Recipes / Message test results to sequence copy: rewrite outbound steps from what buyers said

Message test results to sequence copy: rewrite outbound steps from what buyers said

Wynter + Reply.io

IntermediateMarketingSDRs

Who it's for

Marketers and SDR leads who ran a Wynter message test and want the outbound sequence to reflect what buyers actually said, line by line. Reply.io can be swapped for lemlist, Instantly or Smartlead.

What you get

Strongest and weakest lines, and one revised step waiting for approval. Sample output: every company and person in it is fictional.

Test: 'Outbound message: RevOps leaders' (completed Oct 9), 30 respondents
For: Fernhill Software (fictional), sequence 'Q4 RevOps outbound' (status: active)

Strongest lines
1. "Catch stale deals before the forecast call" (Q2 clarity: 24 of 30 rated it clear)
   "This is the exact pain on Monday mornings." (VP RevOps)
   "Specific. I know what it does." (Sales Ops Manager)

Weakest lines
1. "AI-native pipeline intelligence platform" (Q3 relevance: 7 of 30 rated it relevant)
   "Every vendor says this. Means nothing to me." (Director, Revenue Operations)
   "I stopped reading here." (Head of Sales Ops)

Step 1 email, old vs new (not changed yet)
Old subject: AI-native pipeline intelligence for Quarry Lane Analytics (fictional)
New subject: Stale deals before Thursday's forecast?
Old opener: "Fernhill is the AI-native pipeline intelligence platform..."
New opener: "Most RevOps teams find stale deals the morning of the forecast call. Fernhill flags them the week before."
Why: weakest line removed (7 of 30 relevant); strongest line moved up (24 of 30 clear).

Next: Reply 'approve pause' to pause the sequence while you paste in the new copy. Nothing is started or resumed.

Time and cost

Estimate About 20–30 minutes, most of it reading quotes and the rewrite.

Assumptions: One completed message test with about 30 respondents and one 4-step sequence.

Exact credit costs depend on your plan with each vendor. Check the tool pages before a large run.

MCP servers

Every server is confirmed in: ClaudeChatGPTClaude Code

Set up in a chat app

Every server in this recipe is confirmed for Claude and ChatGPT. Connect each one, then paste the prompts below into one conversation.

Developer setup

Auth setup and mcp.json for Claude Code and Cursor
  • Wynter: Sign in with OAuth in your client (run /mcp in Claude Code). Do it once interactively, and the client stores and refreshes the token.
  • Reply.io: Sign in with OAuth, or pass an API key as a Bearer token for headless use.

Put this in .mcp.json (Claude Code, project root) or ~/.cursor/mcp.json (Cursor). Swap the example keys for your own, and don't commit real ones. Remove the // comment lines if your client requires strict JSON.

{
  "mcpServers": {
    "wynter": {
      "url": "https://api.wynter.io/mcp"
    },
    "reply.io": {
      "url": "https://mcp.reply.io/"
      // headless alternative: "headers": { "Authorization": "Bearer YOUR_REPLY.IO_API_KEY" }
    }
  }
}

Step-by-step prompts

  1. Find my completed Wynter message test {test name}. Show me each question, the per-question stats, and the respondent breakdown by role and seniority. Don't summarize the answers yet.
  2. From the verbatim answers, pick the 3 strongest and 3 weakest lines of our copy. For each, give the per-question stat and quote 2 respondents word for word with their role. Only use answers that are in the test.
  3. Find my Reply.io sequence {sequence name} and tell me its status. Here is the current copy of its steps: {current step copy}. Draft revised steps that keep what tested well and drop or fix what tested badly. Show old and new side by side, with the Wynter quote behind each change. Don't change anything in Reply.io.
  4. If the sequence is active, pause it only after I approve, so prospects don't get the old copy while I update it. Never start or resume it. I'll paste in the approved copy and restart it myself.

Run it on a schedule (optional)

Where this goes wrong

Gotchas

Download

Get the prompts, guardrails and expected output as one Markdown file. Paste it into a Claude Project or ChatGPT project's instructions, or save it as CLAUDE.md for Claude Code.

Download message-test-to-sequence.md