Cross-discipline (paid ads): this is the shared return-math engine for paid ads — paid-measurement-loop, attribution-reconciler, and budget-optimizer delegate ROAS/CPA/payback ratios here rather than recomputing them. Save paid runs under

memory/ad/roi-calculator/

.

• Reads: campaign spend breakdown, results data (reach, impressions, engagements, clicks, conversions, revenue, new customers), AOV and repeat-rate data if LTV is in scope, any prior performance output from

  performance-analyzer

  .
• Writes: ROI calculation file at

  memory/influencer/roi-calculator/YYYY-MM-DD-<topic>.md

  containing direct ROI/ROAS, EMV, cost-efficiency metrics, attribution-modeled revenue, LTV-based ROI, by-influencer ROI, and a summary report block.
• Promotes: only with separate authorization, durable headline numbers with their attribution window, source, and uncertainty; a calculation request alone does not authorize hot-cache writes.
• Done when:
  1. At least one ROI methodology is computed with the inputs and formula shown.
  2. Each headline metric is stated against a declared, source-dated comparison target; no universal benchmark is invented.
  3. A bottom-line assessment (profitable / break-even / loss) and 1-3 recommendations are written.
• Primary next skill: report-generator

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

• ~~social platform analytics

  — reach, impressions, engagements, video views per platform for EMV and cost-per-metric math.

• ~~ecommerce / analytics

  — revenue, conversions, link clicks, and AOV for direct ROI and attribution.

• ~~CRM

  — new-customer counts, repeat-purchase rate, and lifetime value for LTV-based ROI.

• ~~influencer database

  — per-influencer fees and tier data for by-influencer ROI.

1. Gather ROI inputs — campaign details, the investment (total spend) table, and the results-data table. (template)
2. Calculate direct ROI — Simple ROI = (Revenue − Investment) / Investment × 100; ROAS = Revenue / Investment. State profit and a Profitable/Break-even/Loss assessment. (template)
3. Calculate Earned Media Value (EMV) — impression-based (Impressions × CPM / 1000) and engagement-based (Engagements × CPE), then average. Flag EMV as directional, not absolute. (template)
4. Calculate cost-efficiency metrics — CPM, CPR, CPE, CPV, CPC, CPA, and CAC. Compare only against a declared, source-dated target with a compatible market, window, and attribution basis; otherwise report the metric descriptively and mark the comparison pending. (template)
5. Apply attribution modeling — run first-touch, last-touch, linear, time-decay, and position-based; recommend the model that fits the customer journey. (template)
6. Calculate customer lifetime value impact — LTV-Based ROI = (New Customers × Avg LTV − Investment) / Investment; project short- vs. long-term and compare customer quality to organic/paid. (template)
7. Calculate by-influencer ROI — per-influencer ROI/ROAS rank, investment efficiency, and ROI by tier (macro/micro/nano). (template)
8. Generate the ROI report summary — investment, returns, ROI by methodology, key metrics vs. benchmark, bottom line, and 1-3 recommendations. (template)
9. Produce the measured Return (R) evidence for the gate
   The financial outputs from steps 1–8 are the campaign's measured Return (R) evidence for STAR: ROI/ROAS read against the declared target (

   R1

   ) and the alternative-channel baseline (

   R3

   ), CPE/CPM/CPA benchmarked on a normalized window (

   R2

   ), KPI attainment versus the pre-registered target (

   R4

   ), conversions attributed with a stated method and rigor (

   R5

   ), and incremental impact separated from baseline where measurable (

   R6

   ). Measured Return exists only at

   assessment_time: actual

   ; a forecast read has no

   R1

   –

   R6

   . Label each figure Measured / User-provided / Calculated / Estimated.
   Hand this Return evidence to the creator-content-auditor gate — it folds R into the full actual STAR run and computes the profile-weighted SQS. This skill does not run the scorer or emit the composite. Unverified conversions emit

   results-unverified

   : report

   R1

   /

   R2

   /

   R5

   as low-confidence and make no attributable-return claims. These financial numbers are consumed as R evidence; they are not themselves an SQS.
   For a multi-creator campaign, the gate scores each creator partnership separately; a budget-weighted mean of the per-partnership SQS values may summarize the campaign but never replaces the per-partnership diagnosis. This skill supplies the per-partnership Return evidence; it does not aggregate or roll up a composite.
10. Persist only with permission — save under

    memory/influencer/roi-calculator/

    (or the paid path) only after authorization; request separate authorization for hot-cache promotion.

• Profit: $47,000
• ROI: 188%
• ROAS: 2.88:1

• EMV (directional scenario at a declared $8 CPM): $16,800
• EMV Multiple: 0.67x

• CPM: $11.90
• CPA: Unknown (conversion count was not supplied)

• references/roi-templates.md — fill-in templates for every Instructions step, the worked example, and benchmark evidence inputs.
• measurement-protocol.md — read ROI and Return (R) deltas against a control over the readback window; do not over-claim attribution.
• skill-contract.md — shared contract and Handoff Summary format.
• state-model.md — memory tiers and save-path conventions.
• CONNECTORS.md — free/keyless data recipe per connector category.
• STAR scoring: star-benchmark.md — the Return (R) dimension this skill's evidence feeds and the profile-weighted SQS the gate computes.
• performance-analyzer — supplies the results data this skill consumes.
• report-generator — wraps these numbers into a full report.
• budget-optimizer — uses ROI output to reallocate spend.
• campaign-planner — sets the ROI targets these results are checked against.

• performance-analyzer — go back for deeper performance breakdowns if the ROI math exposed gaps.
• budget-optimizer — feed by-influencer and by-tier ROI into the next budget allocation.