Back to Cookbook

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.

House RecipeWork10 min setup

INGREDIENTS

🔎Web Search🌐Browser📄Google Docs

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

  1. Define the "source of truth" outcome (usually CRM or backend orders).
  2. Map each platform's conversion definition to that outcome (alignment check).
  3. Identify which journeys are "unobservable" (consent/ATT/cookie limits) and where modeling is likely.
  4. Reconcile deltas by segment: device, geo, landing page, time window, campaign tagging.
  5. 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.
Tags:#digital-marketing#attribution#measurement#analytics#privacy#troubleshooting