agentleFS
search-atlas-groupsearch-atlas-group/amm-founding-circle/skills/agency-scorecard/SKILL.md

agency-scorecard

Skill0 starsChanged 5 months agoMITView on GitHub

Builds and maintains an EOS-style weekly Scorecard — 5 to 15 leading numbers that tell you Monday whether you'll have a problem Friday — pulled automatically from wherever those numbers already live, with off-track ones flagged before your weekly meeting. Use when your only view of the business is last month's P&L (too late to act on), when you want the Scorecard ready before your L10/weekly meeting instead of built during it, or when you don't know which 10 numbers actually predict your agency's health.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If a proposed number is really an outcome you'd only find out about after it's too late to change — it belongs in the monthly P&L review, …

    line 50
  2. Distinctive:

    Every Monday, the numbers are already sitting there when the meeting starts — nobody spent Sunday night pulling reports.

    line 101
  3. Distinctive:

    Leading over lagging, always: If a candidate metric is really a monthly outcome, say so and route it to the monthly review instead of the …

    line 113
  4. Very specific:

    Never silently redefine a goal range: after a bad week — that's a visible, dated decision, same discipline as quarterly-rocks-tracker's …

    line 117

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

agency-scorecard

The problem this solves: most agency owners only really look at the numbers once a month, when the P&L lands — and by then whatever went wrong already happened four weeks ago. A Scorecard fixes the timing, not just the visibility: a small set of leading numbers, checked every single week, that move before revenue does. Proposals sent, utilization, on-time deliverables, response time to new leads — numbers that tell you Friday's story on Monday, while there's still a week left to do something about it.

The hard part was never "know your numbers" — it's pulling the same 5–15 numbers from 5 different tools, every single week, without it becoming someone's part-time job. That's what this skill hands to the agent.

A Scorecard nobody has to manually assemble is a Scorecard that actually gets reviewed every week. One that takes an hour to build gets skipped the first busy Monday — and a skipped week is exactly the week something was quietly going wrong.


Say this to your agent

"Set up our weekly Scorecard. Here are the numbers that matter: <metric, where it lives, who owns it, the target/goal range>. Every Monday morning, pull the actual number for each from its source, compare it to the goal range, and give me the Scorecard with anything off-track called out — with the specific number, not just a red flag. Keep a running week-over-week view so I can see the trend, not just the snapshot."


Picking the right numbers (this is the part people get wrong)

A Scorecard with 40 metrics gets ignored; a Scorecard with 3 misses real problems. The EOS sweet spot is 5–15 numbers, and every single one should pass this test: if this number goes bad, does it predict a problem before the P&L shows it?

Good Scorecard number (leading)Bad Scorecard number (lagging or vanity)
Proposals sent this weekTotal revenue this month
New-lead response time (hours)Total pipeline value (ever)
Team utilization %Headcount
On-time deliverable rateFollower count / impressions
Client health scores below threshold (from client-health-score)Net margin (this is the output, not a leading signal)

If a proposed number is really an outcome you'd only find out about after it's too late to change — it belongs in the monthly P&L review, not the weekly Scorecard.


The pattern (set once, run every week)

  1. Define the 5–15 numbers, once. For each: the metric, exactly where it lives (which tool, which report, which field), who owns hitting it, and the goal range (a number or band, not "as high as possible").
  2. Weekly pull, same day, same time. The agent goes to each source and gets the actual current value — no manual copy-paste, no "I'll fill it in later."
  3. Compare against the goal range and flag. Anything outside its range gets called out explicitly, with the actual number next to the target, not just a color.
  4. Hand it to the meeting, don't run the meeting. The Scorecard is read for 5 minutes at the top of your weekly meeting — the agent's job ends at "here's this week's read," the owner's job is deciding what to do about anything red.

Weekly Scorecard format

## Scorecard — Week of <date>

| Metric | Owner | Goal | Actual | Status |
|---|---|---|---|---|
| New leads responded to <2hr | Sam | 90%+ | 74% | OFF |
| Proposals sent | Sam | 6+ | 8 | ON |
| Team utilization | Priya | 75-85% | 91% | OFF (over-utilized) |
| On-time deliverables | Priya | 95%+ | 97% | ON |
| Clients below health threshold | Jordan | 0-1 | 2 | AT-RISK |

Flagged for discussion: <the OFF/AT-RISK rows, one line each on why>

What goes on the Scorecard vs. what doesn't

Belongs on the weekly ScorecardDoesn't belong here
Numbers you'd want to know about this week, not next monthAnything you only check quarterly (that's a Rocks or planning input)
A named owner per metricMetrics with no clear owner — assign one before adding it
A real goal range, set in advanceA target invented after seeing the actual number
The trend (up/down vs. last week)Raw historical data dumps — link out to the source instead

What a good result looks like

  • Every Monday, the numbers are already sitting there when the meeting starts — nobody spent Sunday night pulling reports.
  • A problem shows up as an off-track number in week 2, not as a bad month in week 6.
  • The meeting spends its time on the 2–3 flagged rows, not re-deriving all 15 numbers from scratch.
  • Trust in the Scorecard grows because the numbers are always pulled the same way, from the same sources — no quiet redefinition when a metric looks bad.

The rules it runs under

  1. Leading over lagging, always. If a candidate metric is really a monthly outcome, say so and route it to the monthly review instead of the weekly Scorecard.
  2. One owner per number. An unowned metric gets flagged for an owner before it's added, not tracked anonymously.
  3. Never silently redefine a goal range after a bad week — that's a visible, dated decision, same discipline as quarterly-rocks-tracker's no-retroactive-rewrite rule.
  4. Composes with client-health-score (client-risk numbers can be a Scorecard row) and quarterly-rocks-tracker (Rocks are the 90-day plan; the Scorecard is this week's pulse against it).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.