aaron-he-zhu/aaron-marketing-skills / influencer/report/performance-analyzer
performance-analyzer
VerifiedUse when the user asks to "analyze influencer campaign performance", "compare influencers", or "find what content worked"; produces metric scorecards vs target and benchmark, platform/influencer/content rankings, engagement-quality and sentiment reads, conversion-attribution breakdowns, and ranked learnings. Not for dollar-level return math — use roi-calculator. 达人营销效果分析/投放复盘
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 15.1 KB |
| references/analysis-templates.md | markdown | 18.2 KB |
references/analysis-templates.md
View on GitHub# Performance Analyzer — Analysis Templates
Fill-in templates for each step of the influencer performance analysis. Each maps to a numbered step in [../SKILL.md](../SKILL.md) Instructions.
Repo-root links from this file use `../../../`.
- [skill-contract.md](../../../../references/skill-contract.md)
- [state-model.md](../../../../references/state-model.md)
- [CONNECTORS.md](../../../../CONNECTORS.md)
---
## Step 1 — Gather Performance Data
```markdown
### Performance Data Collection
**Campaign**: [name]
**Period**: [start] - [end]
**Influencers**: [count]
**Platforms**: [platforms]
### Data Sources
| Source | Metrics Available | Collection Method |
|--------|-------------------|-------------------|
| Native analytics | Reach, views, engagement | Platform export |
| Influencer reports | Screenshots/exports | From creators |
| Website analytics | Traffic, conversions | GA/tracking |
| Sales data | Revenue, orders | E-commerce platform |
| Promo code data | Redemptions | Sales system |
```
---
## Step 2 — Analyze Core Metrics
```markdown
## Campaign Performance Overview
### Summary Metrics
`Conversions` and `Revenue` below must reference the reconciled counted total from Step 7 under one deduplicated attribution model; until Step 7 passes, both are `Unknown/NEEDS_INPUT`. State whether reach is platform-reported (potentially overlapping) or deduplicated person-level reach. Recompute aggregate ER, CVR, CPA, and other rates from compatible summed numerators/denominators—never sum or simple-average row rates.
| Metric | Result | Target | vs. Target | vs. Benchmark |
|--------|--------|--------|------------|---------------|
| Total Reach | [X] | [X] | [+/-X%] | [+/-X%] |
| Total Impressions | [X] | [X] | [+/-X%] | [+/-X%] |
| Total Engagements | [X] | [X] | [+/-X%] | [+/-X%] |
| Engagement Rate | [X%] | [X%] | [+/-X%] | [+/-X%] |
| Total Video Views | [X] | [X] | [+/-X%] | [+/-X%] |
| Link Clicks | [X] | [X] | [+/-X%] | [+/-X%] |
| Promo Code Uses | [X] | [X] | [+/-X%] | N/A |
| Conversions | [X] | [X] | [+/-X%] | [+/-X%] |
| Revenue | $[X] | $[X] | [+/-X%] | N/A |
### Comparison Status
**Status**: [ABOVE_TARGET / AT_TARGET / BELOW_TARGET / MIXED / UNKNOWN / NOT_COMPARABLE]
Derive this only from the declared metric/target rule and compatible rows above. If a target, source date, window, attribution basis, or better-direction rule is missing, use `UNKNOWN` or `NOT_COMPARABLE`. Do not invent an aggregate `/10` score or adjective verdict.
