Sign in Start free
ANALYTICS & REPORTING

Metric definition reconciliation

Use when two tools report different numbers for the same thing and people are arguing about it.

metric-definition-reconciliation.md
Download .md 
Two sources disagree about the same metric. Explain the gap before anyone picks a favorite.

Metric: {{METRIC_NAME}}
Source A: {{SOURCE_A}} reports {{VALUE_A}}
Source B: {{SOURCE_B}} reports {{VALUE_B}}
Period: {{PERIOD}}
Setup details: {{TRACKING_SETUP}}

Output:
1. Definition table: Source | How this source defines the metric | Counting unit | De-duplication rule | Time zone | Attribution or session rule.
2. Ranked list of likely causes of the gap, largest suspected contribution first, each with the rough size of the effect and how to test it.
3. Which parts of the gap are expected and permanent, and which indicate a genuine setup error.
4. Which source to use for which question. Neither is "correct" for everything.
5. A one-paragraph explanation for a non-technical stakeholder that does not blame either tool.

Rules:
- A gap under {{ACCEPTABLE_VARIANCE}} percent from known definitional differences is normal. Say so plainly rather than hunting for a bug.
- Do not guess at a cause you cannot describe a test for.
- If you need a detail about the setup that I have not given, list it as a question rather than assuming.
- No em dashes.

Fill in before running

Replace each placeholder with your own detail. The more specific you are, the less the model invents.

  • {{METRIC_NAME}}
  • {{SOURCE_A}}
  • {{VALUE_A}}
  • {{SOURCE_B}}
  • {{VALUE_B}}
  • {{PERIOD}}
  • {{TRACKING_SETUP}}
  • {{ACCEPTABLE_VARIANCE}}

Getting a better result

  1. Time zone and de-duplication explain most gaps; check those before touching the tracking code.
  2. Write section 4 into your reporting notes so this argument only happens once.
  3. Ask for the same reconciliation on a single high-traffic day to isolate the effect.

Questions about this prompt

When do I use this rather than picking the tool I trust?

When the gap has become an argument about which tool is broken. Choosing a favourite ends the conversation without answering it, and both numbers usually describe something real. Use it before anyone opens the tracking code, because most gaps are definitional rather than a defect anybody introduced.

What do I need in front of me?

The metric name, both sources with their figures, the period, your tracking setup, and an acceptable variance percentage below which you stop looking. Set that threshold before you see the answer. Without it every gap looks like a bug and someone spends a week on a difference that was always going to exist.

What comes back?

A definition table covering counting unit, de-duplication, time zone and session or attribution rule per source, ranked causes with rough sizes and a test for each, which parts of the gap are permanent, which source answers which question, and a stakeholder paragraph blaming neither tool. Section four is what goes into your reporting notes.

What is the mistake that costs me here?

Going to the tracking code first. Time zone and de-duplication explain most gaps and cost nothing to check, while a tag change made in haste adds a second discontinuity on top of the first. Where it asks for a setup detail you did not supply, answer it rather than letting it assume.