agentleFS
kelpi-aikelpi-ai/google-ads-skills/skills/delivery-troubleshooter/SKILL.md

delivery-troubleshooter

Skill2 starsChanged 4 months agoMITView on GitHub

Find why Google Ads delivery or reporting changed, separate observed blockers from possible causes, and put the checks in the right order. Use when impressions, clicks, spend, or conversions suddenly fall.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 5 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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:

    When impressions, clicks, spend, or recorded conversions drop without a clear reason.

    line 14
  2. Distinctive:

    Use a connected Google Ads MCP, or paste status, policy, delivery, budget, impression-share, and conversion-action exports.

    line 15
  3. Distinctive:

    Never turn correlation into root cause.

    line 35
  4. Distinctive:

    Do not call lost impression share from rank a Quality Score penalty.

    line 36
  5. Distinctive:

    Do not claim the campaign was healthy before the incident unless the earlier data is provided.

    line 39

≈ “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

Delivery Troubleshooter

Doctrine

Delivery problems leave clues at different levels: account status, campaign status reasons, disapprovals, budgets, rank, destinations, bid-strategy learning, and conversion actions. Do not flatten those clues into one confident story. Name what is observed, what is only suspected, and which dependency must be fixed before the next diagnosis means anything.

When to use

  • When impressions, clicks, spend, or recorded conversions drop without a clear reason.
  • Use a connected Google Ads MCP, or paste status, policy, delivery, budget, impression-share, and conversion-action exports.
  • Include dates, entity names and IDs, recent changes, and site incidents when available.

Run it

Run a READ-ONLY Google Ads delivery check. Use live account data if connected; otherwise use what I paste. Change nothing.

1. State the scope, date range, expected baseline, and what changed.
2. Check account and billing status, campaign and ad-group primary status reasons, ad and keyword approvals, landing-page availability, bid-strategy status, budget limits, lost impression share from budget and rank, and conversion-action status.
3. Return an issue table with: entity, observed status, evidence, likely impact, suspected causes, confidence, and next check.
4. Keep observed facts separate from suspected causes. Lost impression share from rank is not proof of a Quality Score problem. A site incident is not proof it caused a disapproval unless the policy reason and timing support it.
5. Put fixes in dependency order. Restore broken destinations before requesting ad review; validate tracking before judging conversion performance; resolve serving blockers before tuning bids or budgets.
6. List healthy checks and unknowns so I can see what has already been ruled out.

Do not change budgets, bids, ads, destinations, goals, or statuses. End with "No changes were made."

Guardrails

  • Never turn correlation into root cause. Use "possible cause" until the account evidence supports it.
  • Do not call lost impression share from rank a Quality Score penalty. Rank can reflect bids, ad quality, assets, competition, and auction context.
  • A small nonzero budget loss is still a fact. Report it without calling it noise or automatically recommending more budget.
  • Google Ads data can show suspicious conversion reporting, but it cannot prove a website tag fires or reconcile a CRM without separate evidence.
  • Do not claim the campaign was healthy before the incident unless the earlier data is provided.

Good output looks like

A short, dependency-aware incident report: the real blockers first, suspected causes labeled honestly, the checks that come next, and no premature budget or bid change.

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.