Attribution Gap Triage
Diagnose missing conversions and "flying blind" measurement fast
Use this when the numbers don't match: ad platforms over/under-report, GA4 looks off, CRM revenue doesn't reconcile, or privacy changes (ATT/cookie loss/consent) have degraded tracking. It produces a root-cause shortlist, a "what to trust" guidance note, and a prioritized fix plan.
INGREDIENTS
PROMPT
You are an OpenClaw agent helping a digital marketer triage an attribution/measurement incident. Ask only for the minimum inputs needed, then: - Define the primary source-of-truth outcome. - List the top likely causes (distinguish privacy/mechanics vs implementation). - Provide a prioritized fix plan (today/this week/this month). - End with a short "what we can and cannot conclude" note. Inputs (fill in): - Business model + primary conversion: - Channels involved: - Tools (GA4, CRM, ecommerce, tag manager): - What changed recently: - What looks wrong (numbers, date range, screenshots if available):
How It Works
This recipe runs a structured attribution incident response: verify whether you have a measurement mechanics
issue (privacy limits, modeling, consent, dedupe rules) vs an implementation issue (UTMs/events/pixels/server-side).
Triggers
- ROAS swings sharply with no clear campaign change
- CRM revenue/leads don't reconcile with ad platform conversions
- You suspect iOS/consent/cookie changes are degrading signal
Inputs
- Channels involved (paid search, paid social, email, organic)
- Conversion definitions (lead, MQL, SQL, purchase) and timestamps
- Current UTM rules + naming conventions (if any)
- Any recent tracking changes (tag manager, pixel, consent banner, server-side)
Outputs
- "What to trust right now" measurement memo (per channel)
- Root-cause hypothesis list (top 3–7)
- Fix plan (quick wins + longer-term infrastructure)
Actions / Steps
- Define the "source of truth" outcome (usually CRM or backend orders).
- Map each platform's conversion definition to that outcome (alignment check).
- Identify which journeys are "unobservable" (consent/ATT/cookie limits) and where modeling is likely.
- Reconcile deltas by segment: device, geo, landing page, time window, campaign tagging.
- Propose fixes: taxonomy, server-side, dedupe rules, improved lead routing, MMM/incrementality fallback.
Parameters
- Reconciliation tolerance (e.g., ±10–20% by channel)
- Lookback window (7/28/90 days)
- Priority outcome (revenue, SQLs, trials, purchases)
Examples
- "Meta shows 200 purchases, Shopify shows 140. Where are the missing 60?"
- "GA4 says paid search is down 30% but spend is flat and CRM leads are up."
Tips
- Always separate "tracking is broken" from "tracking is limited by design."
- Produce a client/executive-safe narrative that explains uncertainty and next steps.