### Key Highlights
✅ **What Exceeded Expectations**:
- [Highlight 1]
- [Highlight 2]
⚠️ **What Underperformed**:
- [Issue 1]
- [Issue 2]
```
---
## Step 3 — Analyze by Platform
```markdown
## Platform Performance
### Platform Comparison
| Platform | Reach | Engagements | ER | Clicks | Conversions | CPA |
|----------|-------|-------------|-------|--------|-------------|-----|
| Instagram | [X] | [X] | [%] | [X] | [X] | $[X] |
| TikTok | [X] | [X] | [%] | [X] | [X] | $[X] |
| YouTube | [X] | [X] | [%] | [X] | [X] | $[X] |
| **Total** | **[X]** | **[X]** | **[%]** | **[X]** | **[X]** | **$[X]** |
### Platform Insights
**Observed highest platform under the declared comparable metric**: [Platform / NOT_RANKED]
- Evidence: [metric, window, attribution basis, source_ref]
- Possible explanation: [explicit hypothesis or `not tested`]
**Observed lowest platform under the declared comparable metric**: [Platform / NOT_RANKED]
- Evidence: [metric, window, attribution basis, source_ref]
- Possible explanation: [explicit hypothesis or `not tested`]
- Improvement opportunity: [suggestion]
### Platform-Specific Metrics
#### Instagram
| Metric | Feed Posts | Reels | Stories |
|--------|------------|-------|---------|
| Reach | [X] | [X] | [X] |
| Engagements | [X] | [X] | [X] |
| ER | [%] | [%] | [%] |
| Saves | [X] | [X] | N/A |
| Shares | [X] | [X] | [X] |
#### TikTok
| Metric | Result | Benchmark |
|--------|--------|-----------|
| Views | [X] | |
| Likes | [X] | |
| Comments | [X] | |
| Shares | [X] | |
| Average Watch Time | [X]s | |
| Completion Rate | [%] | |
```
---
## Step 4 — Analyze by Influencer
```markdown
## Influencer Performance
### Creator Comparison
**Declared ranking metric/rule**: [metric, better direction, window, attribution basis, complete creator scope, decision/significance rule or missing]
| Rank/status | Creator ref | Reach | ER | Conversions | ROAS artifact value/ref | Comparability |
|-------------|-------------|-------|----|-------------|-------------------------|---------------|
| [1 / NOT_RANKED] | creator-<UUIDv4> | [X] | [%] | [X] | [X]:1 / [roi_ref] | [comparable/gap] |
| [2 / NOT_RANKED] | creator-<UUIDv4> | [X] | [%] | [X] | [X]:1 / [roi_ref] | [comparable/gap] |
| [3 / NOT_RANKED] | creator-<UUIDv4> | [X] | [%] | [X] | [X]:1 / [roi_ref] | [comparable/gap] |
### Highest Observed Comparable Row
#### [creator_ref / NOT_RANKED]
| Metric | Result | vs. Campaign Avg |
|--------|--------|------------------|
| Reach | [X] | [+/-X%] |
| Engagement Rate | [%] | [+/-X%] |
| Video Completion | [%] | [+/-X%] |
| Click-through Rate | [%] | [+/-X%] |
| Conversion Rate | [%] | [+/-X%] |
| Cost per Conversion | $[X] | [+/-X%] |
**Observed differences**:
- [dated metric difference + evidence ref]
- [dated metric difference + evidence ref]
**Causal explanation**: [supported by supplied experimental/quasi-experimental evidence, or `not established`]
**Hypothesis for next test**: [one falsifiable hypothesis or none]
**Content Analysis**:
- Format: [what they posted]
- Hook: [how they opened]
- Message: [how they communicated]
- CTA: [what they asked viewers to do]
**Retro decision**: [renew/retest/retire/unknown under the preregistered rule; default unknown]
### Lower Observed Comparable Row
#### [creator_ref / NOT_RANKED]
**Results**: [summary]
**Observed difference**: [dated metric + evidence ref]
**Causal explanation**: [supported evidence or `not established`]
**Learning**: [what to do differently]
```
---
## Step 5 — Content Performance Analysis
```markdown
## Content Performance
### Top Performing Content
| Rank/status | Creator ref | Platform | Format | Reach | ER | Observed feature |
|-------------|-------------|----------|--------|-------|----|------------------|
| [1/NOT_RANKED] | creator-<UUIDv4> | [platform] | [format] | [X] | [%] | [visible feature + source_ref] |
