Data — is it sound?
Brief 90

Metric Definition Consistency

Whether the numbers the product reports mean the same thing everywhere — so active user, revenue, and signup are not defined three different ways.

METRICS.md4 phases7 lenses~2.7k chars

Free & open · no signup · read-only — it ends by asking · nothing leaves your machine

What it does

You are working inside this repo. Mission: check that the metrics the product computes and reports are defined once and consistently — that the same word means the same calculation in the app, the analytics, and the dashboards.

Read-only pass. Read the queries, event definitions, and reporting code; change nothing. Your only write is the report file.

What it looks for

7 lenses it looks through

Phase 2 sweeps the codebase through every one of these, citing file and line for each finding.

1

Duplicate definitions

the same metric computed differently across code, analytics, and dashboards

2

Ambiguous terms

"active," "churned," "converted" with no single written definition

3

Source-of-truth drift

app database, warehouse, and analytics tool disagreeing on the same count

4

Window boundaries

day/week/month cutoffs and timezones that differ across reports

5

Filter & dedup logic

inconsistent handling of test accounts, bots, refunds, internal traffic

6

Event coverage

metrics built on events that fire inconsistently or are missing on some paths

7

Documentation

whether definitions are written down or live only in memory

The deliverable

What lands in your repo

One structured report at the repo root — or in reports/, if you keep one — the same shape every time, ready for a teammate — or the optional Studio — to act on.

METRICS.md
1
Metric dictionary
term · canonical definition · source of truth · owner
2
Conflicts
each: metric · the differing definitions · where each lives · the impact
3
Canonical choices
the definition to standardize on, and why
4
Consolidation plan
the duplicate computations to retire or align
One file. Evidence-backed. It ends by asking before touching anything.
How it works

What it makes the agent do

Every Goal Prompt follows the same four steps, so results are consistent and repeatable — no matter which one you run.

1
Phase 1

Collect the definitions

List the metrics that drive decisions.

2
Phase 2

Audit through 7 lenses

Cite each definition's location for every finding.

3
Phase 3

Curate

Rank by decision weight.

4
Phase 4

Report

Create METRICS.md at repo root.

Get started

Three ways to run this Goal Prompt

01 · COPY

Paste it in

Copy it and paste it into your agent inside the repo you want checked.

02 · INSTALL

As a slash command

Install the goal plugin once — two commands — then just type /goal:metric-definition-consistency.

1/plugin marketplace add GhostlyGawd/goal-prompts
2/plugin install goal@goal-prompts

Or install only this Goal Prompt as /goal-metric-definition-consistency:

curl -fsSL https://goal-prompts.vercel.app/install | BRIEF=90 sh
03 · AGENT

From an agent (MCP)

Let an agent fetch it mid-conversation, or pull the raw Goal Prompt by URL.

https://goal-prompts.vercel.app/raw/90.md
Transparency

The exact prompt

Nothing hidden — this is the whole Goal Prompt, verbatim. Read it in a minute, edit it, or copy it as-is.

Read the full brief (2,722 characters)
# Goal: Metric Definition Consistency

You are working inside this repo. Mission: check that the metrics the product computes and reports are defined once and consistently — that the same word means the same calculation in the app, the analytics, and the dashboards.

Read-only pass. Read the queries, event definitions, and reporting code; change nothing. Your only write is the report file.

## Phase 1 — Collect the definitions
- List the metrics that drive decisions: active users, revenue, conversion, retention, churn, signups.
- For each, find every place it is computed — app code, SQL, analytics events, dashboards.
- Note where the number is shown to users, leadership, or investors.

## Phase 2 — Audit through 7 lenses
Cite each definition's location for every finding.
1. **Duplicate definitions** — the same metric computed differently across code, analytics, and dashboards
2. **Ambiguous terms** — "active," "churned," "converted" with no single written definition
3. **Source-of-truth drift** — app database, warehouse, and analytics tool disagreeing on the same count
4. **Window boundaries** — day/week/month cutoffs and timezones that differ across reports
5. **Filter & dedup logic** — inconsistent handling of test accounts, bots, refunds, internal traffic
6. **Event coverage** — metrics built on events that fire inconsistently or are missing on some paths
7. **Documentation** — whether definitions are written down or live only in memory

## Phase 3 — Curate
- Rank by decision weight: a metric on the board deck outranks one on a debug page.
- For each conflict, propose the canonical definition and the source of truth.
- Separate a genuine disagreement from two names for the same thing.

## Phase 4 — Report
Create `METRICS.md` at repo root:
1. **Metric dictionary** — term · canonical definition · source of truth · owner
2. **Conflicts** — each: metric · the differing definitions · where each lives · the impact
3. **Canonical choices** — the definition to standardize on, and why
4. **Consolidation plan** — the duplicate computations to retire or align

Start the report with today's date. If `METRICS.md` already exists from a previous run, read it first and lead with what changed since.

## Rules
- One metric, one definition, one source of truth
- A number shown to a decision-maker with no written definition is a liability
- No computed metrics in this repo? Say so in a one-paragraph null report and stop — a null result is a valid finding.
- If a `reports/` directory exists at the repo root, write the report there instead of the root.
- Before asking, present the top findings as a ranked list in plain words
- Report only — end by asking which definitions to standardize first

House rules for this Goal Prompt

  • One metric, one definition, one source of truth
  • A number shown to a decision-maker with no written definition is a liability
  • No computed metrics in this repo? Say so in a one-paragraph null report and stop — a null result is a valid finding.
  • If a reports/ directory exists at the repo root, write the report there instead of the root.
  • Before asking, present the top findings as a ranked list in plain words
  • Report only — end by asking which definitions to standardize first
Keep exploring

More Data briefs