Brand Sentiment Analyzer — Multi-Source Mentions + AI Insights

Monitor brand sentiment across Reddit, Google News, and Trustpilot in one run. Get an executive summary with a brand health score, top complaints, top praise, crisis signals, and share-of-voice vs an optional competitor — plus per-mention details with individual sentiment classification. AI-powered theme extraction via Anthropic Claude.

Built as a drop-in replacement for Brand24 ($79+/mo), Mention ($41+/mo), and Meltwater ($5K-$24K/yr). Pay only for what you run — typical brand report is ~$0.80.

What this Actor does

You give it a brand name (e.g., OpenAI, Tesla, Apify). It:

1. Searches Reddit (posts + comments mentioning the brand)
2. Searches Google News (articles published about the brand)
3. Searches Trustpilot (customer reviews of the brand)
4. Aggregates and deduplicates the mentions
5. Sends them to Claude AI for theme extraction, complaint clustering, and crisis signal detection
6. Outputs one executive summary + N raw mentions with individual sentiment

Optionally, it does the same for a competitor brand and computes share-of-voice + sentiment delta.

Twitter/X and YouTube comment sources are reserved for v1.1.

Why use this Actor

• Replaces $79+/mo brand-monitoring SaaS — pay only when you run a report
• AI-powered insights — not just "60% positive / 40% negative" (lexicon junk), but real themes like "customers complain about late Black Friday delivery"
• Crisis detection — flags boycott calls, lawsuits, viral negative posts, safety/regulatory concerns
• Share-of-voice — head-to-head sentiment comparison with a competitor in one run
• MCP-ready for AI agents — Claude/GPT/Gemini can call this actor via Apify MCP for real-time brand-reputation context
• No API keys to manage — Claude API runs server-side; you just provide the brand name

How to use it

1. Open the Apify Console and find this Actor.
2. Enter your brand name in the Brand name or keyword field.
3. (Optional) Add a competitor brand for share-of-voice comparison.
4. (Optional) Adjust mentions-per-platform (default 25) and time window (default 7 days).
5. Click Start.
6. Read the summary (first dataset record) — get the brand health score, top complaints, top praise, crisis signals, and executive summary at a glance.
7. Drill into the per-mention data for evidence and source URLs.

Typical run: ~30-60 seconds, ~$0.80 per brand report.

Input

Field	Type	Default	Description
brandName	string	(required)	Brand or keyword to monitor (e.g., "OpenAI", "Tesla")
competitorBrand	string	""	Optional competitor for share-of-voice comparison
platforms	array	["reddit", "google-news", "trustpilot"]	Sources to monitor
maxMentionsPerPlatform	int	25	Hard cap per source (5-500)
timeRange	enum	"7d"	24h, 7d, 30d, or all
language	enum	"en"	ISO 639-1 code (en/es/de/fr/pt/it/ja) or all
includeAiInsights	bool	true	Run Claude for themes + crisis signals (adds $0.30)
responseFormat	enum	"detailed"	concise (~80 tokens/item, agent-friendly) or detailed
maxCostUsd	int	0	Hard cost cap; 0 = no cap

Example input

{
  "brandName": "Apify",
  "competitorBrand": "Bright Data",
  "platforms": ["reddit", "google-news", "trustpilot"],
  "maxMentionsPerPlatform": 30,
  "timeRange": "7d",
  "language": "en",
  "includeAiInsights": true,
  "responseFormat": "detailed"
}

Output

The dataset contains:

• One summary record (first item, _summary: true) — executive analysis
• Many mention records (recordType: "mention") — individual mentions with per-mention sentiment

You can download the dataset in JSON, CSV, HTML, or Excel. The full summary is also saved as a standalone file in the key-value store under summary-<brand>-<timestamp>.

Example summary record

{
  "_summary": true,
  "recordType": "summary",
  "brandName": "Apify",
  "competitorBrand": "Bright Data",
  "brandHealthScore": 7.4,
  "sentimentBreakdown": {
    "positive": 42, "neutral": 28, "negative": 20,
    "positivePercent": 46.7, "negativePercent": 22.2
  },
  "topComplaints": [
    { "theme": "Pricing confusion", "frequency": 6,
      "exampleQuote": "Took me 30 minutes to figure out how compute units work" },
    { "theme": "Cold-start latency", "frequency": 4,
      "exampleQuote": "First run takes 90s before the actor even starts" }
  ],
  "topPraise": [
    { "theme": "MCP integration", "frequency": 14,
      "exampleQuote": "Apify MCP is the cleanest way to give Claude scraping superpowers" },
    { "theme": "Crawlee quality", "frequency": 11,
      "exampleQuote": "Crawlee handles anti-bot stuff that took us months to build in-house" }
  ],
  "crisisSignals": [],
  "shareOfVoice": {
    "brandMentions": 90, "competitorMentions": 64,
    "brandSharePercent": 58.4, "brandSentimentDelta": 1.8
  },
  "executiveSummary": "Apify is in healthy positive territory (7.4/10), driven by strong developer enthusiasm for MCP integration and Crawlee. Two friction points stand out: pricing transparency and cold-start latency. No crisis signals. Apify holds 58% share-of-voice vs Bright Data with a +1.8 sentiment lead.",
  "mentionCounts": { "twitter": 0, "reddit": 38, "youtube": 0, "googleNews": 27, "trustpilot": 25, "total": 90 },
  "warnings": [],
  "scrapedAt": "2026-05-03T14:00:00.000Z"
}

