Leaderboard Ad728 × 90AdSense placeholder — will activate after approval

Funnel Analysis Query Diagnosing the Worst Drop-Off Step

Data Product Analytics Advanced 🤖 ChatGPT 👁 1 views

📝 The Prompt

I have an events table (event_name, user_id, event_time, properties JSON). Write a Postgres query that computes a funnel for these ordered steps (within a 7-day window per user, in order, no skipping): [step_1] → [step_2] → [step_3] → [step_4]. Output: total users entering the funnel, count + conversion rate at each step (vs prior step AND vs top of funnel), median time-to-next-step. Then a second query segmenting by [dimension, e.g. signup channel] to show which segment has the worst step-to-step rate. Use LATERAL joins or window functions — avoid repeated full-table scans. Include comments noting where ordering ambiguity (same timestamp) is handled.

⚙️ Replace 5 placeholders: [step_1] [step_2] [step_3] [step_4] [dimension, e.g. signup channel]

🎯 What this prompt does

This AI prompt helps you funnel analysis query diagnosing the worst drop-off step. Designed for product analytics workflows in the data category, it's a advanced-level prompt you can copy directly into ChatGPT to get instant, production-ready results.

Use it when you need a advanced prompt that produces clear, actionable output without wrestling with trial-and-error wording. Just copy, customize, and run.

In-article Ad #1336 × 280AdSense placeholder — will activate after approval

🚀 How to use this prompt

  1. Copy the prompt using the 📋 button above.
  2. Open ChatGPT (or Claude, Gemini, Perplexity, or your preferred LLM).
  3. Paste the prompt into a new chat. Replace 5 bracketed placeholders ([step_1] [step_2] [step_3] ) with your own details.
  4. Run the prompt and review the AI's response. Most outputs are usable immediately.
  5. Iterate if needed — if the tone, length, or structure isn't quite right, reply with "make it shorter", "use bullet points", or "make it more formal" and the AI will refine it.

💡 Tips for better results

  • Replace the bracketed placeholders ([step_1], [step_2], [step_3], [step_4]) with your own specifics before sending.
  • If the first output isn't quite right, ask the AI to refine, rewrite, or add more detail — iteration is key.
  • For long outputs, ask for a section at a time (e.g. 'start with the introduction only') to keep quality high.
  • Combine this with other data prompts to build an end-to-end workflow.
  • Save your favorite variations — small wording tweaks often produce noticeably different results.
In-article Ad #2336 × 280AdSense placeholder — will activate after approval

✨ What you'll get

When you run this prompt, expect ChatGPT to return:

  • A directly usable product analytics output tailored to the details you provided
  • Clear structure (headings, bullets, or numbered sections) that you can drop into your workflow
  • Content that matches your specified tone and context
  • Results in under 30 seconds — no manual drafting required

Need a different angle? Just ask follow-up questions. The AI will adjust without you starting over.

🔄 3 variations to try

1

Make it more formal

Add "Use a formal, professional tone suitable for enterprise clients" at the start of the prompt.

2

Ask for multiple options

Append "Give me 5 alternative versions, each with a different angle or approach." after the main instruction.

3

Request structured output

Add "Return the response as a markdown table (or bullet list, or JSON)" so you can paste the result directly into your docs or code.

🏷 Tags

🔎 Find more prompts like this

Browse 92 more data prompts or search the full library.

End-of-content Ad728 × 90AdSense placeholder — will activate after approval
Mobile Sticky320 × 50AdSense placeholder — will activate after approval