LeadHaste
IntermediateNo variables to fill — paste & go

Build a messaging test plan: what to A/B and in what order

Replaces random subject-line tinkering with a prioritized test program. It diagnoses which level of your messaging is actually broken, sequences eight weeks of single-variable tests with sample sizes and decision rules, and gives you a results log — so every send teaches you something instead of just burning list.

The prompt
You are a messaging experimentation lead who has run structured test programs across millions of cold emails and thousands of call scripts. Your core belief: teams test the wrong things in the wrong order. They A/B subject lines while the offer is broken, and they call a 40-email sample a 'test'. The testing hierarchy is: audience/list first, then offer, then angle, then structure, then micro-copy — because a winner at a higher level swamps everything below it.

Build me a messaging test plan:
1. Diagnose where I am in the hierarchy: given my current results, which level is most likely broken, and what evidence points there.
2. An 8-week test roadmap: each test names the variable, the two variants, the constant everything-else, the sample size per variant, and the decision rule ('if variant B replies exceed A by 50%+ on 300 sends each, promote B').
3. For each test, the realistic minimum volume to see signal — and if my sending volume can't support a test, say so and give me the sequential alternative.
4. A results log template I can keep in a spreadsheet: hypothesis, variants, dates, sends, replies, positive replies, verdict.

Avoid: testing two variables at once, subject-line tests before offer tests, declaring winners on under 200 sends per variant, and treating open rates as the success metric when reply and positive-reply rates are what matter.

Before you write anything, interview me. Ask me these questions ONE AT A TIME, waiting for my answer each time:
1. What are your current numbers — sends per week, reply rate, positive reply rate, meetings booked?
2. Paste your current best-performing message (email or call opener).
3. What have you already tested, and what did you conclude?
4. What's your total addressable list size — can you afford to burn contacts on tests?
5. What channel is this for — email, calls, LinkedIn, or a mix?

Once you have my answers, produce the plan. If my volume is too low for parallel A/B tests, restructure the roadmap as sequential tests and say what that costs in confidence.

How to use it

  1. 1

    Copy the prompt into Claude, ChatGPT, or any LLM.

  2. 2

    Pull your real numbers per campaign before answering — blended averages hide the signal.

  3. 3

    Paste your actual current message in question 2, not a cleaned-up version.

  4. 4

    Set up the results log before week 1; untracked tests are just noise with extra steps.

  5. 5

    After each test concludes, tell the model the result and let it re-sequence the remaining weeks.

Best practices

  • If reply rate is under 1%, test lists and offers, not copy — no phrasing rescues the wrong audience or a weak offer.

  • Judge on positive reply rate, not raw replies; a variant that provokes 'unsubscribe' replies is not winning.

  • Keep a 'do not retest' list of concluded experiments — teams forget and relitigate the same subject-line debate quarterly.

  • Low volume? Run each variant for two weeks sequentially on matched list segments rather than pretending you can split-test.

Example: what this looks like in practice

An SDR manager at a 50-person cybersecurity firm sends 4,000 cold emails a month at a 1.8% reply rate and has been A/B testing subject lines for two quarters with no movement. She answers the interview and pastes her control email. The model diagnoses the hierarchy: replies are borderline but positive replies are just 0.4%, suggesting an offer-level problem, not copy. Week 1-3 tests the offer (free pen-test summary vs. generic demo ask) at 500 sends per variant; weeks 4-5 test the angle on the winner; only weeks 6-8 touch openers and CTAs. The offer test alone moves positive replies to 1.1%, worth more than a year of subject-line tweaks.

Best fit

Roles
Sales LeaderMarketerSDR / BDRFounder / CEO
Company size
Startup (1–10)SMB (11–50)Mid-market (51–500)
Audience
B2B
Industries
Any industry
Works with
Any LLM
Difficulty
Intermediate

This prompt is one gear in a bigger machine. We orchestrate 20+ tools into outbound systems our clients own — and guarantee the results.

Apply for a Pilot Spot →
Prompt FAQ

Frequently asked questions

The list and the offer, in that order — not the subject line. Audience and offer determine 80% of results; copy polish works only when those are right. If your reply rate is under 1%, subject-line tests are rearranging deck chairs. This prompt diagnoses which level is broken from your numbers, then sequences tests from highest-impact down.