Performance Analyzer

Analyze influencer campaign performance past surface metrics — score results vs target/benchmark, rank platforms/creators/content, read engagement quality and sentiment, attribute conversions, and write ranked learnings.

Cross-discipline (paid ads): this is also the cross-channel paid-ads scorecard/anomaly lens — account-wide metric rollups vs target/benchmark that feed ad-test-designer (what to test) and paid-measurement-loop (what to read back). Save paid runs under memory/ad/performance-analyzer/.

Quick Start

Analyze performance of [campaign name] influencer campaign

Compare creators within one campaign:

Compare performance of these influencers from [campaign]: @handle1, @handle2, @handle3

Skill Contract

Reads: campaign name and date range; native platform analytics (reach, views, engagement); influencer-supplied reports or screenshots; website/GA traffic and conversion data; sales and promo-code redemption data; targets, benchmarks, and the preregistered decision rule/readback window if supplied; the optional lightweight campaign tracker and its evidence_refs; and any ROI/ROAS artifact already computed by roi-calculator. Reuse each explicit upstream opaque creator_ref or a verified creator-registry aggregate ID; a raw handle/name/URL/provider ID is transient lookup input only and never becomes a saved identity. Per-creator baselines come from memory/creators/<aggregate-id>.md only when an authorized artifact or verified registry link resolves that ref. Never derive the path from a raw locator.
Writes: return the performance analysis inline by default. When a current non-forked tracker-state artifact proves measured or closed, include the compact Campaign Retro Card from step 8 bound to that campaign, creator, measurement contract, and decision rule. Save the analysis and card together to memory/influencer/performance-analyzer/YYYY-MM-DD-<campaign>.md only with exact WARM-save authorization; saved tables, headings, evidence, and handoffs use creator_ref plus opaque source refs, never raw handles, names, profile URLs, email addresses, or provider IDs.
Promotes: only with separate exact authorization, promote durable evidence-backed campaign facts (verified metric results and descriptive format/platform associations) to memory/hot-cache.md; any ROI/ROAS value remains tied to its exact roi-calculator artifact. The Retro Card's qualitative renew | retest | retire | unknown decision, rationale, next hypothesis, and limitations remain WARM and are never promoted as registry truth. This skill makes no creator-registry proposal: after a creator row is closed, the existing boundary still permits only a separately authorized, evidence-backed actual rate, signed rights window/expiry, or measured performance baseline to be proposed by the owning workflow; creator-registry alone decides whether it becomes canonical.
Done when:
- Core metrics are compared against compatible source-dated targets/benchmarks. Missing or incompatible context is Unknown/NOT_SCORED, never an invented /10 score or adjective verdict.
- Creators/platforms/content are ranked only under a declared metric, compatible window/basis, complete candidate set, and preregistered decision rule; descriptive associations and causal hypotheses stay visibly separate.
- Conversions use one declared attribution model with deduplicated, mutually exclusive counted buckets; overlapping promo/UTM/direct observations remain reconciliation evidence, and modeled influence stays Estimated outside the counted total.
- With verified current measured or closed state, each requested next-cycle decision has a scope-bound Campaign Retro Card with campaign/creator/state/measurement/decision-rule refs, evidence-backed rationale, evidence_refs, next-campaign hypothesis, and unresolved limitations; insufficient decision evidence resolves to unknown, while missing/forked state blocks the card.
Primary next skill: roi-calculator — turn measured performance into dollar-level return.

Handoff Summary

Emit the standard shape from skill-contract.md §Handoff Summary Format.

Data Sources

This family needs no live integrations (Tier 1). The skill runs entirely on inputs you provide — paste platform exports, influencer report screenshots, GA numbers, and promo-code redemption counts, and it analyzes the supported fields. Missing inputs do not block a partial descriptive read, but any dependent score, verdict, rank, causal explanation, attribution total, or decision becomes Unknown/NOT_SCORED/NEEDS_INPUT rather than being filled in.

Where a connector could speed the work, the skill marks it with a ~~ placeholder:

~~social platform analytics — native reach/engagement/video metrics per post.
~~web analytics — site traffic, click-through, and on-site conversion data.

Measured YouTube post-performance (free key): when campaign content lives on YouTube, python3 "${CLAUDE_PLUGIN_ROOT}/scripts/connectors/youtube.py" videos @creator --limit 20 pulls the actual per-video views/likes/comments for the campaign window — Measured platform metrics without waiting for the creator's screenshot export. Keep both labels honest: API numbers are Measured, creator-supplied numbers are User-provided, and the two can legitimately disagree (display rounding, timing). Free YOUTUBE_API_KEY. See scripts/connectors/README.md.

~~ecommerce / sales platform — revenue, orders, AOV, promo-code redemptions.
~~influencer database — historical creator benchmarks for comparison.

No placeholder is required to run. See CONNECTORS.md for the verified free/keyless data recipe per category.

Instructions

Work the steps below as one dependency-aware pass. Each fill-in template lives in references/analysis-templates.md. Build the Step 2 shell after intake, but run Step 7 before populating or publishing Step 2 Conversions, Revenue, or any rate/cost that depends on them; those fields must cite Step 7's reconciled counted total or remain Unknown/NEEDS_INPUT.

