agentleFS
discontinuitythesisdiscontinuitythesis/google-ads-skills/skills/weekly-exception-note/SKILL.md

weekly-exception-note

Skill0 starsChanged 47 days agoMITView on GitHub

Write the weekly Google Ads exception note for a client or account owner, reporting only what moved outside its normal range, the likely cause, and what needs a decision. Use for Monday account checks, weekly PPC reports or when asked what changed in a Google Ads account last week.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask for what is missing, then work with what you have and say what you did not have.

    line 14
  2. Distinctive:

    Is a movement real demand, or attribution?: If conversions moved, did impressions and search impression share move with them?

    line 36
  3. Distinctive:

    Use exactly this structure.

    line 43
  4. Distinctive:

    Do not call a movement inside the conversion lag window a trend.

    line 70
  5. Distinctive:

    Do not change the account, even if a tool allows it.

    line 71
  6. Distinctive:

    Do not include personal data (names, emails, phone numbers) from the CRM in the note.

    line 72

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Weekly exception note

You compare last week with its baseline and write a short note that a business owner can read in two minutes. You report exceptions, not everything. You propose actions; you do not make them.

A note that says nothing moved is a valid and useful note. Send it anyway. Silence reads as inactivity.

Inputs

Ask for what is missing, then work with what you have and say what you did not have.

  1. Campaign weekly data (required): queries.gaql query 1 for the last 5 complete weeks (Monday to Sunday).
  2. Conversions by action (required): query 2 for the same weeks.
  3. Change history (recommended): query 3 for the last 14 days.
  4. CRM outcomes (recommended): leads by week with status (qualified, won, lost, open) and value, and the source of each (Google Ads by click ID or UTM).
  5. Ecommerce only: Merchant Center disapprovals or out-of-stock counts for the week, if the user has them.
  6. Known events (ask once): site releases, form or CRM changes, promotions, stock problems, bank holidays, anything the business changed.
  7. Conversion lag: days until about 90% of qualified leads are recorded. If search-term-review has been run, reuse its figure. Otherwise ask, or assume 30 days for lead generation and 7 for ecommerce and say so.

Baseline and thresholds

  • Baseline: the average of the 4 complete weeks before last week.
  • Flag a metric when last week differs from baseline by more than 20% and the absolute change matters (at least 5 conversions, or at least 10% of weekly spend, or at least one qualified lead where weekly qualified leads average under 10). Small accounts wobble; do not flag noise.
  • Offline conversions for the last lag days are incomplete. Compare them only against the same point in previous weeks' maturity, or leave them out and say why.

The questions to answer

Work through these in order. Each becomes a section of the note only if it produced an exception.

  1. Did lead or order quality move differently from platform CPA? Compare cost per platform conversion with cost per qualified lead (or per won deal, or gross profit per order) using matured weeks. The classic exception: CPA improved while qualified rate fell, which usually means cheaper, weaker leads or a conversion action counting something new.
  2. Did the conversion signal change rather than the business? A step change in one conversion action while others hold steady, a new action appearing, an action dropping to zero, or a change in conversions per click on one landing page. Cross-check with change history and known site releases. A tracking break is the most expensive thing this note can catch.
  3. Is a movement real demand, or attribution? If conversions moved, did impressions and search impression share move with them? Did CRM leads from all sources move the same way? A drop in Google-attributed conversions with flat CRM leads points at attribution or tracking, not demand.
  4. What is each campaign limited by? For campaigns that under-delivered: budget (search budget lost impression share up, spend at cap), rank (rank lost impression share up), demand (impressions down with impression share steady), inventory (ecommerce: disapprovals or stock). Say which.
  5. Did anyone change anything? List account changes from change history in the last 14 days that could explain an exception, by who or what made them (interface, API, Google automated). Remember a bidding change needs one or two conversion cycles before it can be judged.
  6. Is spend on track? Spend against budget for the month to date, if a monthly budget is known.

Output

Use exactly this structure. Plain language, no jargon the client would not use. Numbers with the previous number beside them.

# <Account name>: week of <Monday date>

**Summary:** <one or two sentences. If nothing moved: "Spend, qualified leads and conversion mix within normal range. No account changes made.">

## Exceptions
### <Short title of exception 1>
- **What moved:** <metric, last week vs baseline, change>
- **Evidence:** <the one or two figures that show it>
- **Likely cause:** <cause, with confidence: high, medium or low>
- **Proposed action:** <action, or "watch another week"> · **Needs approval:** yes/no

(repeat for each exception, most costly first, at most five)

## Changes made last week
<list, or "none">

## Not checked this week
<what data was missing, in one line>

Rules

  • At most five exceptions. If there are more, keep the five with the most money at stake and list the rest in one line.
  • Confidence is honest. "Low" is fine. Never present a guess as a cause.
  • Do not call a movement inside the conversion lag window a trend.
  • Do not change the account, even if a tool allows it.
  • Do not include personal data (names, emails, phone numbers) from the CRM in the note.

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.