Diagnose Your Ad Panel Data: Find Where Your Money Is Leaking in the Funnel
Opening your Meta or Google ad panel and seeing ROAS drop is easy. Understanding WHY it dropped is the hard part. Most people look at one metric and make a panic decision: CTR drops, they swap the creative immediately; CPA rises, they cut the budget immediately. But metrics are not independent, they show different layers of a funnel. Low CTR is a creative problem, but a low conversion rate paired with a high CTR means the person who clicked was not convinced on the page. Those are two different diseases needing two different treatments. High frequency is a completely different signal, it says the audience is tired of seeing the same ad. This prompt takes a panel screenshot or raw numbers and tells you exactly which metric is a symptom of what, which layer of the funnel the real problem sits in, and whether the right move right now is to kill it, scale it, or wait.
You are a paid media analyst who reads ad performance data for a living. Your job is not to read the numbers on my panel back to me. It is to tell me what those numbers say together, which layer of the funnel the problem sits in, and the one concrete action I should take right now: kill it, scale it, or wait. Instead of judging a single number in isolation, you build the real diagnosis by reading the numbers against each other. Use the following information: - Platform: [Meta, Google Ads, TikTok Ads, etc.] - Time range: [last X days] - Spend: [total, in dollars] - Impressions and clicks: [if available] - CTR: [percentage] - CPA or cost per acquisition: [number] - ROAS: [if available, as a ratio] - Frequency: [if available, average times shown] - Context: [is this a new campaign, how many days has it been live, did you change anything recently] - Goal of this campaign: [sales, leads, or traffic; which metric matters most to you] GIVE THE OUTPUT IN THIS ORDER, WITH THESE EXACT HEADINGS: ═══ FUNNEL MAP ═══ Place the metrics I gave you into the three layers of the funnel: VISIBILITY (impressions, frequency), INTEREST (CTR, clicks), CONVERSION (CPA, ROAS). Mark each layer as healthy or problematic, and justify in one sentence what threshold makes you call it problematic. ═══ WHERE THE REAL PROBLEM IS ═══ Read the metrics together, not one at a time, and name in a single sentence which layer the problem is concentrated in (example: creative fatigue, message-to-page mismatch, audience narrowing, bid competition). Explain why the other layers are secondary right now. ═══ SYMPTOM VS. ROOT CAUSE ═══ Separate whether the number I am seeing is the problem itself, or a SYMPTOM of something deeper (example: a low CTR may not be a creative problem on its own, it may be a symptom of the message not resonating with the audience). Say which additional observation, visual, copy, or targeting, would confirm the root cause. ═══ WHAT I SHOULD DO RIGHT NOW ═══ Recommend one of three options clearly: KILL (why), SCALE (by how much, how gradually), WAIT (for how long, what threshold triggers intervention). Flag the risk of deciding too early, and state honestly how this recommendation could turn out wrong. ═══ WHAT TO WATCH NEXT WEEK ═══ Tell me which metric to check after taking this action, to see whether it moved. Be specific about what I should be tracking for the next diagnosis, and what the next step is if that metric improves versus if it gets worse. RULES: - Do not invent industry benchmarks or fake comparison numbers (like a made-up "a good ROAS is X" threshold). Judge only the internal consistency of the data I gave you. - Never make a panic decision based on a single metric. Always read at least two metrics together. - Where you are not sure, say so: "this cannot be determined with certainty from this data, you would need to check X." - Prioritize based on my campaign goal. Do not put click-through rate first if the goal is sales, and do not put ROAS first if the goal is leads. - Write in English, plain and direct. Do not leave the required action vague.
How to use it
- Copy the core metrics from your ad panel (Meta Ads Manager, Google Ads, TikTok Ads) for the last 7 to 14 days: spend, impressions, clicks, CTR, CPA, ROAS, frequency. The more complete the numbers, the sharper the diagnosis.
- Run the prompt and get the funnel layer the problem sits in plus the recommended action. Before acting on it, check it against context only you know: new product, seasonality, a competitor move.
- If the diagnosis says wait, do not intervene early. If it says kill, actually kill it. If it says scale, scale by the suggested amount. Feed it the same data again next week and track whether the direction changed.
Example / tip
Input: Meta campaign, last 10 days, $130 spend, CTR 2.1%, CPA $11, ROAS 1.4, frequency 5.8. AI output flags the high frequency, calls it a sign of creative fatigue, notes the CTR still looks reasonable on its own but reads as audience narrowing once paired with the frequency number; recommends refreshing the creative and widening the audience first, and advises against cutting or raising budget immediately.
Usage example
Copy the prompt as is, fill in the bracketed fields for your own business, and paste it into ChatGPT or Claude. The more concrete your input, the more useful the output. Take the first draft, ask for a one-sentence fix on whatever you don't like, then give it a final pass in your own voice.
When not to use it
This prompt interprets the numbers you already see in the panel, it does not manage the campaign for you, and it cannot guarantee results without real account access. You always make the final call.
Output quality checklist
- Does the diagnosis rest on reading multiple metrics together, not just one number in isolation.
- Is the funnel layer the problem sits in (visual, message, page, audience fatigue) clearly named.
- Is the recommended action concrete (kill, scale, wait), not a vague suggestion.
