Cross-channel playbooks

1  Ground it      get_business_context · get_guardrails · get_brand_profile
                  search_knowledge        ← what happened last launch
2  Test the hook  create_marketing_asset_draft (social_post) × 3–5, one variable
                  social_create_post   (HIGH)     ← free hook test, 3–5 days
3  Build content  create_marketing_asset_draft (blog_post) — the launch page,
                  targetKeywords from google_search_performance's top queries
4  Build paid     meta_list_pages · list_media_files
                  meta_draft_campaign_plan (low) using the winning hook
                  → show the plan → meta_deploy_campaign_plan (HIGH, paused)
                  → meta_activate_campaign_plan (HIGH) on launch day
5  Local / AI     localfalcon_list_reports — baseline the launch keyword now,
                  so the after-scan has a before
6  Record it      add_knowledge (past_campaign) — the plan and the baseline

1  The code     promotions_list → promotions_create_draft (low) with the code
                and discount the user stated — never invented
                → promotions_publish_to_store (HIGH) if a store is connected
2  Paid         meta_draft_campaign_plan (low) with promotionId → show the plan
                → meta_deploy_campaign_plan (HIGH, paused)
3  Organic      social_create_post (HIGH) with the same promotionId, one post
                per platform in its own voice
4  Content      create_marketing_asset_draft (email / blog_post) — the code in
                the copy, the link from promotions_list's landingUrl
5  Judge it     promotions_performance — orders and revenue on the code
6  Record it    add_knowledge (past_campaign) — partner, code, the result

1  marketing-audit              full read across every channel
2  marketing-diagnose           on the two or three worst findings only
3  marketing-scale              winners up, losers paused
4  creative-testing             one test for next month, one variable
5  ai-search-visibility         re-scan the same grid; trend vs. last month
6  add_knowledge                the month's conclusions, so the next one starts
                                from evidence

1  Confirm      meta_list_ads (7d and 30d) — ROAS ≥ 1.3 × break-even in BOTH
                windows, ≥ 50 conversions in 7d, no fatigue
2  Paid         meta_set_budget (HIGH) — clamped to 2 × and the daily cap
3  Same hook,   the winner is a message, not an ad: run it on the other paid
   other paid   channels too, each created PAUSED and enabled separately
                google_ads_create_campaign → google_ads_create_ad_group (HIGH)
                → google_ads_create_responsive_search_ad (HIGH)
                + google_ads_add_keywords (LOW) — no ad or keyword, no serving
                tiktok_ads_create_campaign → tiktok_ads_create_ad_group (HIGH)
4  Organic      the same claim, native to each platform:
                create_marketing_asset_draft → social_create_post (HIGH)
5  Content      create_marketing_asset_draft (blog_post) — the hook is the angle
6  Email        create_marketing_asset_draft (email) — the hook is the subject
7  AI search    the winning language belongs in the page AI surfaces will quote
8  Record       add_knowledge (past_campaign) — the hook, the numbers, the window

1  Baseline     localfalcon_list_locations → is the new place_id even there?
                localfalcon_run_scan (HIGH) — chatgpt + gaio + google,
                the head keyword, a grid matched to the real service area
2  Content      create_marketing_asset_draft (blog_post) — a page that names the
                city and the service area in words
3  Social       social_create_post (HIGH) — location-specific, on the connected
                accounts
4  Paid         meta_draft_campaign_plan (low) — countries checked against the
                geo guardrails BEFORE drafting
5  Re-scan      same grid, same radius, 30 days later. ATRP and SoLV before/after

1  Measure      localfalcon_list_reports / _get_report — who is named, at which
                points, on which surfaces. Name them from the data, not from
                the user's impression
2  Read them    what do they say about themselves in the text these surfaces
                quote
3  Answer back  create_marketing_asset_draft (blog_post) — the question,
                answered directly, with the proof search_knowledge can support
4  Be current   social_create_post (HIGH) — quiet businesses are hard to name
5  Compete paid meta_draft_campaign_plan (low) on the same intent, if the
                economics support it
6  Re-scan      same grid, 30 days

1  Find the waste   marketing-audit, then pause what is bleeding, on every
                    connected channel (meta_pause_entity /
                    google_ads_pause_entity / tiktok_ads_pause_entity —
                    low, immediate)
2  Redeploy it      the freed budget into entities that pass the scale gate,
                    wherever they are (meta_set_budget /
                    google_ads_set_campaign_budget / tiktok_ads_set_budget
                    — HIGH). The cap is one number across all three, so what
                    step 1 freed on one channel is spendable on another
3  Free channels    social_create_post, create_marketing_asset_draft,
                    the AI-search work in ai-search-visibility — all of it
                    costs no media budget
4  Fix the target   if targetCplCents is null, derive a candidate from deal
                    value × close rate × margin × ad-spend share, and ask the
                    user to SET it. Do not adopt it silently

## <playbook> — <business>

**Done now** (low tier, already applied)

- <draft created, entity paused, plan drafted>

**Waiting for your approval** (nothing has happened yet)

- <the exact change, per item, with the number>

**Skipped**

- <channel> — not connected

**Next checkpoint**

- <date> — <what to look at, and why that date>