Example mention record

{
  "recordType": "mention",
  "brandName": "Apify",
  "platform": "reddit",
  "sourceUrl": "https://www.reddit.com/r/webscraping/comments/...",
  "author": "scrape_dev_42",
  "text": "Apify MCP changed how I integrate scrapers with Claude — finally stable.",
  "title": "Apify MCP review after 3 months",
  "publishedAt": "2026-05-01T09:23:00.000Z",
  "engagement": { "score": 87, "comments": 14, "subreddit": "webscraping" },
  "sentiment": "positive",
  "sentimentScore": 0.78,
  "scrapedAt": "2026-05-03T14:00:00.000Z"
}

Data fields

Field	Where	Description
recordType	both	summary or mention
brandHealthScore	summary	0–10 overall sentiment score
sentimentBreakdown	summary	counts + % positive/negative
topComplaints	summary	top 5 negative themes (Claude)
topPraise	summary	top 5 positive themes (Claude)
crisisSignals	summary	boycott/lawsuit/safety risks (Claude)
shareOfVoice	summary	mention-share + sentiment delta vs competitor
executiveSummary	summary	3–5 sentence narrative (Claude)
mentionCounts	summary	per-source mention counts
platform	mention	reddit, google-news, trustpilot
sourceUrl	mention	direct link to mention
text	mention	full mention text
sentiment	mention	positive / neutral / negative
sentimentScore	mention	-1.0 to +1.0
engagement	mention	platform-specific (score/rating/source)
publishedAt	mention	ISO 8601

Pricing — How much does it cost to monitor a brand?

This Actor uses Apify pay-per-event pricing. You only pay for what you actually use:

Event	Price	Description
apify-actor-start	$0.005	Once per run (covers orchestration overhead)
mention-found	$0.005	Per mention pushed to dataset
ai-insights-generated	$0.30	Once per run when AI report is generated

Typical run cost (3 sources × 25 mentions = 75 mentions + AI report): 0.005 + (75 × 0.005) + 0.30 = $0.68 per brand report

Comparison:

• Brand24: $79+/mo for ~2K mentions = $0.04/mention (no AI summary)
• Mention.com: $41+/mo for 250 mentions = $0.16/mention (no AI summary)
• This actor: $0.005/mention + AI = up to 30× cheaper per mention with deeper analysis

Within the x402 agent-payment prepay limit of $1, so AI agents can call this actor on demand.

Tips & advanced options

• Crisis monitoring — set timeRange: "24h" and schedule the actor every 4 hours. AI flags boycotts and lawsuits in crisisSignals.
• Launch tracking — set timeRange: "7d" and run daily after launch.
• Steady-state brand health — set timeRange: "30d" and run weekly.
• Lower cost — set includeAiInsights: false to skip the $0.30 AI report and rely on lexicon-based per-mention sentiment only.
• Compact agent output — set responseFormat: "concise" to drop heavy fields; ~80 tokens/mention vs 250.
• Cost cap — set maxCostUsd: 5 to guarantee no run exceeds $5.

FAQ, disclaimers, and support

Is brand monitoring legal? Yes — this actor only queries publicly indexed data (Reddit JSON search, Google News RSS, Trustpilot's public review pages). It does not bypass authentication walls. Users are responsible for compliance with their own jurisdiction's data-protection rules.

Why aren't Twitter and YouTube included yet? Twitter/X requires either authenticated scrapers or paid API access; we're integrating it via Actor.call to a vetted X scraper in v1.1. YouTube requires a 2-step pipeline (search videos → scrape comments), planned for the same release.

What if Trustpilot doesn't have my brand? The Trustpilot adapter resolves brand names to Trustpilot business slugs (e.g., amazon → amazon.com). If no business unit is found, it returns 0 mentions for that source — Reddit and Google News continue normally. Note: Trustpilot uses Cloudflare bot management which blocks non-browser TLS fingerprints on many brand pages — coverage is best for indie / smaller brands and may fail for major retailers (Amazon, Apple, etc.). v1.3 will add browser-fingerprint scraping for full Trustpilot coverage.

What if Claude is unavailable? The actor falls back to lexicon-based per-mention sentiment scoring. The summary record will still ship with brandHealthScore, sentimentBreakdown, and an executiveSummary derived from the lexicon counts. AI-only fields (topComplaints, topPraise, crisisSignals) will be null and a warning will be added.

Can I run this on a schedule? Yes. Use the Apify Schedules feature to run daily/hourly. Set a maxCostUsd cap to avoid surprise bills.

Found a bug or want a new source? Use the ../issues on this Actor's page. Custom integrations available — DM the developer.

---

Built by khadinakbar — see the full portfolio of 50+ data actors.