google-ads-audit

Files (6)

• SKILL.md
• evals.json · evals/
• account-health-scoring.md · references/
• business-context.md · references/
• conversion-network-audit.md · references/
• persona-discovery.md · references/

SKILL.md (first lines)

# Google Ads Audit

Diagnose account health and persist business context for downstream skills (`/google-ads`, `/google-ads-copy`, `/google-ads-landing`). **Read-only** — never mutates the account. The user runs `/google-ads` to execute fixes you recommend.

## Setup

Follow `../shared/preamble.md` (MCP detection, account selection) and `../shared/analysis-principles.md` (evidence requirement, guardrails). Both apply throughout this skill.

## Filesystem contract (must persist)

| Artifact | Path | When |
|---|---|---|
| Business context | `{data_dir}/business-context.json` | First full audit, or refresh when `audit_date` is >90 days old. Skip on scoped audits if file is fresh. |
| Personas | `{data_dir}/personas/{accountId}.json` | Every full audit. |

These are the handoff to every other ads skill — write them even if the report is short. Otherwise `/google-ads-copy` and `/google-ads-landing` operate without business context and produce generic output.

**business-context.json schema:** `business_name, industry, website, services[], locations[], target_audience, brand_voice{tone, words_to_use[], words_to_avoid[]}, differentiators[], competitors[], seasonality{peak_months[], slow_months[], seasonal_hooks[]}, keyword_landscape{high_intent_terms[], competitive_terms[], long_tail_opportunities[]}, social_proof[], offers_or_promotions[], landing_pages{}, unit_economics{aov_usd, profit_margin, source}, notes, audit_date, account_id`.

**personas JSON schema:** `{account_id, saved_at, personas: [{name, demographics, primary_goal, pain_points[], search_terms[], decision_trigger, value}]}`. See `references/persona-discovery.md`.

## Policy freshness check (run first)

Read `../shared/policy-registry.json`. For each entry where `last_verified + stale_after_days < today`:
- Any entry without a direct current first-party Google source is a hypothesis, not an audit rule or benchmark. Do not use it for a finding or recommendation without verification.
- **High-volatility** → search the official Google Ads Help, Ads & Commerce blog, or Google Ads developer documentation for the `category`; compare the source with the recorded `rule`. If it drifted, omit the stale rule and banner the limitation.
- **Moderate-volatility** → verify it when it could affect a material finding; otherwise omit it rather than repeating a stale caveat.
- **Stable** → skip silently.

## Phase 1 — Pull the audit dataset

Choose available read capabilities for the requested audit scope. Batch related reads where useful and supported; consult current server guidance for schemas and limits.

You decide the exact GAQL shape, but a defensible audit needs to see, at minimum:

- Account-level rollups (`customer`)
- Campaign performance with bidding strategy, network, and impression-share metrics (`campaign`, 90-day cap for impression-share data)
- Ad-group performance (`ad_group`)
- Keyword performance with Quality Score and components (`keyword_view`)
- Search terms (`search_term_view`)
- Negative keywords and shared lists (`campaign_criterion` + shared sets)
- Conversion actions (`conversion_action`) — including counting type, attribution model, primary/secondary
- Network segmentation (`segments.ad_network_type`) when diagnosing CPA/CVR shifts or Search Partners
- RSA assets (`ad_group_ad`)
- Geo targeting (`campaign_criterion` LOCATION + PROXIMITY)
- Recent change events (`change_event`, last 30 days) — for explaining regressions

Aggregate inside the script. Return summarized JSON, not raw rows. The agent narrates; the script does the math.

Use platform recommendations or account-setup diagnostics as optional cross-checks when available and relevant to the question.

If a read fails, follow actionable recovery guidance. Clearly report missing evidence; continue independent findings only when the available data supports them.

**Skip scoring entirely if** `totalSpend == 0` or `activeCampaigns == 0`. Go straight to business context.

## Phase 2 — Scope handling

If the user narrows the audit ("focus on one campaign", "campaign X", "just check waste"):

- Match campaign names by case-insensitive substring. If no match, list available campaigns and ask.
- Filter the in-memory dataset before analysis — no extra API calls.
- Account-level dimensions (conversion tracking, account guardrails) stay account-wide. Note "Scoped to: X" in the report.
- Skip Phase 4 (business context refresh) on scoped audits if `business-context.json` is fresh.

