evaluate-attribution-models — Claude Code command overview
evaluate-attribution-models compares marketing attribution models across KPIs, outlines trade-offs, and generates a rollout plan tailored to specific stakeholders.
Apache-2.0 ★ 450 updated 2026-04-03
.claude/commands/evaluate-attribution-models.mdWhat evaluate-attribution-models does
This command benchmarks campaign outcomes across several attribution frameworks: first-touch, last-touch, linear, position-based, time-decay, and data-driven. Inputs specify target campaigns, selected KPIs like CAC, payback, or ROAS, and the intended audience, including finance, marketing leads, ops, or executives.
The workflow executes across five stages: data normalization, weighting application, sensitivity benchmarking, narrative drafting, and recommendations. It orchestrates the attribution-architect and marketing-intelligence-lead agents, using rules defined in attribution-playbook and exec-dashboard-blueprint. Without historical cost and pipeline figures, sensitivity analysis cannot run.
Deliverables include a table detailing KPI variances across models, a memo justifying the recommended baseline, and an operational checklist with milestone owners.
What it helps with
- compare revenue attribution across multiple models
- build a CAC and payback comparison memo for finance leads
- generate a phased rollout plan with monitoring hooks for data-driven models
- isolate variance drivers across running marketing programs
Example prompts for the agent
- /marketing-analytics:evaluate-attribution-models --campaigns launch_q1,evergreen_abm --models first,last,position,data-driven --audience finance
- Evaluate attribution trade-offs between first and position-based models for executive review.
- Compare linear and data-driven attribution for
evergreen_abmfocused on payback period.
How to install evaluate-attribution-models
Manually: put the files in place
.claude/commands/evaluate-attribution-models.mdCall it with /evaluate-attribution-models. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/evaluate-attribution-models.mdCall it with /prompts:evaluate-attribution-models. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/evaluate-attribution-models.prompt.mdManually: put the files in place
.gemini/commands/evaluate-attribution-models.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about evaluate-attribution-models
How to install evaluate-attribution-models in Claude Code?
Save the command configuration to .claude/commands/evaluate-attribution-models.md in your project repository.
Which attribution models are supported?
The command evaluates first, last, linear, position, time-decay, and data-driven attribution methods.
Which subagents and skills does the workflow trigger?
It invokes attribution-architect and marketing-intelligence-lead agents alongside attribution-playbook and exec-dashboard-blueprint skills.
Source text
evaluate-attribution-models.md · Apache-2.0
--- name: evaluate-attribution-models description: Compares attribution models, highlights trade-offs, and recommends rollout plans. usage: /marketing-analytics:evaluate-attribution-models --campaigns launch_q1,evergreen_abm --models first,last,position,data-driven --audience finance --- # Command: evaluate-attribution-models ## Inputs - **campaigns** – list of campaigns/programs to analyze. - **models** – attribution models to compare (first, last, linear, position, time-decay, data-driven). - **audience** – finance | marketing-lead | ops | exec. - **metrics** – choose KPIs (pipeline, revenue, CAC, payback, LTV, ROAS). - **confidence** – optional minimum data confidence threshold to highlight gaps. ## Workflow 1. **Data Preparation** – pull campaign performance, cost, and pipeline/revenue outcomes. 2. **Model Execution** – run requested models, normalize windows, and apply weighting rules. 3. **Sensitivity Analysis** – compare outcomes vs benchmarks, highlight variance drivers. 4. **Narrative Assembly** – contextualize trade-offs, governance considerations, and risks. 5. **Recommendation Engine** – propose primary model, fallback, and rollout/QA checklist. ## Outputs - Attribution comparison deck/table with KPI deltas per model. - Recommendation memo with decision, rationale, and risk mitigations. - Rollout plan including QA steps, owner assignments, and monitoring hooks. ## Agent/Skill Invocations - `attribution-architect` – leads methodology comparison. - `marketing-intelligence-lead` – ensures narrative + stakeholder alignment. - `attribution-playbook` skill – documents rules + templates. - `exec-dashboard-blueprint` skill – packages executive summary. ---
© gtmagents · Apache-2.0 · gtmagents/gtm-agents