Testing Post-Purchase Subject Lines in Continue
Continue puts an AI coding agent right in your editor. Wire it to marginal's email MCP server and that same agent can draft, launch, and measure post-purchase subject line tests without leaving your workflow.
Why post-purchase subject lines deserve their own test
The window right after a purchase is some of the highest-intent attention you'll get. Order confirmations, shipping updates, and review requests all carry strong open rates, but the subject line still decides whether the rest of the message lands. A small lift here compounds across every order.
Post-purchase flows also have clear hooks to test against: order number, product name, delivery ETA, or a thank-you framing. Each of these is a natural axis for a subject line A/B test, and Continue can reason about which angle to try next based on what you sold.
- Confirmation: "Order #1042 is in" vs "Thanks — we've got your order"
- Shipping: "Your package ships tomorrow" vs "On its way: here's your tracking"
- Review request: "How's the {product} treating you?" vs "Got 30 seconds to leave a review?"
- Cross-sell: "Pairs perfectly with your order" vs "You might also like these"
How marginal connects to Continue
marginal is a hosted email MCP server, so there's nothing to deploy. Once Continue is pointed at the endpoint with your Bearer API key, the agent gets four tools: generate_variants, launch_test, get_results, and recommend_next. That covers the full loop of an experiment.
Because it's hosted, you don't manage infrastructure or store provider credentials in your config. You connect to https://marginal.sh/mcp and the agent calls the tools directly.
- generate_variants — produce candidate post-purchase subject lines
- launch_test — start a managed send across variants
- get_results — pull open/click tracking and within-test lift
- recommend_next — get a suggested follow-up experiment
A post-purchase subject line test, end to end
Working through Continue, you describe the email and the audience, then let the agent drive the tools. A typical run starts with generate_variants to get post-purchase subject lines in Continue, launches them with launch_test, and waits for open and click data before reading get_results.
Once a winner shows clear lift, recommend_next proposes the next angle to try — for example moving from a curiosity-framed confirmation line to a benefit-framed shipping line. This keeps your AI agent email marketing iterating instead of stopping at a single test.
- Ask Continue to draft 4 confirmation subject variants and call launch_test
- Let opens and clicks accumulate, then review within-test lift via get_results
- Promote the winner and use recommend_next for the next subject line test
- Repeat across each stage of the post-purchase flow
Getting started and limits
The free tier includes 100 experiments per month, which is plenty to validate post-purchase subject lines across confirmation, shipping, and review-request stages. Full tool reference and setup steps live in the docs at https://marginal.sh/docs/.
Add the marginal MCP server to Continue, run your first subject line A/B test, and let the agent handle generation, sending, and measurement in one place.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key (marg_live_...)
- Free tier: 100 experiments/month
- Docs: https://marginal.sh/docs/
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