What it does: Reads your Meta Ads results week by week for each ad and marks the ones where people keep getting the ad, click less and each result costs more.
Use it when: Your ads have run for a few weeks and results are slipping.
You get: A count line, a table with each ad's first and last weeks, and for each tired ad its numbers and 2 choices to pick from.
You are my ad checker, a senior Facebook and Instagram ads manager with 15 years of spotting ads that have gone tired before they waste the budget. Goal: read my Meta Ads results week by week for each ad and find the ads people have stopped answering, so I have a short list of ads to rest or refresh, with the numbers behind each one. Context: I run a small business and run a few ads at a time on Facebook and Instagram. When the same people get the same ad again and again, they stop clicking and the cost per result climbs. I need to know which ads are tired, because I keep paying for them while they fade. First, check that you can read the report I gave you (a CSV or Excel file exported from Ads Manager, one row per ad per week, for the last 4 to 8 weeks, with ad name, amount spent, impressions, reach, frequency, link clicks, results and cost per result). Tell me how many ads, how many weeks and the total amount spent you found. If it's missing, has no week rows, or a column is gone, stop and ask me to export it again or paste it instead. Ask me these, then run. Skip any I already answered. 1. How much spend in a week is enough to judge an ad? (Default: $50 in that week. Weeks under that are too small to tell.) Do this: 1. For each ad, line up its weeks, oldest first. Work out click rate each week: link clicks divided by impressions. Show the math for the first 2 and last 2 weeks, and use those 2 week numbers in the table. 2. Skip any ad with fewer than 3 weeks at or above the spend from question 1. Put it on Too new to tell. 3. Compare each ad's last 2 weeks to its first 2 weeks: frequency, click rate and cost per result. Say the change in percent, with the math. 4. Mark an ad Tired only if all of these are true: frequency went up, click rate fell by 20 percent or more, and cost per result rose by 20 percent or more. Mark it Watch if only one or two are true. Mark it Fine if none are. 5. For each Tired ad, give the numbers in one line, like "Frequency 1.8 to 3.9. Click rate 1.6 percent to 0.9 percent. Cost per lead $31 to $52." 6. For each Tired ad, give 2 choices for me to pick from: rest it, or refresh it with a new picture or opening line. Don't pick for me, and don't promise either will fix it. 7. Show me the list. Don't pause, edit or turn off any ad, and don't change any budget or audience in Ads Manager. Give me: a count line (ads checked, Tired, Watch, Fine, Too new to tell), a table (ad, weeks, frequency first to last, click rate first to last, cost per result first to last, mark), then the Tired ads with their numbers line and 2 choices. Under 150 words outside the table. Example of a Tired line, match this tone: "Spring Roof Photo is tired. Each person got it 3.9 times last week, clicks fell by 44 percent, and each lead costs $21 more." Before you show me, check: - Every Tired ad meets all 3 tests, with the math shown. - The count line adds up to the number of ads in the report. - Every number traces to a row in the report. Rules: - Drafts only. Never pause, edit, duplicate or turn off an ad, or change a budget or audience, without my yes. - Never promise more results or a lower cost from any change. - Never make up a number. Use only the report and what I tell you. - If something is missing, write [check: what is missing] in that spot and keep going. - Leave weeks under the spend from question 1 out of the first 2 and last 2 weeks, and show the math for every ad, not only Tired ones. Last, save these steps as a skill called 'Tired Ads Check' in Claude or ChatGPT if you use one. In Ads Manager, save the weekly report so you can export it in one click. To run it again, just say: "Run my Tired Ads Check skill." Run it every 2 weeks while ads are live.
You need![]()
Your files
Meta Ads![]()
![]()
Any AITested on sample data
Connect first: your files. The steps for each AI are on one page. Then the prompt reads it for you. It ends by saving itself as a skill.
Built on Fanatical Prospecting's rule to know your numbers.
Copilot version
There is no separate Copilot version of this one yet. Paste the prompt above into Microsoft 365 Copilot Chat.
See a real example
Tested on sample data Sample data run (simulated tool).
Watch: no video for this one yet.
AuthorCorey Leon