Traffic Analyzer Agent Skill

Traffic Analyzer

Analyze website traffic, engagement, and traffic sources for any domain. Goes beyond

raw data — scores the domain, interprets what the traffic patterns mean for affiliate

promotion, and recommends whether the program is worth your time.

A tool returns numbers. This skill returns a verdict.

Use cases:

• Is this affiliate program's website healthy? (High traffic = more brand awareness = easier conversions)
• Where does a competitor get their traffic? (Find channels they're ignoring)
• Compare 2-3 affiliate programs by advertiser website strength
• Validate a niche by checking traffic to the top programs in it

Stage

This skill belongs to Stage S1: Research

When to Use

• Before committing to promote an affiliate program — check if the advertiser is legit
• When comparing multiple programs — traffic is a proxy for brand strength
• When competitor-spy identifies competitor sites — analyze their traffic sources
• When evaluating a niche — check if the top programs have healthy traffic
• When an advertiser claims "millions of users" — verify with data

Input Schema

domains: string[]             # (required) 1-5 domains to analyze — "heygen.com", "synthesia.io"
compare: boolean              # (optional, default: true if 2+ domains) Side-by-side comparison
focus: string                 # (optional, default: "affiliate") 
                              # "affiliate" — score from promoter perspective
                              # "competitor" — analyze as a competitor site
                              # "advertiser" — evaluate advertiser health

Workflow

Step 1: Gather Traffic Data

With SimilarWeb API (see shared/references/social-data-providers.md):

If social_data_config.similarweb is configured:

• Call SimilarWeb API for each domain
• Returns: global rank, country rank, visits, pages/visit, avg duration, bounce rate, traffic sources

Without API (web_search fallback):

For each domain:

• web_search "[domain] traffic similarweb" → often shows rank and visit estimates in snippets
• web_search "[domain] site traffic statistics" → third-party reports
• web_search "site:[domain]" → Google index count as proxy for content depth
• web_search "[domain] alexa rank" OR "[domain] semrush traffic" → alternative sources
• web_fetch "https://www.similarweb.com/website/[domain]/" → extract visible data from SimilarWeb free page (may be limited)

Note: web_search data is approximate. SimilarWeb API provides exact metrics.

Step 2: Analyze Core Metrics

For each domain, analyze and interpret:

Traffic Volume:

global_rank: number            # Lower = better. <10K = major site, <100K = solid, <1M = niche
country_rank: number           # Rank in primary country
monthly_visits: string         # "1.2M", "350K", "45K"
visits_trend: string           # "growing" | "stable" | "declining" (if historical data available)

Engagement Quality:

pages_per_visit: number        # >3 = good engagement, <2 = bouncy
avg_visit_duration: string     # ">3 min" = engaged, "<1 min" = low quality
bounce_rate: number            # <40% = excellent, 40-60% = normal, >60% = concerning

Traffic Sources Breakdown:

direct: number                 # % — brand strength indicator
search: number                 # % — SEO strength
social: number                 # % — social media presence
referral: number               # % — partnership/affiliate ecosystem
paid: number                   # % — ad spend (high paid = advertiser invests in acquisition)

Step 3: Interpret for Use Case

For affiliate promoters (focus: "affiliate"):

Score the domain as an affiliate promotion target:

| Signal | Good (8-10) | OK (5-7) | Red Flag (1-4) |

|--------|-------------|----------|-----------------|

| Monthly visits | >500K | 50K-500K | <50K |

| Bounce rate | <40% | 40-60% | >70% |

| Search traffic | >30% | 15-30% | <10% (overly dependent on ads) |

| Brand (direct) | >30% | 15-30% | <10% (nobody knows them) |

| Pages/visit | >4 | 2-4 | <2 |

Why this matters for affiliates:

• High traffic = people already search for this brand → easier to convert your referrals
• Strong brand (high direct traffic) = trust → higher conversion rate
• Good engagement = product delivers value → lower refund rate → your commissions stick
• Healthy search traffic = sustainable business → long-term commission potential
• High paid traffic = advertiser invests in growth → good sign for program longevity

For competitor analysis (focus: "competitor"):

• Identify their strongest traffic channel → where are they winning?
• Find their weakest channel → opportunity for you
• Compare bounce rate → are they retaining visitors better than you?
• Check referral traffic → which sites link to them? (potential partnership targets)