| [2/NOT_RANKED] | creator-<UUIDv4> | [platform] | [format] | [X] | [%] | [visible feature + source_ref] |
| [3/NOT_RANKED] | creator-<UUIDv4> | [platform] | [format] | [X] | [%] | [visible feature + source_ref] |
### Content Format Analysis
| Format | Pieces | Avg Reach | Avg ER | Best Performer |
|--------|--------|-----------|--------|----------------|
| Video (Reels/TikTok) | [#] | [X] | [%] | [creator_ref] |
| Static Images | [#] | [X] | [%] | [creator_ref] |
| Carousels | [#] | [X] | [%] | [creator_ref] |
| Stories | [#] | [X] | [%] | [creator_ref] |
| YouTube Videos | [#] | [X] | [%] | [creator_ref] |
### Content Theme Analysis
| Theme | Pieces | Avg ER | Conversion Rate | Notes |
|-------|--------|--------|-----------------|-------|
| Product demo | [#] | [%] | [%] | [notes] |
| Lifestyle | [#] | [%] | [%] | [notes] |
| Tutorial | [#] | [%] | [%] | [notes] |
| Review | [#] | [%] | [%] | [notes] |
| Unboxing | [#] | [%] | [%] | [notes] |
### Observed Content Associations
**Hook patterns associated with the observed result**:
- [Pattern 1]: [examples + comparable metric/evidence]
- [Pattern 2]: [examples + comparable metric/evidence]
**Messaging associations**:
- [Message type 1]: [observed result; causality not established / supplied causal evidence]
- [Message type 2]: [observed result; causality not established / supplied causal evidence]
**Visual Elements That Performed**:
- [Element 1]
- [Element 2]
```
---
## Step 6 — Engagement Quality Analysis
```markdown
## Engagement Quality
### Engagement Breakdown
| Type | Volume | % of declared sample | Quality state | Rubric/evidence ref |
|------|--------|----------------------|---------------|---------------------|
| Likes | [X] | [%] | [typed state or Unknown] | [ref] |
| Comments | [X] | [%] | [typed state or Unknown] | [ref] |
| Saves | [X] | [%] | [typed state or Unknown] | [ref] |
| Shares | [X] | [%] | [typed state or Unknown] | [ref] |
| Link clicks | [X] | [%] | [typed state or Unknown] | [ref] |
### Comment Sentiment Analysis
**Sampling/coding contract**: [platform/content scope, start/end, sampled and eligible counts, sampling method, coding rubric/model/version, denominator, source/evidence refs]. Without this contract, sentiment percentages and intent labels are `Unknown/NOT_SCORED`.
| Sentiment | % | Examples |
|-----------|---|----------|
| Positive | [%] | "[example]", "[example]" |
| Neutral/Questions | [%] | "[example]", "[example]" |
| Negative | [%] | "[example]", "[example]" |
**Key Themes in Comments**:
- [Theme 1]: [frequency] mentions
- [Theme 2]: [frequency] mentions
- [Theme 3]: [frequency] mentions
### Purchase Intent Signals
| Signal | Count | Examples |
|--------|-------|----------|
| "Where to buy" questions | [#] | |
| Price questions | [#] | |
| Code requests | [#] | |
| "Just ordered" | [#] | |
| Tagged friends | [#] | |
### Engagement Quality Read
**Status**: [SUPPORTED / MIXED / UNKNOWN / NOT_SCORED]
**Rubric/calculation ref**: [supplied rubric + inputs + formula, or none]
Do not emit a `/10` score unless the user supplies the rubric, required inputs, and calculation. Missing comment/source evidence is `UNKNOWN`/`NOT_SCORED`, not an estimated score.
```
---
## Step 7 — Conversion & Attribution Analysis
```markdown
## Conversion Analysis
### Conversion Funnel
```
Reach [bar from supplied data] [reach] ([share])
↓
Engagements [bar from supplied data] [engagements] ([rate])
↓
Link Clicks [bar from supplied data] [clicks] ([rate])
↓
Site Visits [bar from supplied data] [visits] ([rate])
↓
Add to Cart [bar from supplied data] [adds] ([rate])
↓
Purchases [bar from supplied data] [purchases] ([rate])
```
Use `Unknown` for any stage not supplied or observed. Do not infer missing counts from the shape of the illustrative bars.
