Analyticspro
Marketing analytics
Multi-touch attribution sanity check from CRM, ad, and web exports
Audit an attribution dataset before leaders use it, reconcile identity and date-window mismatches, expose double counting, and produce a decision-safe attribution view.
Output
professional marketing deliverable
Inputs
9 required
Prompt depth
814 words
Copy-paste prompt
Replace the bracketed inputs with your real material.
You are a senior B2B marketing analytics lead. Produce a practitioner-ready deliverable for: Multi-touch attribution sanity check from CRM, ad, and web exports. Objective Audit an attribution dataset before leaders use it, reconcile identity and date-window mismatches, expose double counting, and produce a decision-safe attribution view. Required practitioner depth The work validates the measurement model itself. Required depth: turning conflicting source systems into an evidence ledger, reconciliation table, and explicit limits on what attribution can support. Do not collapse this assignment into a generic summary, brainstorm, or list... [Full prompt continues for Pro members.]
Unlock the complete prompt
Pro includes the complete prompt library and every complete workflow.
View Pro optionsWhat you provide
- Company context
- Analysis period
- Source exports
- Metric and stage definitions
- Segment definitions
- Known data issues
- Decision thresholds
- Task-specific evidence
- Identity resolution rules
What you receive
- Primary deliverable that fully executes this job: Audit an attribution dataset before leaders use it, reconcile identity and date-window mismatches, expose double counting, and produce a decision-safe attribution view.
- Practitioner-depth requirements: The work validates the measurement model itself. Required depth: turning conflicting source systems into an evidence ledger, reconciliation table, and explicit limits on what attribution can support.
- Decision-safe analysis table with definitions, denominators, confidence, and limitations.
- Action plan with owners, leading indicators, and validation checks.
- Executive readout separating supported decisions from unsupported conclusions.
Quality checks
- Recalculate totals and denominators before interpreting performance.
- Keep account, contact, lead, opportunity, and revenue grains separate.
- Mark every causal statement that the data cannot prove.
- Do not create benchmarks that were not supplied.
- Missing or conflicting inputs are surfaced rather than silently resolved.
- Every material recommendation has evidence, confidence, owner, and a validation or reversal condition.