The Ad Comment Harvester Checklist

Drop 02 ended with a homework assignment most people skipped: collect 20+ pieces of real customer language and feed it to Prompt 4. Prompt 4 is the best prompt in the pack, and it’s useless without that input. “Go collect customer language” is not instructions. This drop is the instructions.

The Ad Comment Harvester is a timed 15-minute checklist. Ad Library pass, find the live posts, harvest verbatim quotes, tag and tally. Once a week, one competitor per session, one file that gets more valuable every time you run it.

The part every guide gets wrong

The Meta Ad Library does not show comments. It shows the ads, the start dates, the platforms. That’s it. Plenty of tutorials wave at “mine the Ad Library comments” as if the button exists. It doesn’t.

The comments live on the actual post, on the brand’s Facebook or Instagram page. So the routine uses the Ad Library for the only thing it’s good for here: telling you which ads deserve your attention. An ad that’s been running 60+ days is an ad the brand keeps paying for, which means it works, and it’s had two months to collect comments. The checklist has you find those, then walk over to the brand’s page and find the matching posts. If the ad is a dark post that never hit the page, you skip it and move on. There’s a two-minute cap on hunting, because the routine only survives if it stays short.

What you’re actually collecting

Not sentiment. Not inspiration. Vocabulary. The checklist has explicit take-and-skip rules: take pain in their own words, objections, desired outcomes, specific results, repeated questions. Skip tag-only comments, bots, brand replies, and one-word reactions.

Everything goes into one file, customer-language.md, verbatim, typos included. Each quote gets one of five tags (PAIN, OBJECTION, DESIRE, RESULT, QUESTION) and a frequency tally. The tally is the whole game. One vivid quote is a curiosity. The same complaint five times is a creative brief.

And for niches where nobody comments on ads, the checklist includes four backup sources with the same capture rules: Amazon 3-star reviews, Reddit threads, TikTok comments, and your own support inbox, which is the highest-grade source and the one everyone forgets.

Why the manual way costs you

Without a routine, customer research happens in one of two ways. You pay for it: an agency discovery phase or a VoC tool subscription, hundreds a month to be told what your customers say. Or you binge it: one three-hour scroll through comments the night before a creative session, no capture system, and by ad number 40 everything sounds the same. The binge doesn’t repeat, so the file never compounds, and 90 days later your personas are stale and you start over.

Fifteen minutes a week fixes both. Four sessions in, you’ve crossed the 20-quote threshold that Prompt 4 needs. Twelve weeks in, you have a frequency-ranked map of what your market complains about, in its own words, that no competitor can download.

How to use it

1. Do the one-time setup: create customer-language.md next to brand-kit.md, paste in the capture table from the checklist, list 3 to 5 competitors, bookmark their Ad Library and Facebook pages. Ten minutes.
2. Run the 15-minute routine on one competitor. Set an actual timer. The time cap is a feature, not a suggestion.
3. At 20+ quotes, run Prompt 4 from the Persona Generator Prompt Pack (Drop 02) and paste the file in. Your personas get rewritten in real customer phrasing.
4. Keep the weekly loop going and rotate competitors. High-frequency PAIN quotes become Pillar 1 headline tests; repeated QUESTIONs become the objections your ads answer up front.

One rule from the checklist worth repeating here: never lift a comment into your ads word for word. You’re harvesting phrasing patterns, not sentences. Rewrite before you publish.

Next Monday’s Tool Teardown: verdict as always, one real test, keep or replace.

Next Friday’s drop: the UGC Script Prompt Pack. Turn your customer-language.md file into 30-second video ad scripts that sound like a customer talking, because the words started as one.