Testing Referral Program Emails for Beauty Brands
Referral programs live or die on the invite email. This guide shows beauty teams how to run a beauty referral program email test with marginal — generating subject line variants, launching managed sends, and reading within-test lift to find the message that actually drives shares.
Why referral emails matter in beauty
Beauty buyers trust friends more than ads. A well-timed refer-a-friend email — sent after a first order or a glowing review — turns loyal customers into a low-cost acquisition channel. But generic copy gets ignored, and small wording changes can swing share rates dramatically.
That makes the referral program email an ideal candidate for an email A/B test. The reward is concrete, the audience is engaged, and the subject line carries most of the open-rate weight.
- Trigger invites after a positive moment: delivery, repeat purchase, or a left review
- Test how you frame the reward — free product vs. percentage off vs. shared credit
- Beauty audiences respond to specifics: a named shade, a hero serum, a routine
Designing the subject line A/B test
Most of your referral test budget should go into the subject line A/B test. Open rate is the first gate — if the invite isn't opened, the reward never gets seen. Keep the body consistent across variants so you isolate the subject's effect.
Good beauty referral subject lines tend to cluster around a few angles. Test one angle per variant rather than mixing several ideas into one line.
- Generosity: 'Give your friend 20% off their first glow-up'
- Self-interest: 'Get a free serum for every friend you refer'
- Social proof: 'The product your friends keep asking about'
- Mutual: 'You both get $15 — share the love'
Running the test with marginal
marginal is a hosted email MCP server, so your AI agent can run the whole experiment from your editor without you wiring up infrastructure. Point your client at the endpoint, drop in your Bearer key, and the four tools cover the full loop.
Use generate_variants to draft referral subject lines, launch_test to send them and split your audience, get_results to compare open and click metrics with within-test lift, and recommend_next to decide the follow-on test. The free tier includes 100 experiments per month — enough to iterate on referral messaging steadily.
- generate_variants — produce on-brand referral subject lines
- launch_test — managed send with open/click tracking
- get_results — within-test lift on the winning variant
- recommend_next — what to try in the following round
Connecting your AI agent
AI agent email marketing works best when the agent can act, not just suggest. marginal connects to clients like Cursor, Claude Code, Claude Desktop, Windsurf, Cline, Continue, Zed, and OpenAI Codex via the MCP endpoint at https://marginal.sh/mcp.
Once connected, you can ask your agent in plain language to draft, launch, and evaluate a beauty referral campaign — then loop in the recommendation for the next test. See the docs at https://marginal.sh/docs/ for setup details and tool reference.
- Registry: sh.marginal/mcp
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key (marg_live_...)
- Hosted — nothing to self-host or maintain
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