Audit Your Recent Social Posts to Find What Your Top Performers Share

Paste your recent posts with their reach or likes and get a ranked table, the patterns behind your best and worst posts, and five data-backed changes to test.

You getRanked table, what top posts share and what bottom posts share

Prompt

Prompt
Act as a social media analyst who audits accounts using their own performance numbers, not best-practice lists.

Right after this message I will paste my recent posts (the text or a short description of each, plus the platform) with their reach, likes, comments or saves. If I paste fewer than 8 posts or leave out numbers, ask me for them before you start.

Your job: find what separates my top performers from my bottom performers and tell me what to change.

Rules:
1. Rank posts by the metric I give most consistently. If I give reach plus interactions, use engagement rate (interactions divided by reach).
2. Every pattern must cite at least two posts by number as evidence.
3. Compare traits that actually vary across my posts: hook, format, topic, length, posting day or time, call to action.
4. No generic advice. Every recommendation must trace back to a pattern in my data.
5. Label any pattern backed by only 2-3 posts as weak.

Output format:
1. Ranked table: Post #, short description, metric, tier (top, middle, bottom).
2. What top posts share: 3-5 bullets with evidence.
3. What bottom posts share: 3-5 bullets with evidence.
4. Change list: 5 actions, each with the pattern behind it and a simple test for next week.
5. Two post ideas built on the strongest pattern.

End with one line on what this data cannot tell us, such as sample size, algorithm shifts or missing metrics.

Example output

At a glanceEngagement rate by LinkedIn post10 recent posts ranked, top tier above 4.8%
  • Post 4: £3,200 VAT story: 6.1%
  • Post 7: HMRC receipts carousel: 5.4%
  • Post 9: Monthly vs January question: 4.8%
  • Post 2: Mileage rates tip: 3.0%
  • Post 6: Desk setup photo: 2.7%
  • Post 1: Why I went freelance: 2.5%
  • Post 10: Blog link: 1.9%
  • Post 8: Quote graphic: 0.7%

Example for: a freelance bookkeeper's 10 most recent LinkedIn posts, with impressions and reactions.

1. Ranked Table (engagement rate)

Post #DescriptionRateTier
4Story: client found £3,200 in missed VAT reclaims6.1%Top
7Carousel: 5 receipts HMRC queries most5.4%Top
9Question: "Do you do your books monthly or in January panic?"4.8%Top
2Tip: mileage rates explained3.0%Middle
6Photo of my desk setup2.7%Middle
1Text: why I went freelance2.5%Middle
10Link to my blog on Making Tax Digital1.9%Middle
3Link to a news article on tax deadlines1.1%Bottom
5Promo: "Spots open for Q3 clients"0.9%Bottom
8Quote graphic about hard work0.7%Bottom

2. What Top Posts Share

  • A number in the first line (Posts 4, 7). Both open with a specific figure or count.
  • Speak to a fear owners already have (Posts 4, 7, 9): HMRC queries, missed money, January stress.
  • End with a question (Posts 4, 9). Comments were 3x your average.

3. What Bottom Posts Share

  • External links (Posts 3, 10). Both sit in the lower half.
  • No personal angle (Posts 3, 8). Nothing shows your work or your clients.
  • Direct selling with no story (Post 5). Weak: only one example.

4. Change List

  1. Put a number in line one. From: Posts 4, 7. Test: 2 posts next week opening with a figure.
  2. Move links to the first comment. From: Posts 3, 10. Test: repost the Making Tax Digital blog with the link in comments.
  3. Close with a yes/no or either/or question. From: Posts 4, 9. Test: add one to every post this week.
  4. Wrap offers in a client story. From: Posts 4 vs 5. Weak pattern. Test: announce Q3 spots inside an anonymised client result.
  5. Drop quote graphics. From: Post 8. Test: replace with one carousel.

5. Post Ideas

  • "7 expenses sole traders forget to claim (No. 4 cost one client £1,100)" as a carousel.
  • "A client sent me 412 receipts in a shoebox. Here's what we found." End with: "Shoebox or spreadsheet?"

Limits: 10 posts is a small sample, posting times weren't provided, and saves and profile views would sharpen the ranking.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Requiring two cited posts per pattern stops the model from falling back on generic social media tips.
  • Ranking by engagement rate instead of raw likes keeps high-reach posts from skewing the results.
  • Pairing each change with a one-week test turns the audit into experiments you can actually run.

Use it in ChatGPT, Claude or Gemini once you have 8-20 recent posts. Copy each post's text or a short description and its reach, likes, comments or saves from your platform analytics.

Common edits

  • Add a goal, such as profile visits or link clicks, and ask the model to rank by that metric instead.
  • Ask for the post ideas written out in full, ready to schedule.

Prompt details

By
Prompts-GPT editors
Published

Did this prompt work?

Keep exploring

More in Social Media