For advertiser evaluation (focus: "advertiser"):

• Is the advertiser's website healthy? (declining traffic = risky to promote)
• Do they invest in marketing? (paid traffic % shows ad budget)
• Is their product sticky? (engagement metrics reveal product quality)
• How established are they? (global rank trajectory)

Step 4: Generate Traffic Score

Calculate an overall Traffic Health Score (0-100):

traffic_score = (
  rank_score × 0.20 +           # Based on global rank
  volume_score × 0.25 +          # Based on monthly visits  
  engagement_score × 0.25 +      # Based on bounce rate + pages/visit + duration
  diversity_score × 0.15 +       # Traffic source diversity (not overly dependent on one channel)
  brand_score × 0.15             # Direct traffic % (brand recognition)
)

Score interpretation:

• 80-100: Excellent. Strong, established brand. Safe to promote long-term.
• 60-79: Good. Healthy traffic. Solid promotion candidate.
• 40-59: Fair. Growing or niche site. Evaluate other factors (commission, product quality).
• 20-39: Weak. Low traffic or declining. Proceed with caution.
• 0-19: Red flag. Very low traffic, new, or declining fast. Not recommended unless early-stage with high commission.

Step 5: Compare (if multiple domains)

If 2+ domains provided, create side-by-side comparison:

• Which has more traffic?
• Which has better engagement?
• Which has more diverse traffic sources?
• Which is growing faster?
• Overall winner with reasoning

Step 6: Self-Validation

Before presenting output, verify:

• [ ] Data source clearly stated (API vs web_search estimate)
• [ ] Scores are calibrated (not all 8/10 — differentiate clearly)
• [ ] Interpretation matches the focus (affiliate vs competitor vs advertiser)
• [ ] Red flags explicitly called out, not buried
• [ ] Recommendation is actionable and specific

If any check fails, fix before delivering. Do not flag checklist to user.

Output Schema

output_schema_version: "1.0.0"
domains_analyzed:
  - domain: string
    data_source: "similarweb_api" | "web_search_estimate"
    metrics:
      global_rank: number | null
      country_rank: number | null
      country: string | null
      monthly_visits: string
      pages_per_visit: number | null
      avg_duration: string | null
      bounce_rate: number | null
    traffic_sources:
      direct: number | null        # percentage
      search: number | null
      social: number | null
      referral: number | null
      paid: number | null
    traffic_score: number          # 0-100
    verdict: string                # "excellent" | "good" | "fair" | "weak" | "red_flag"
    interpretation: string         # 2-3 sentence analysis based on focus
comparison: object | null          # if 2+ domains
  winner: string
  reasoning: string
recommended_next_skill: string

Output Format

## Traffic Analysis: [Domain(s)]

### Data Source
📊 **[SimilarWeb API | Web search estimates (approximate)]**

---

### [domain1.com] — Traffic Score: [XX]/100 — [Verdict]

| Metric | Value | Assessment |
|--------|-------|------------|
| Global Rank | #XX,XXX | [Good/Fair/Low] |
| Monthly Visits | X.XM | [High/Medium/Low] |
| Pages/Visit | X.X | [Engaged/Average/Bouncy] |
| Avg Duration | Xm Xs | [Good/Low] |
| Bounce Rate | XX% | [Healthy/Concerning/High] |

**Traffic Sources:**

Direct: ██████████░░░░░░ 35% (strong brand)

Search: ████████░░░░░░░░ 28% (good SEO)

Social: ████░░░░░░░░░░░░ 15% (social presence)

Referral: ███░░░░░░░░░░░░░ 12% (affiliate ecosystem)

Paid: ██░░░░░░░░░░░░░░ 10% (moderate ad spend)

**What This Means for You:**
[2-3 sentences interpreting metrics for the user's focus — affiliate/competitor/advertiser]

---

### [If comparing 2+ domains]

### Head-to-Head: [domain1] vs [domain2]

| Metric | [domain1] | [domain2] | Winner |
|--------|-----------|-----------|--------|
| Traffic Score | XX/100 | XX/100 | [domain] |
| Monthly Visits | X.XM | XXK | [domain] |
| Engagement | X.X pg/visit | X.X pg/visit | [domain] |
| Brand Strength | XX% direct | XX% direct | [domain] |
| SEO | XX% search | XX% search | [domain] |

