The 5-Minute Prompt That Turns Your Meta Ads Data Into a Copy Strategy

Most advertisers look at their age and gender breakdown, nod at the chart, and move on. That's a wasted export. Buried in that one report is the answer to three questions that should shape every campaign you run: who is actually buying, who are you paying to reach for nothing, and what should your next ad say?

You don't need a data analyst to answer them. You need a decent export and a well-built prompt.

The export (two minutes in Ads Manager)

Before the prompt can do its job, it needs the right data underneath it:

1. In Meta Ads Manager, set your date range to the last 30 days

2. Apply Breakdown → Age and Gender

3. Make sure your columns include purchases, cost per purchase, CPM, CPC and link clicks

4. Export to CSV

That's it. Thirty days is long enough to smooth out daily noise, short enough that the picture still reflects your current creative and audiences. If you don’t have enough sales data then extend as needed.

The prompt

Paste this into your AI tool of choice along with an upload of the .csv file which you have downloaded from Ads Manager:

You are a Meta ads analyst helping me build a copy strategy for a UK ecommerce brand. Below is 30 days of Meta ads performance data broken down by age group and gender.

Please analyse the data and give me:

1. Purchase performance by demographic — which age/gender segments are actually buying, ranked by number of purchases and cost per purchase. Flag any segment with a cost per purchase significantly better or worse than the account average.

2. Efficiency signals — compare CPM, CPC and link clicks across segments. Identify where we're paying a premium to reach people who don't convert, and where cheap traffic is also converting (the sweet spot).

3. Intent vs conversion gaps — highlight any segments with high link clicks but low purchases (interested but not converting — possible copy/landing page mismatch) and any with low clicks but strong conversion (small but high-intent audience worth scaling).

4. Primary audience recommendation — based on the above, tell me which age group (and gender split if relevant) I should prioritise in future campaigns, with the data reasoning behind it. Include a secondary audience worth testing.

5. Copy strategy implications — for the recommended primary audience, suggest: the tone and messaging angles most likely to resonate with this demographic, what pain points or motivations to lead with, and 3 example hook lines (first ~50 characters of primary text) I could test.

Present the analysis in clear sections with a short summary table of the key metrics per segment at the top. Use UK English. Be direct about what the data does and doesn't support — if sample sizes are too small to draw conclusions for a segment, say so rather than guessing.

Why each section earns its place

Purchase performance is the anchor. Everything else in your account is a means to this number, so the analysis starts with who converts and what each conversion costs — benchmarked against your account average, not judged in isolation.

Efficiency signals catch the trap most accounts fall into: cheap reach that never converts. A low CPM feels like a win until you notice that segment hasn't bought anything in a month. This section forces the comparison between what you pay to reach people and what you get back.

Intent vs conversion gaps is where the diagnostic value lives. A segment clicking heavily but not buying isn't a dead audience — it's a mismatch. The creative is promising something the landing page isn't delivering, or the message resonates but the price point doesn't. Meanwhile, a segment with modest clicks and a strong conversion rate is quietly telling you it's ready for more budget.

The audience recommendation turns analysis into a decision. Not five options, not "it depends" — one primary audience with the reasoning behind it, plus one secondary worth testing. That's a media plan, not a report.

Copy strategy implications closes the loop. Knowing that women 45–64 drive your purchases is useful; knowing what to say to them is money. Asking for hook lines such as the first ~50 characters of primary text, the only part most people read, means you leave the analysis with something you can put live the same day.

The line that makes it all trustworthy

The most important sentence in the prompt is the last one: "if sample sizes are too small to draw conclusions for a segment, say so rather than guessing."

AI tools are eager to please, and an eager analyst is a dangerous one. Eight purchases in a segment can produce a spectacular-looking cost per purchase that evaporates the following week. Explicitly instructing the model to flag thin data is the difference between an analysis you can act on and one that sends you scaling a snapshot.

Make it a habit

Run this monthly and the value compounds. One export tells you where you stand; three in a row tell you whether your new creative is actually shifting your buyer profile, whether that promising small segment holds up under more spend, and whether your targeting has drifted away from the people who pay the bills.

Five minutes of exporting, one paste, and you've replaced gut feel with a data-backed answer to the only question that matters: who's buying, and what do we say to them next?