## Phase 3 — Diagnose

The audit's headline output is **three pulse metrics** — Waste ($/mo), Demand captured (%), CPA ($) — each annotated with its top contributor and a pointer to the fix. Read `references/account-health-scoring.md` for the formula, annotation rules, signal-failure overrides, and `audit-history.json` schema. The pulse metric IS the verdict; you don't add a letter grade or 0–5 score on top.

To compute and back the pulse metrics, you'll need to look across these seven areas. They are diagnostic surface area, not graded dimensions:

1. **Signal Quality** *(account-level)* — measurement integrity. If broken, **STOP** here and recommend pausing spend until it's fixed. Pulse metrics are meaningless without measurement (apply the signal-failure override on the Waste line per the reference).
2. **Campaign Structure** — keywords per ad group, brand vs. non-brand separation, channel mixing, naming, budget logic.
3. **Keyword Health** — Quality Score weighted by spend, zombie keywords, match-type discipline.
4. **Search-Term Quality** — wasted spend, brand-leakage, negative coverage, conversion-worthy terms not yet keywords.
5. **Ad Copy & Creative** — RSA coverage, asset variety, sitelink/callout/structured-snippet completeness, PMax asset-group health.
6. **Impression Share** — read rank-lost vs budget-lost together (see the 2×2 matrix in `account-health-scoring.md`); they're different problems with different fixes.
7. **Spend Efficiency** — waste vs. headroom, brand vs. non-brand split, concentration risk.

For Signal Quality and network-mix questions, read `references/conversion-network-audit.md`. It adds the prerequisite checks for conversion-action integrity, Search Partners, Display leakage in Search campaigns, and regression decomposition.

Per-area findings only show up in the report when the area surfaced something material. Cite specific entities, dollars, and time windows. "Some keywords are underperforming" is not a finding; "Campaign X has $1,840 in last-30-day spend on 12 keywords with 0 conversions and QS ≤ 4" is.

For unit-economics-aware framing: if `business-context.json.unit_economics.aov_usd` and `profit_margin` exist, frame waste and headroom in dollars saved / captured per month, not "above account average". See `../shared/ppc-math.md`.

## Phase 4 — Business context

Derive what you can from data already pulled:

| Field | Source |
|---|---|
| `business_name` | `customer.descriptive_name` |
| `services` | Campaign + ad-group names, top converting keywords |
| `locations` | `campaign_criterion` LOCATION + PROXIMITY |
| `brand_voice` | Top-performing RSA headlines / descriptions |
| `keyword_landscape.high_intent_terms` | Converting keywords with strong CVR |
| `keyword_landscape.competitive_terms` | Keywords in campaigns with high rank-lost-IS |
| `keyword_landscape.long_tail_opportunities` | Converting search terms not yet promoted to keywords |
| `website` | Apex domain from ad final URLs |

Then crawl the website (homepage + about + services + top 3 ad landing pages, parallel `WebFetch`) and merge into the schema. See `references/business-context.md`.

Ask the user — it's faster than guessing — for: differentiators, competitors, seasonality, unit economics (AOV, margin). Ask for everything else only if the data + crawl can't answer it.

## Phase 5 — Personas

Discover 2–3 personas from search terms, top keywords, ad-group themes, landing pages, geo, and device split — all from the dataset already in memory. Persist to `{data_dir}/personas/{accountId}.json`. Each persona must be grounded in **5+ actual search terms**; if not, drop it. See `references/persona-discovery.md`.

## Phase 6 — Report

Structure: pulse metrics (3 lines, each with number + top contributor + fix pointer) → per-area findings (only those that surfaced something material) → Quick Wins section (per the rules in `references/account-health-scoring.md`). Cap at ~80 lines. Every claim cites a specific entity, number, and window.

End with a single closing line after the handoff to `/google-ads`:

State where any audit artifacts were actually saved. Do not claim hosted audit history unless a live result confirms it.

## Guardrails

1. **Read-only skill.** Diagnose; don't mutate. Every fix routes through `/google-ads` (or `/google-ads-copy`, `/google-ads-landing`). End the report with one handoff tied to the #1 action.
2. **STOP condition.** If conversion tracking is broken, recommend pausing spend until it's fixed before recommending anything else.

From Swastik023/marketing-api-automation-mcp, MIT.