Your Meta ads just fell off a cliff. Killing the winner feels right — and it's usually wrong. This skill turns account panic into a five-phase diagnosis, backed in real time by 500+ professional media buyers seeing what you're seeing.
A proven winner. Five weeks of 2.5x. Then one morning the dashboard is red and the numbers don't make sense — CPMs went down and performance still collapsed.
The urge is to flip twenty switches: kill it, dupe it, raise the budget, change the bid, rebuild the campaign. The skill's job is to slow that down to exactly one change at a time — so you can actually tell what worked.
TRIGGER PHRASES IT LIVES FOR: “MY ADS DIED” · “META IS ACTING WEIRD” · “WHAT'S WRONG WITH MY ACCOUNT”
A full diagnostic feels like a conversation with a sharp colleague — not an intake form. Drag through the flow (KEEP SCROLLING →)
Separate what you know from what you suspect. Everything you already said is extracted first — then one batch of questions, five or fewer, never re-asking. The timeline alone splits creative fatigue from algorithmic breaks: fatigue is gradual; cliffs are not.
The symptom is matched to one of six diagnostic playbooks. Checks run cheapest-first and most-common-first — you check tracking before you rebuild a single campaign. Every check is answerable in Ads Manager in under two minutes.
The part generic advice can't do. Meta's behavior shifts month to month, and the Foxwell Founders community notices within days. A platform-wide cost-cap breakage looks identical to an account problem from inside one account — this phase is how you tell them apart.
EVIDENCE IS WEIGHTED BY AGE — PLATFORM QUESTIONS DEMAND FRESH THREADS.
Before anything is recommended, facts are separated from hypotheses. This is the discipline step — it kills the classic failure mode of treating a guess like a diagnosis and rebuilding the account over it. Every likely cause gets a confirmation step you can actually perform.
Ordered cheapest-and-most-reversible first. One structural change at a time — change bids, budget and creative together and you learn nothing. Every action gets a monitoring window and a tripwire, scaled to spend level. And when the community says it's platform weather, the plan is: protect spend and wait it out.
Inside one ad account, a platform-wide meltdown and a self-inflicted wound look exactly the same. The skill queries the Foxwell Founders community live — 500+ professional media buyers whose field reports become the evidence layer of every diagnosis.
Two to four distinct queries per case: one phrased like the symptom (“cost cap not spending”), one for the suspected mechanism (“CPM spike placement shift”), one scoped to the time window. source_filter="slack" catches what's happening right now; unfiltered pulls playbook and blog depth.
Called whenever the change is sudden or unexplained. If “cost controls failing” is this week's top theme, the entire diagnosis reframes — from “what did I break?” to “how do I operate through platform instability?”
For high-stakes or ambiguous cases, the skill routes you to a named specialist in the community — because sometimes the right next action is a better question, asked of the right person.
For delivery, cost controls and CPM questions, only FRESH and RECENT threads count. For evergreen craft — fatigue signals, testing cadence — the archive is gold. Every citation carries author, source, age and link.
A real diagnostic, condensed — the exact case from the skill's eval suite. No account access. Just what the buyer said, plus community evidence.
Each playbook is an ordered decision framework — not a script. Click one open.
MONITORING: 48–72H PER INTERVENTION · LONGER UNDER $200/DAY
DISCIPLINE: WORKING CAPS ARE FRAGILE — A $1 DECREASE HAS CHOKED DELIVERY ENTIRELY
JUDGE A STRUCTURE TEST ON 7+ DAYS / FULL LEARNING EXIT — NOT DAY 2
NEVER DUPE AND CHANGE THREE OTHER THINGS AT ONCE
ONE FATIGUED AD IS A CREATIVE PROBLEM · MONTHLY FATIGUE IS A SYSTEMS PROBLEM
CANARY: UNBLOCK ONE HIGH-CONFIDENCE CAMPAIGN — IF IT SPENDS, IT'S CONTROLS; IF NOT, IT'S ACCOUNT/PLATFORM-LEVEL
After the plan lands in chat, the skill saves a markdown brief to the workspace. Share it with a teammate or client — then come back after the monitoring window and continue from recorded state instead of memory. Return with results and the same brief gets updated, not restarted.