Testing Referral Program Subject Lines in Continue
Referral campaigns live or die on the open. With marginal's email MCP server connected to Continue, you can draft, test, and ship referral program subject lines without leaving your editor — the AI agent handles the variant generation and A/B test orchestration for you.
Why referral emails need their own subject line test
Referral program messaging has a tricky job: it has to nudge an existing customer to vouch for you to someone else. The incentive, the social proof, and the call to action all compete for attention in the subject line, and small wording changes move open rates more than you'd expect.
Rather than guessing which angle wins, a subject line A/B test lets the data decide. Inside Continue, you can ask the agent to spin up several referral angles and pit them against each other on a slice of your list before committing to a full send.
- Test incentive-forward lines ("Give $20, get $20") against curiosity-driven ones
- Compare urgency framing versus relationship framing
- Validate emoji and personalization choices with real opens, not opinions
How marginal works inside Continue
marginal is a hosted MCP server, so there's nothing to deploy. Once it's added to Continue with your Bearer API key, the agent gains four tools for running referral program subject line experiments end to end.
A typical flow is conversational: you describe the referral campaign, the agent calls generate_variants for a set of candidate subject lines, launches the test, and reports back on which one is pulling ahead.
- generate_variants — produce candidate referral subject lines from your brief
- launch_test — start a managed subject line A/B test with real sends
- get_results — pull open and click tracking plus within-test lift metrics
- recommend_next — get a data-backed suggestion for the next iteration
A practical workflow for referral subject lines
Start by telling Continue the goal of your referral program email and any constraints — incentive amount, tone, and audience segment. The AI agent uses that context to generate distinct variants rather than minor rewordings, which gives the subject line test more signal.
After launch_test runs, check back with get_results to see open and click performance and the within-test lift between variants. When a winner emerges, recommend_next can propose a follow-up test so your referral subject lines keep improving across campaigns.
- Brief the agent with your referral offer and audience
- Generate 4–6 contrasting subject line variants
- Launch the A/B test on a sample, then scale the winner
- Iterate using lift metrics and the next-step recommendation
Getting set up
Point Continue at the marginal endpoint at https://marginal.sh/mcp and authenticate with your marg_live_ API key. Because marginal is fully hosted, you don't connect any external ESP keys to get started — sends and tracking are managed for you.
The free tier covers 100 experiments per month, which is plenty to validate a referral program subject line strategy. Full tool references and configuration details live in the docs at https://marginal.sh/docs/.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key (marg_live_...)
- Free tier: 100 experiments/month 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