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.
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
Copy the prompt into Claude, ChatGPT, or any LLM.
- 2
Pull your real numbers per campaign before answering — blended averages hide the signal.
- 3
Paste your actual current message in question 2, not a cleaned-up version.
- 4
Set up the results log before week 1; untracked tests are just noise with extra steps.
- 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
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 → →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.
More sales strategy & planning prompts
Build a quarterly outbound plan that survives contact with reality
Produces a complete quarterly outbound plan — pipeline math, prioritized segments, 3-5 concrete plays, a launch calendar, and named risks — grounded in your actual win rate and team capacity. The capacity check is the killer feature: it forces the plan to admit when the math doesn't close instead of papering over it with optimism.
Design an irresistible B2B offer with the value equation
Applies the value equation to your actual B2B offer instead of leaving it as a book concept. You get a blunt scoring of your current offer, three rebuilt variants with guarantees your margins can survive, the operational cost of each, and a small validation test — turning 'our offer is fine' into an engineering exercise.
Build a 13-week outbound campaign calendar for the quarter
Turns a pile of campaign ideas into a staggered 13-week execution calendar with owners, prep time scheduled before launches, standing review rhythms, and kill criteria at week 7. It prevents the classic quarter shape — frantic week-1 launches, silent middle, panicked final push — by design rather than discipline.