**Verdict:** [domain1] is the stronger affiliate promotion target because [reasoning].

---

### 🎯 Recommendation

[Specific, actionable recommendation based on focus]

### Next Steps
- `affiliate-program-search` — check commission details for [domain]
- `competitor-spy` — deep dive into their affiliate strategy
- `trending-content-scout` — find what content about [domain/product] is performing

Error Handling

• No API and web_search returns limited data: Present what's available. Note: *"Limited data available via web search. For accurate metrics, configure SimilarWeb API — see shared/references/social-data-providers.md."* Still provide estimated score.
• Domain not found / too new: Note: *"[domain] has insufficient traffic data. This could mean: (1) very new site, (2) very low traffic, (3) data not yet indexed. This is itself useful information — proceed with caution."* Score: 10/100.
• Domain is a subdomain: Analyze the root domain instead. Note the adjustment.
• More than 5 domains requested: Analyze top 5, suggest running again for the rest.
• SimilarWeb API rate limited: Fall back to web_search for remaining domains.

Examples

Example 1:

User: "Is HeyGen worth promoting? Check their traffic."

→ domain: "heygen.com", focus: "affiliate"

→ SimilarWeb or web_search → Global rank: ~15K, 2.1M monthly visits

→ Score: 82/100 — "Excellent. HeyGen has strong traffic with healthy engagement. 35% direct traffic shows strong brand recognition. Your referral links benefit from existing brand awareness."

→ Next: affiliate-program-search for HeyGen commission details

Example 2:

User: "Compare Notion vs ClickUp vs Monday.com traffic for my productivity niche"

→ domains: ["notion.so", "clickup.com", "monday.com"]

→ Analyze all 3, side-by-side comparison

→ Winner: Notion (highest traffic, best engagement)

→ But: ClickUp has highest referral % (12%) = strongest affiliate ecosystem → may convert better

Example 3:

User: "I found this small SaaS tool — screenpal.com. Is the advertiser legit?"

→ domain: "screenpal.com", focus: "advertiser"

→ Global rank: ~180K, ~300K monthly visits

→ Score: 55/100 — "Fair. Niche tool with moderate traffic. Growing steadily. Low paid traffic (2%) suggests bootstrapped. Engagement is good (3.8 pages/visit). Worth promoting if commission is strong, but don't expect brand-name conversion rates."

Feedback & Issue Reporting

When this skill produces unexpected, incomplete, or incorrect output, generate a

skill_feedback block (see shared/references/feedback-protocol.md for full schema).

Skill-specific failure modes:

• Domain not found in SimilarWeb: Very new or very small site. Report as data_quality, note domain.
• All metrics null from web_search: No traffic data findable. Report as data_quality, severity: medium.
• Traffic score seems wrong: Score doesn't match known reality (e.g., Google.com scored 40/100). Report as wrong_output.

Auto-detect triggers:

• traffic_score is 0 or null for a well-known domain
• All traffic_sources percentages are null
• Comparison requested but only 1 domain returned data

Report issues: GitHub Issues | Discussions

References

• shared/references/social-data-providers.md — SimilarWeb API configuration
• shared/references/flywheel-connections.md — master flywheel connection map
• shared/references/affiliate-glossary.md — affiliate marketing terminology
• shared/references/feedback-protocol.md — issue detection and reporting standard

Flywheel Connections

Feeds Into

• affiliate-program-search (S1) — traffic score as program evaluation factor
• competitor-spy (S1) — traffic sources reveal competitor strategy
• niche-opportunity-finder (S1) — traffic data validates niche demand
• content-angle-ranker (S1) — traffic source breakdown informs platform prioritization
• trending-content-scout (S1) — social traffic % hints at which platforms to scout

Fed By

• competitor-spy (S1) — competitor domains to analyze
• affiliate-program-search (S1) — program URLs to evaluate
• niche-opportunity-finder (S1) — top program domains in a niche

Feedback Loop

• S6 performance-report shows your referral contribution to the advertiser → compare your traffic impact over time → prioritize programs where you move the needle

chain_metadata:
  skill_slug: "traffic-analyzer"
  stage: "research"
  timestamp: string
  suggested_next:
    - "affiliate-program-search"
    - "competitor-spy"
    - "trending-content-scout"