)

• “Pick 5 KOCs worth partnering with for Anker on YouTube, with reasons.”
• “Give a three-step PR response plan for the recent negative feedback.”
• “If we could only focus on two platforms next quarter, which two do you recommend and why?”

Comparative (A vs B)

• “What’s the reputation gap between us and competitors on after-sales topics?”
• “How do the user profiles of the two competitors differ?”

Decision-making

• “If the budget is cut by 30%, which platforms should we keep first?”
• “If we can only fix one product issue, what should we solve first?”
• “Which directions are most worth investing in next quarter?”

​

⚠️ Important: how to ask for the best results

The value of AI deep follow-up depends on the quality of your question. Three principles:

1. Ask “why” and “how,” not “what”

The data report already answers “what” (how much volume, what the trend is). The value of deep follow-up lies in explaining causes and giving advice.

• ❌ “How much volume this month?” → just check the data report
• ✅ “Volume rose this month — what are the main drivers, and which content contributed most?”

2. Give specific scenarios and constraints

The more specific the question, the more actionable the answer.

• ❌ “Help me analyze competitors”
• ✅ “How does competitor X’s content strategy on TikTok differ from ours? If we want to catch up, what’s the first step?”

3. State your question directly — don’t rewrite or condense it

There’s no need to rephrase, summarize, or trim your question. Just tell the AI what you actually want to solve; AI deep follow-up will understand your intent from the complete data and analyze accordingly.

​

Recommended analysis workflow

A typical end-to-end analysis workflow:

1. Create a knowledge space and collect data
(create_space / refresh_brand)
        │
        ▼
2. Read the brand social analysis report
(get_social_voc_report_kit)
Understand what happened in the market
        │
        ▼
3. Spot questions worth digging into
For example:
• Why is negative sentiment increasing?
• Why is a competitor growing faster?
• Which platform is worth continued investment?
        │
        ▼
4. AI deep follow-up analysis
(report_follow_up_analysis)
Understand causes, find opportunities, form strategy
        │
        ▼
5. Keep following up around the same report
Continuously refine analysis and decisions

​

Type 3: Personas and Simulated Interviews

Use the persona workflow when the question becomes “who will buy, why do they buy, and why do they not buy?”

1. Read the current overview with free get_persona_voc.
2. Call get_persona_section only for sections listed in availableSections, such as persona details, evidence, demographics, discussion topics, or competitor reports.
3. If no persona report exists, or the user explicitly requests regeneration and confirms 99 credits = 59,400 points, call generate_persona_voc, then check the asynchronous run with get_persona_generation_progress.
4. Use list_persona_questions when the user needs prompts. Use ask_persona for one persona and ask_all_personas for the same question across all personas. Both cost 1 credit = 600 points; the broadcast call is billed per request, not per persona.
5. After at least two assistant replies in a first-person interview, use get_persona_interview_summary to summarize it.

ask_persona, ask_all_personas, and interview summaries return simulated content, not real user testimony. On REPORT_SUPERSEDED or CURSOR_STALE, discard the old reportId/cursor and reread the overview.

​

Report-wide conventions

Three rules that run through the whole report:

• Competitor benchmarking first: key metrics show your brand alongside competitors wherever possible. For brand decisions, relative gaps are usually more meaningful than absolute values alone.
• Shared global basis: global controls such as time range and data basis affect the entire report, so confirm the current analysis window first.
• Drill-down by evidence: trends, sentiment, keywords, and AI citations should trace back to source posts — conclusions need to be traceable to evidence.

​

Global controls

​

Brand header card

The top section of the page carries brand identity and global actions. Switching time range or data basis here refreshes all modules below; high-impact actions such as re-collection, configuration, and export are usually grouped here too.

• Brand name / description: the main monitored brand and a brief business positioning.
• Last refresh time: when the previous collection finished — the basis for judging data freshness; not updated while collecting.
• Monitoring keywords / competitor tags / owned accounts: define the monitoring scope. Adding or removing competitors usually requires a re-collection to take effect.
• Collection schedule: scheduled collection mode, date, and time.
• Brand monitoring in progress: indicates collection is running, so the data may not be the latest.

​

The Five Modules in Detail

​

1. AI Market Insight and Key Metrics

Answers:

How is the brand performing overall?
What are the most important opportunities and risks right now?

Common content includes an overall judgment, growth opportunities, risk alerts, competitive pressure, key metric cards, and a platform distribution summary. Useful for quickly gauging the brand’s current overall state.

​

2. Competitor Comparison · Trends and Share

Answers:

How far apart are we and competitors in voice, engagement, and sentiment?
Who's growing, who's declining?

Useful for judging whether your voice share is expanding, which competitor is more active recently, on which platforms competitors have an edge, and whether you have any clear weak spots.

​

3. User Voice · Sentiment and VOC

Answers:

What are users praising?
What are users complaining about?
What's behind the negative feedback?

Common content includes sentiment distribution, positive drivers, negative drivers, representative posts, and user feedback themes you can follow up on. Useful for product, operations, support, and content teams.

​

4. Platform Analysis · Reputation Board

Answers:

Which platform has a better reputation?
Which platform is more prone to risk?
Do discussion themes differ across platforms?

Useful for deciding where to focus content and investment next, prioritizing risk monitoring, and identifying which platform needs support or PR follow-up.

​

5. Data Center · Posts and Drill-down

Answers:

What are the original posts supporting this conclusion?
Can I filter by a specific platform, sentiment, or theme?

The data center is suited to evidence review and secondary analysis. Through MCP, you can also have the AI find a specific type of post directly:

Find 10 negative posts about the price being too high.
Only show discussions about after-sales on YouTube and Reddit.