### Conversion Metrics
| Metric | Result | Benchmark + source/date | Status |
|--------|--------|-------------------------|--------|
| Click-through Rate | [%] | [%] ([source/date]) | [above/below/equal/pending] |
| Landing Page CVR | [%] | [%] ([source/date]) | [above/below/equal/pending] |
| Overall CVR | [%] | [%] ([source/date]) | [above/below/equal/pending] |
| Cost per Click | $[X] | $[X] ([source/date]) | [above/below/equal/pending] |
| Cost per Conversion | $[X] | $[X] ([source/date]) | [above/below/equal/pending] |
### Deduplicated Attribution
**Declared model/window**: [model + dates]
**Deduplication key/rule**: [order/event/customer key + precedence rule]
| Mutually exclusive counted bucket | Conversions | Attributed revenue | % of counted total | Evidence refs |
|-------------------------------------|-------------|--------------------|--------------------|---------------|
| Promo-only after reconciliation | [X] | $[X] | [%] | [refs] |
| UTM-only after reconciliation | [X] | $[X] | [%] | [refs] |
| Other direct bucket under model | [X] | $[X] | [%] | [refs] |
| Non-additive evidence | Value | Status |
|-----------------------|-------|--------|
| Overlapping promo + UTM/direct observations | [X] | [reconciled/unresolved] |
| Estimated influence | [X] | Estimated; excluded from counted total and percentages |
The mutually exclusive counted buckets must sum exactly to the declared counted total. Never add overlapping promo, UTM, and direct observations; unresolved overlaps return `NEEDS_INPUT` for the attribution total.
### Promo Code Performance
| Code ref | Creator ref | Uses | Attributed revenue | AOV |
|----------|-------------|------|--------------------|-----|
| [opaque code_ref] | [creator_ref] | [X] | $[X] | $[X] |
| [opaque code_ref] | [creator_ref] | [X] | $[X] | $[X] |
| [opaque code_ref] | [creator_ref] | [X] | $[X] | $[X] |
```
---
## Step 8 — Generate Insights & Recommendations
```markdown
## Insights & Recommendations
### Top 5 Learnings
1. **[Learning 1]**
- What we observed: [data]
- Why it matters: [significance]
- Future application: [how to use this]
2. **[Learning 2]**
- What we observed: [data]
- Why it matters: [significance]
- Future application: [how to use this]
[Continue for top 5]
### Observed Higher Results
| Element | Performance | Recommendation |
|---------|-------------|----------------|
| [Element 1] | [comparable metric + evidence] | [bounded retest or rule-backed action] |
| [Element 2] | [comparable metric + evidence] | [bounded retest or rule-backed action] |
### Observed Lower Results
| Element | Performance | Recommendation |
|---------|-------------|----------------|
| [Element 1] | [comparable metric + evidence] | [bounded retest or stop-rule action] |
| [Element 2] | [comparable metric + evidence] | [bounded retest or stop-rule action] |
### Optimization Opportunities
| Opportunity | Impact hypothesis | Evidence/model ref | Effort | Priority rule |
|-------------|-------------------|--------------------|--------|---------------|
| [Opportunity 1] | [testable hypothesis] | [ref or Unknown] | [effort] | [declared rule] |
| [Opportunity 2] | [testable hypothesis] | [ref or Unknown] | [effort] | [declared rule] |
### Influencer Roster Recommendations
| Creator ref | Next-cycle decision | Evidence-backed rationale |
|-------------|---------------------|---------------------------|
| [creator_ref] | `renew` | [decision rule cleared + evidence ref] |
| [creator_ref] | `retest` | [mixed/inconclusive evidence + testable change] |
| [creator_ref] | `retire` | [declared stop rule or hard constraint + evidence ref] |
| [creator_ref] | `unknown` | [missing/unaligned evidence] |
### Future Campaign Recommendations
1. **Platform Mix**: [recommendation]
2. **Influencer Tier**: [recommendation]
3. **Content Format**: [recommendation]
4. **Messaging**: [recommendation]
5. **Budget Allocation**: [recommendation]
### Campaign Retro Card — only with verified current tracker state `measured` or `closed`
Repeat this compact card for each creator decision requested. Do not inherit scope implicitly or trust a bare stage string: cite the current non-forked tracker-state evidence, measurement contract, and decision rule.
