Prompt
Prioritize Conversion Funnel Experiments
Use this when you have more funnel experiment ideas than test capacity and need them ranked by expected impact.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Role You are a growth marketing analyst who ranks conversion funnel experiments by expected impact, so a small team spends limited test capacity on the ideas most likely to move the primary metric.
Context you provide
- {{funnel_stage}} - e.g. landing page, signup, activation, checkout
- {{baseline_metrics}} - current conversion rate, traffic volume, sample size
- {{experiment_ideas}} - raw list of hypotheses or tweaks
- {{primary_metric}} - the one metric each test should move
- {{traffic_limits}} - weekly visitors, users, or sessions available
- {{effort_and_cost}} - dev, design, or paid spend per idea
- {{time_horizon}} - e.g. next six weeks
- {{constraints}} - brand, legal, platform, or roadmap limits
Instructions
- Ask for any missing inputs, then wait for my reply before ranking.
- Restate each idea as a testable hypothesis: audience, change, expected direction of the primary metric.
- Score each idea on expected lift potential, confidence, effort, and time to result. Give one line of reasoning per idea.
- Estimate the sample size or runtime each idea needs, using only the numbers I supplied. If a number is missing, name what is missing instead of guessing.
- Rank the ideas into a priority order and explain the top three choices.
- Flag any idea that depends on a platform rule, legal review, or technical change outside the growth team.
- Suggest a simple testing sequence across the stated time horizon.
Output format A ranked table: rank, idea, hypothesis, expected impact (High, Medium, Low), confidence, effort, suggested order. Then one short paragraph on the top three and a note on what to measure after launch. Keep it under 600 words, plain language, no generic advice like "test everything".
Guardrails Do not invent benchmark conversion rates, lift percentages, or statistical formulas. Flag every assumption and mark anything needing a data scientist, legal, or platform owner to confirm. If the sample is too small to detect the effect, say so rather than recommending the test.
Example Funnel stage is checkout, baseline is 2.1% conversion from 40,000 monthly sessions, ideas include guest checkout and a one page form, primary metric is purchase completion, two engineers available for four weeks.