Testing Referral Program Emails for Marketplace Sellers
Referral invites convert best when the subject line earns the open and the offer lands fast. This guide shows marketplace sellers how to run a referral program email test with marginal, the hosted email MCP server your AI agent can drive end to end.
Why referral emails need testing on marketplaces
Marketplace sellers operate in crowded inboxes. Your buyers and fellow sellers already receive order updates, payout notices, and promotions, so a referral invite has to stand out before anyone reads the reward terms. The right subject line A/B test can move open rates several points, and on a list of active marketplace sellers that compounds into more referral signups per send.
Referral programs also have a unique structure worth testing: the ask, the reward framing, and the urgency. A subject that leads with the payout ('Earn $25 per seller you refer') often behaves differently than one that leads with community ('Help another seller get started'). You won't know which fits your audience until you put both in front of real recipients.
- Test reward-led vs. relationship-led subject lines
- Compare one-time bonus framing against recurring commission
- Check whether naming the marketplace in the subject lifts opens
Setting up the email A/B test with marginal
marginal exposes four MCP tools your agent calls directly: generate_variants, launch_test, get_results, and recommend_next. For a referral campaign, you start by having the agent generate subject line variants tuned to your seller audience, then launch the managed send and let marginal handle open and click tracking automatically.
- generate_variants — produce subject line options for the referral invite
- launch_test — send the A/B test to your marketplace seller segment
- get_results — pull open/click rates and within-test lift metrics
- recommend_next — get a data-backed suggestion for the follow-up send
Running it from your AI agent
Because marginal is a hosted email MCP server, there's no infrastructure to stand up. Point a supported client at https://marginal.sh/mcp, authenticate with your Bearer API key, and ask the agent to run the referral subject line A/B test in plain language. AI agent email marketing means the loop — draft, launch, measure, iterate — happens inside the same tool where you already work.
Clients that connect today include Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Zed, and OpenAI Codex. The free tier covers 100 experiments per month, which is plenty to iterate on a referral campaign across several seller segments before scaling.
- Connect via the endpoint at https://marginal.sh/mcp
- Authenticate with a marg_live_ API key
- Read tool details in the docs at https://marginal.sh/docs/
Reading results and iterating
After the send, get_results returns within-test lift so you can see which referral subject line drove more opens and clicks among marketplace sellers, not just a raw rate. Use recommend_next to decide your follow-up — whether to resend the winner to non-openers or test a new reward angle entirely.
Treat each referral push as a learning cycle. Over a few experiments you'll build a reliable picture of what motivates your sellers to invite others, and your agent can carry those insights into the next campaign automatically.
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