- `campaign_id`: [exact campaign ID]
- `creator_ref`: [exact opaque creator ref]
- `tracker_state_ref`: [current non-forked state artifact/receipt ref proving measured or closed]
- `measurement_contract_ref`: [current locked contract ref]
- `decision_rule_ref`: [precommitted rule ref]
- `decision`: `renew | retest | retire | unknown`
- `evidence_backed_rationale`: [decision rule + observed result; cite, do not infer]
- `evidence_refs`: [[dated source/artifact reference], ...]
- `next_campaign_hypothesis`: [one falsifiable change to test next, or `none`]
- `unresolved_limitations`: [[missing source/window/attribution/comparability issue], ...]
```
The card is WARM working state, not a STAR/SQS or creator-content-auditor verdict. A missing/mismatched/forked scope or state ref blocks the card; missing comparison/decision evidence forces `unknown`. After the analysis/card is saved with exact authorization, offer a separate handoff to `campaign-planner` to append that artifact reference to the relevant tracker row's `evidence_refs`; do not update the tracker or advance its stage from this skill. An optional next-cycle `fit-scorer` handoff passes only cited evidence and the hypothesis, requires an explicit user choice, and never carries a simulated STAR/SQS verdict.
---
## Worked Example (full)
**User**: "Analyze our summer skincare campaign using this supplied dated export: total reach 2.4M (target 2M) and engagement rate 4.2% (target 3.5%). Attribution artifact `attr-reconciliation-ref-01` covers the complete conversion universe, uses unique `order_id`, the predeclared last-non-direct model/window, and reports a reconciled counted total of 1,847 conversions and $142,500 attributed revenue (targets: 2,000 and $150,000). ROI-calculator artifact `roi-ref-01` reports campaign ROAS 2.8:1 (target 3:1) and creator-ref-01 ROAS 4.2:1 under that same counted universe/cost window. The supplied creator rows say creator-ref-01 has the highest conversion count, creator-ref-02 has the highest ER at 6.8%, and creator-ref-03 has the highest reach per dollar. No matched platform comparison, significance result, complete roster decision rule, or other evidence refs were supplied."
**Output**:
```markdown
# Summer Skincare Campaign Performance Analysis
## Executive Summary
**Campaign Performance**: Mixed against the supplied targets
**Provenance**: reach/engagement results and targets are user-provided from the stated export; conversions/revenue cite `attr-reconciliation-ref-01`; ROAS cites `roi-ref-01`; status deltas are Calculated. This analysis does not promote them to independently verified, causal, or incremental results.
| Metric | Result | Target | Status | Evidence ref |
|--------|--------|--------|--------|--------------|
| Total Reach | 2.4M | 2M | ✅ +20% | supplied export |
| Engagement Rate | 4.2% | 3.5% | ✅ +20% | supplied export |
| Conversions | 1,847 | 2,000 | ⚠️ -7.7% | attr-reconciliation-ref-01 |
| Revenue | $142,500 | $150,000 | ⚠️ -5% | attr-reconciliation-ref-01 |
| ROAS | 2.8:1 | 3:1 | ⚠️ -6.7% | roi-ref-01 |
## Supplied Creator Highlights
1. **creator-ref-01** — ROAS 4.2:1 from `roi-ref-01`; supplied highest conversion count under `attr-reconciliation-ref-01`
2. **creator-ref-02** — supplied highest engagement rate, 6.8%
3. **creator-ref-03** — supplied highest reach per dollar
## Key Learning
**Platform comparison**: `NEEDS_INPUT` — no matched TikTok/Instagram result, comparable attribution basis, or significance test was supplied. A possible platform-mix explanation may be recorded only as a hypothesis for a future test, not as an observed cause.
## Recommendation
Do not renew, replace, or reallocate from this input alone. Request the full creator/platform breakdown, comparable windows and attribution definitions, evidence refs, and the precommitted decision rule; until then, the creator decision is `unknown` and any next-cycle change is an explicitly labeled hypothesis.
```
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |