Testing upsell subject lines in Claude Code
Upsell campaigns live or die on the subject line. With marginal connected to Claude Code, you can draft, test, and measure upsell subject lines without leaving your editor — your agent handles the variant generation and the A/B test, and reports back the winner.
Why run upsell subject line tests from Claude Code
Upsell messages target existing customers, so a small lift in opens compounds quickly across a warm list. Claude Code is already where you reason about copy, segments, and timing — wiring in an email MCP server means you can move from idea to live subject line A/B test in one conversation.
marginal exposes its tools over MCP, so Claude Code can call them directly. You describe the upsell angle, and the agent does the rest using the four core tools.
- generate_variants — produce multiple upsell subject line angles in one call
- launch_test — split-test variants against your audience
- get_results — pull open/click rates and within-test lift
- recommend_next — get a data-backed suggestion for the next round
A practical upsell workflow
Frame the prompt around the offer and the customer's current state. For example: "Generate five upsell subject lines nudging Starter-plan users to Pro, then launch a subject line A/B test." Claude Code calls generate_variants, then launch_test, and you have a live experiment in seconds.
- Anchor each variant on a distinct hook: value gained, time saved, social proof, scarcity, or a direct ask
- Keep the audience tight — recent active users convert better on upsell offers
- Let the test run long enough to clear daily open-rate noise before checking lift
- Ask Claude Code to call recommend_next so the following send builds on the winner
Reading the results that matter
marginal reports open and click tracking plus within-test lift, so you can see which upsell subject line genuinely outperformed rather than guessing from raw totals. Ask Claude Code to summarize get_results and call out the lift percentage and confidence in plain language.
Because the metrics are scoped to the test, you avoid comparing across different audiences or send times — the comparison stays apples to apples within a single experiment.
Connecting marginal to Claude Code
marginal is a hosted MCP server, so there's nothing to deploy or self-host. Add the endpoint https://marginal.sh/mcp with your Bearer API key (marg_live_...) and Claude Code can call the tools immediately.
The free tier covers 100 experiments per month — plenty to validate an upsell subject line test cadence. See the docs at https://marginal.sh/docs/ for tool schemas and setup details for AI agent email marketing.
Get started with marginal
marginal is a hosted email marketing MCP server at marginal.sh. Sign up free, create an API key, and connect https://marginal.sh/mcp from Cursor, Codex, or Claude Desktop.
- 100 experiments/month on the free tier
- Four MCP tools: generate_variants, launch_test, get_results, recommend_next
- Listed in the MCP Registry as sh.marginal/mcp