Paid Social Ads & Creative (Meta)

Fixes: the client runs no paid acquisition, or burns budget on bad creative, or pays an agency $10–15k/mo for four things a skill stack now does — research, creative, campaign, reporting. This is a demand-generation play: once the front office captures leads well, paid ads pour fuel on it. An effectiveness fix (more qualified demand) with a hard, trackable ROI (ROAS/CPA).

Sequencing rule (opinionated): do not turn on paid ads into a leaky funnel. Fix capture first — missed-call-text-back.md + speed-to-lead-followup.md — then buy attention. Driving paid clicks to a business that doesn’t answer its phone or follow up is setting money on fire. Ads are the last front-office system you turn on, not the first.

The stack (with alternatives)

The gooseworks open-source Claude skill library packages the four agency functions as skills. Each maps to a phase:

Agency function	gooseworks skill	What it does
Research	`competitor-ad-intelligence`	Tear down what competitors run on Meta (via the Ad Library) + reverse-engineer their funnels
Creative	`remix-graphic-ad-from-reference`	Remix proven winning ads onto the client’s brand — real static creative
Campaign	`meta-ads-campaign-builder`	Launch-ready Meta campaign: targeting, ad sets, copy
Reporting	`meta-ads-analyzer`	Drop in the ad data → what to scale, what to cut

Install: npx gooseworks install --all (routes ad work to the goose-ads entry skill; the rest of the 100+ catalog is fetched on demand).

Verified 2026-07-05 — it is NOT free. The npm package (gooseworks@0.3.x, built by the Athina AI team, gooseworks.ai) is a credentialed, credit-metered service: skills require gooseworks login + GOOSEWORKS_API_KEY, are fetched from their hosted catalog, and the underlying data providers/generators bill against their credits (gooseworks credits checks balance). “Less than $100/mo” in the launch post = their paid credit plan, not open-source freeware. The install is free; the usage is metered. Budget it like any other tool line-item, and note the dependency risk: skills live in their catalog, not your repo — an account or pricing change upstream changes your delivery cost.

Job	Recommended	Alternatives	Why / when to switch
The four ad functions in Claude	J&M’s own `jm-ads` skill pack (built 2026-07-06, `~/projects/jm-ads`) — router + brand-pack + competitor-intel ($0, public ad libraries) + angle-miner + campaign-builder (spec generator, no Meta API by design) + analyzer + remix (edit-the-reference on our keys, structural human-QA gate). Reverse-engineered from gooseworks’ MIT recipes (vendored in `jm-ads/research/vendor/`) and adapted for local SMBs	gooseworks (credit-metered, 2-4× markup over raw providers — the reference implementation); AdCreative.ai, Foreplay, Motion; a human freelancer	Self-owned skills are the default: no metered middleman, recipes adapted to local-service clients (the SaaS-flavored originals assume B2B). A human editor remains non-negotiable — the remix QA gate is structural (agents can’t approve creative). Remix quality path needs a FAL key (~$0.33/creative).
Run the media	Meta Ads Manager (client’s own Business account + pixel/CAPI)	—	Ownership of the ad account and pixel stays with the client, always.

What’s cheap vs. what isn’t: the skills make the work (research, creative, build, analysis) nearly free. They do not make the media free — the client still needs a Meta Business account and a real ad budget. Sell the labor savings honestly; don’t imply $99 buys results.

AOA build steps (Audit → Optimize → Automate)

Audit — competitor-ad-intelligence: pull competitors’ live Meta ads, spot the angles/offers/funnels that are working; audit the client’s current ads + pixel/analytics and their actual CPA/ROAS if any.
Optimize — remix-graphic-ad-from-reference to put proven-winner structures onto the client’s brand (inspiration, not copying — respect IP), then meta-ads-campaign-builder for a launch-ready campaign (targeting, ad sets, copy). Human review before anything goes live.
Automate — meta-ads-analyzer on performance data each week → scale winners, cut losers, feed learnings back into new creative. Standing weekly reporting loop.

Time to ship

Competitor teardown + first creative concepts week 1. Campaign live within 1–2 weeks (gated on ad account access, pixel/CAPI setup, and budget). Optimization is an ongoing weekly loop.

Compliance / cautions

Meta advertising policies — creative and targeting must comply (especially in regulated categories: health, finance, housing/employment special ad categories).
IP — “remix” means inspired-by structure, never lifting a competitor’s exact assets. Keep it clean.
Account ownership — client owns the ad account, pixel, and audiences; J&M operates with delegated access. Never run client spend through a J&M account.

What to log (renewal mechanism)

Creatives shipped, CTR / CPM / CPA / ROAS per campaign, scale/cut decisions and their result, and spend-to-revenue. The renewal story is “$X in returned ad spend for $Y in fees” — the cleanest ROI in the whole delivery menu.