Use it
Agent playbooks
TL;DR
Playbooks are ready-to-run goals for common ad-ops jobs. Each one spans several tools and accounts and routes every write through an approval gate. Paste a goal into the Workspace, adjust the thresholds, and review the plan.
Analyze & report (read-only, no approval needed)
- Weekly performance review: “Query the last 7 days for cost, conversions, CPA, ROAS and CTR; list the top 10 campaigns by cost and top 10 search terms; flag any campaign whose CPA moved more than 25% week-over-week; give me 3 ranked next actions.”
- Account change-history review: “Summarize the last 7 days of change events grouped by resource type and user; highlight status changes and budget changes over 20%.”
- Ad disapproval monitor: “Find any ads that are DISAPPROVED or SITE_SUSPENDED with their policy topics; if any, email me the full list.”
Optimize & operate (writes — each step asks for approval)
- Mine search terms → negatives: “Audit the 14-day search-term report; flag terms with cost over $20 and zero conversions; propose campaign/ad-group negatives and wait for my approval.”
- Rebalance budgets to ROAS: “Compare 14-day campaign ROAS and move 20% of budget from campaigns under 1.5× to my top 3 above 3×; show a before/after table and ask before each change.”
- Tune top keyword bids: “Take the 20 keywords with the most conversions in 30 days and set max-CPC toward my campaign average CPA.”
- Geo cleanup: “Per campaign, find locations with cost over $50 and zero conversions in 30 days and exclude them, keeping existing inclusions.”
- Switch bidding strategy: “Change this campaign’s bidding to Target CPA at $30.”
Build & launch
- Brand search campaign: “Create a paused Search campaign ‘Brand_Test’ at 50/day, one ad group with 3 brand phrase keywords and one responsive search ad.”
- Performance Max: “Create a paused PMax campaign with an asset group — 5 headlines, 1 long headline, 4 descriptions, business name, images and logo; generate any missing images and link them.”
- Remarketing from analytics: “Read my GA4 behavior data, build a matching GA audience, and wire it into a remarketing campaign.”
- PMax asset-strength audit: “For each PMax asset group, count headlines/descriptions/images against Google’s targets and propose the missing assets.”
Cross-account at scale
Pick several accounts in the context picker and the agent fans one goal across them, with a dry-run blast-radius preview before anything writes.
- Bulk pause across an MCC: “Across the selected accounts, pause every enabled campaign matching ‘Test_’ or ‘Legacy_’ but never anything containing ‘brand’; show a dry-run first.”
- Device/geo/time tuning + experiment: “Pull 30-day device, geo and ad-schedule performance, recommend bid modifiers, and convert the campaign into a 50/50 experiment for 14 days.”
Cross-account writes always preview with dryRun first, then re-validate live state before applying.
Make a playbook reusable
Turn any goal into a Skill so the agent reuses your method on future runs, or into a Schedule so it runs on a cron (hourly, daily, weekly) and re-evaluates against live data every time.
Frequently asked questions
- Are these playbooks safe to run as-is?
- The read-only ones run immediately. The write playbooks plan the changes and pause at an approval gate so you review every mutation before it applies — and cross-account writes show a dry-run preview first.
- Can I change the thresholds?
- Yes — edit any number (ROAS, CPA, cost, date range) in the goal text. The orchestrator adapts the plan to whatever you specify.