agentsclimarketplace
Sign in

Weekly account auditor

Skill kelpi-ai/google-ads-skills/skills/weekly-account-auditor

Compare two complete Google Ads periods, explain the few changes that matter, and rank the next checks without changing the account. Use for a weekly review or when performance suddenly feels off.From its SKILL.md

Install
npx -y skills add kelpi-ai/google-ads-skills --skill weekly-account-auditor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 28 days oldThe repository was created 28 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.0 KB, 569 tokens by cl100k_base, as published. Nobody here has run it

Weekly Account Auditor

Doctrine

A useful account review explains movement, not just totals. Compare equal, complete periods in the account timezone, find the campaigns or ad groups driving the change, and put tracking gaps ahead of confident conclusions. Recent conversions may still arrive, so a bad-looking week is not final until conversion delay is accounted for.

When to use

  • Run once a week with a current period and an equal comparison period.
  • Use a connected Google Ads MCP, or paste exports with campaign metrics and date ranges.
  • Bring the business target, primary conversion definition, currency, timezone, and known conversion delay when you have them.

Run it

Run a READ-ONLY weekly Google Ads audit. Use live account data if connected; otherwise use the exports I paste. Change nothing.

1. Confirm the account, currency, timezone, primary conversion, target CPA or ROAS, and two equal complete date ranges. If any are missing, mark the gap before interpreting performance.
2. Compare spend, impressions, clicks, conversions, conversion value, CPA, and ROAS only where the required inputs exist. Show the formula behind every calculated metric.
3. Find the campaign or ad-group drivers behind the largest useful changes. Use entity names and IDs, current value, comparison value, and absolute change.
4. Check conversion delay, tracking interruptions, partial days, and major account changes before calling a trend real. Never estimate missing conversions from the length of an outage.
5. Return at most 5 findings, ranked by business impact and confidence. Separate observed facts, likely explanations, and next checks.

End with a short action list labeled: investigate, keep watching, or ready for approval. Do not pause, scale, or edit anything. Say "No changes were made."

Guardrails

  • Equal complete periods only. Do not compare a partial day with a finished day.
  • A tracking outage or unresolved conversion delay lowers confidence. It does not justify pausing campaigns or scaling winners.
  • Do not turn hours of missing tracking into an estimated conversion count or "corrected" CPA. The missing number is unknown until it is reconciled.
  • Do not invent revenue, profit, attribution, or a target the user did not provide.
  • Never recommend a fixed percentage budget move from summary metrics alone.
  • A low CPA does not prove a campaign has room to scale. Ask for impression share, budget limits, query coverage, and marginal performance before proposing more spend.
  • Search-term reports can omit low-activity queries. State the coverage limit when search-term evidence is used.

Good output looks like

A five-minute review with the real movement, the entities driving it, the caveats that could change the verdict, and a short next-check list. No dashboard dump and no account changes.

What ships with it: 1 file

554 B alongside SKILL.md

Gives 0 of the 12 instructions most audit compliance skills give in 569 tokens

Counted across 937 of the 1,487 authors here whose files we hold, read 2026-08-07

  • Fetch latest guidelines before each reviewin 43 of 937, across 3 files
  • Group findings by severityin 43 of 937
  • Check files against all fetched rulesin 42 of 937, across 2 files
  • Output findings in terse file:line formatin 41 of 937, across 3 files
  • Ask user which files to review if none specifiedin 41 of 937, across 3 files
  • Read specified files or prompt user for filesin 39 of 937, across 1 file
  • Generate the audit reportin 33 of 937, across 30 files
  • Assign a severity to every findingin 25 of 937
  • Run automated accessibility scansin 23 of 937, across 13 files
  • Output a markdown audit reportin 22 of 937
  • Map findings to WCAG criteriain 20 of 937, across 10 files
  • Confirm audit scopein 19 of 937, across 9 files

Said here and by no other author read

  • Confirm account settings, targets, and two equal date ranges
  • Mark any missing settings before interpreting performance
  • Compare spend, clicks, conversions, and calculated metrics
  • Show the formula behind every calculated metric
  • Identify the campaigns driving the largest changes
  • Check conversion delay and tracking interruptions before concluding

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,790. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.