Gather performance data — log campaign/period/influencers/platforms and the available sources (native analytics, influencer reports, web analytics, sales, promo codes). Template: step 1.
Analyze core metrics — compare reach, impressions, engagements, ER, video views, clicks, promo uses, conversions, and revenue against compatible source-dated targets/benchmarks. Emit field-level comparison states; do not invent an aggregate score or adjective verdict. Template: step 2.
Analyze by platform — compare platforms on compatible reach/ER/click/conversion/CPA windows and state observed differences. Put any explanation in a separately labeled hypothesis unless a designed comparison supports it. Template: step 3.
Analyze by creator — use opaque creator_ref; rank only comparable rows under the declared rule. Consume ROI/ROAS only from a cited roi-calculator artifact, do not compute it here, and separate observed content anatomy from causal hypotheses. A renew/retest/retire call comes only from the Retro decision gate. Template: step 4.
Content performance analysis — compare formats/themes under compatible exposure and attribution bases. Name observed higher/lower associations; describe a hook/message/visual as causal or "winning" only when the supplied design clears the measurement protocol. Template: step 5.
Engagement quality analysis — break engagement by type/intent, run evidenced comment sentiment, and surface purchase-intent signals. Use typed observations or Unknown; emit no /10 quality score without a supplied rubric, inputs, and calculation. Template: step 6.
Conversion & attribution analysis — draw the observed funnel and use one declared attribution model. Deduplicate events into mutually exclusive counted buckets; preserve promo/UTM/direct overlap as reconciliation evidence, and report Estimated influence outside the counted total. Template: step 7.
Generate insights & recommendations — write 3–5 evidence-backed observations, separately labeled hypotheses, and bounded next tests. Add one compact Campaign Retro Card per creator decision requested only when a verified current, non-forked tracker-state artifact proves that exact campaign/creator is measured or closed and the matching measurement-contract and decision-rule refs are supplied; a bare stage string never qualifies. Use only renew | retest | retire | unknown. Template: step 8.

Before naming any creator/format/platform a real winner, clear the comparability, complete-scope, preregistered-rule, and significance bars in measurement-protocol.md — otherwise mark it Keep-testing or NOT_RANKED. When a structured score is needed, apply per-dimension STAR analysis (Suitability/Trust/Appeal/Return dimension reads) from star-benchmark.md, and hand financial inputs to roi-calculator for Return (R) math — this skill contributes inputs but does not compute ROI/ROAS or SQS (the creator-content-auditor gate computes SQS).

For the Retro Card, use renew only when comparable measured evidence clears the preregistered decision rule without a material unresolved limitation; use retest for a plausible but inconclusive or correctable test; use retire only when measured evidence or a documented hard constraint clears the declared stop rule; otherwise use unknown. This operating decision is not a STAR dimension, SQS, or creator-content-auditor verdict—do not simulate or carry forward one.

After an authorized WARM save, offer a handoff to campaign-planner to append the saved analysis/card reference to the relevant tracker row's evidence_refs; the tracker edit needs its own exact authorization, and neither the card nor this skill advances stage. Also offer fit-scorer as an explicit next-cycle handoff with the card's evidence references and hypothesis. Do not invoke it automatically, and do not translate the Retro decision into a STAR/SQS verdict.

Example

User: "Analyze this dated summer-skincare export for 10 creators. It contains opaque creator refs, the metric/target table below, per-creator and per-platform results, one deduplicated attribution model, and a completed significance read. ROI comes from roi-calculator artifact roi-ref-01."

Output (abridged — full version in references/analysis-templates.md):

# Summer Skincare Campaign Performance Analysis — illustrative export-backed read

| Metric | Result | Target | Status |
|--------|--------|--------|--------|
| Total Reach | 2.4M | 2M | ✅ +20% |
| Engagement Rate | 4.2% | 3.5% | ✅ +20% |
| Conversions | 1,847 | 2,000 | ⚠️ -8% |
| Revenue | $142,500 | $150,000 | ⚠️ -5% |
| ROAS (from `roi-ref-01`) | 2.8:1 | 3:1 | ⚠️ -7% |

**Top 3**: the three `creator_ref` rows that clear the declared ranking and significance rule, using only comparable metrics in the export.
**Key learning**: report the export-backed TikTok/Instagram delta only if the comparison windows and attribution bases match; otherwise mark it Keep-testing.
**Recommendation**: renew/drop and reallocation calls remain conditional on the predeclared decision rule rather than invented from the campaign count alone.

Reference Materials

references/analysis-templates.md — the eight fill-in step templates plus the full worked example.
skill-contract.md — shared contract and handoff format.
state-model.md — memory tiers and save-path conventions.
CONNECTORS.md — verified free/keyless data recipes per connector category.
measurement-protocol.md — preregistered readback windows, outcome unit, alpha, practical-effect boundary, multiplicity/sequential policy, guardrails, and decision owner. Report statistical and practical flags separately; use experiment.py for deterministic Calculated evidence, and never substitute a universal p-value/lift rule or attribute a business action to the helper.
The STAR benchmark at references/star-benchmark.md — scoring architecture when a structured score is needed.
Sibling skills: roi-calculator, report-generator, fit-scorer, campaign-planner.

Next Best Skill

Primary: roi-calculator — convert measured performance into dollar-level ROI, cost-per-result, and payback math.

Alternates (same Report family):

report-generator — package the analysis into a formal stakeholder report.
fit-scorer — feed proven performers back into creator scoring for the next round.

Termination note: Maintain a visited-set. If a skill has already been invoked this session, stop and report chain-complete rather than re-running it. Cap the chain at max-depth 3 hops; if results are inconclusive after that, surface the open loops to the user instead of continuing.