Google Ads Draft Summary

Tracking Fix (P0)
  ├── blocks → Budget Scaling (P2)
  ├── blocks → Bid Strategy Changes (P2)
  └── informs → Structure Changes (P2)

Negative Keywords (P1)
  └── before → Structure Changes (P2)
      └── before → RSA Refresh (P3)
          └── before → Budget Scaling (P2)

# Draft Summary — [Date]
Account: [Name] (CID: [ID])
Tracking confidence: [HIGH / MEDIUM / LOW / BROKEN]
Pending drafts: [N]
Total estimated waste addressable: $[amount]/month

## Recommended Apply Order

### Step 1: [Draft name] — P0
- **File:** `workspace/ads/drafts/[filename]`
- **Type:** [tracking-fix / negatives / structure / budget / rsa / landing-review]
- **Summary:** [one sentence]
- **Impact:** [Critical / High / Medium / Low]
- **Risk:** [High / Medium / Low]
- **Reversibility:** [Easy / Moderate / Hard]
- **Estimated waste stopped:** $[amount]/month
- **Blocked by:** [nothing / other draft]
- **Blocks:** [list of drafts that depend on this one]

### Step 2: [Draft name] — P1
[same format]

### Step 3: [Draft name] — P2
[same format]

[... repeat for all pending drafts]

## Dependency Map

```text
[visual or text representation of dependencies]

---

## How to Generate the Summary

### Step 1: Read the Index
Load `workspace/ads/drafts/_index.md` to get the list of all drafts and their current status.

### Step 2: Read Each Pending Draft
For each draft with status `proposed` or `approved`, read the full file to extract:
- Action type(s)
- Confidence level
- Risk and reversibility
- Dependencies
- Estimated impact (waste $ if quantified)

### Step 3: Check Tracking Confidence
Load the most recent tracking diagnosis from `workspace/ads/findings.md`.
Tracking confidence determines which drafts are blocked.

### Step 4: Build the Dependency Chain
Map out which drafts depend on which. Apply the standard ordering:
1. Tracking fixes
2. Keyword pauses (stop active bleeding)
3. Negative keyword additions
4. Structure changes
5. RSA refreshes
6. Landing page improvements
7. Budget reallocation

### Step 5: Assign Priority and Impact
Using the P0-P3 framework and impact/risk tables above.

### Step 6: Write the Summary
Output the Draft Summary document to `workspace/ads/drafts/_summary.md`

---

## Output Locations

- **Summary document:** `workspace/ads/drafts/_summary.md` (overwritten each time)
- **Audit packets:** `workspace/ads/drafts/_batch-YYYY-MM-DD-[account-slug].md` (created by `/google-ads audit` when one run emits 2+ drafts)
- **Index update:** Update `workspace/ads/drafts/_index.md` if any status changes are warranted
- **Console output:** Also display the summary inline for immediate review

---

## Multi-Account Support

When `workspace/ads/drafts/_index.md` contains drafts for multiple accounts (identified by account slug in the filename):

- Group the summary by account
- Show per-account tracking confidence
- Show per-account total impact
- Cross-account dependencies are rare but possible (shared negative lists, MCC-level changes)

---

## Rules

- **Read every pending draft.** Don't summarize from the index alone — the index is a title list, not a summary.
- **Always check tracking confidence.** It gates everything.
- **Be honest about confidence.** If a draft says "Medium confidence," don't promote it to P1.
- **Surface the 20-minute review.** The operator should be able to read the summary, pick the top 2-3 actions, and apply them in one session.
- **Don't create new drafts.** This skill summarizes existing drafts — it doesn't generate new recommendations.
- **Flag stale drafts.** If a draft is >30 days old, note it. The data it was based on may no longer be current.
- **Flag conflicting drafts.** If two drafts recommend contradictory actions, call it out explicitly.