Online Courses Upsell Email Test
Selling the next tier, a bundle, or a cohort upgrade to existing students works best when the message is tuned. This guide shows how to run an online courses upsell email test with marginal, the email MCP server your AI agent calls to draft variants, send, and read results.
Why upsell emails for online courses need testing
Course students convert on upsells for reasons that are hard to guess: some respond to outcome framing ("finish the certification"), others to scarcity ("cohort closes Friday"), and others to price anchoring. A single subject line rarely wins across all of them.
An email A/B test removes the guesswork. Instead of shipping one upsell blast, you compare subject lines and angles against real opens and clicks from your own learners, then double down on what actually moves enrollments.
- Upgrade prompts: free trial to paid, single course to full path
- Bundle offers: pair a completed course with a logical next module
- Cohort or live-session upsells with time-bound enrollment windows
- Renewal and continuation nudges for subscription-based programs
Running the test with marginal
marginal is a hosted MCP server, so your AI agent drives the whole loop through four tools. You describe the upsell campaign in plain language and the agent handles variant generation, the send, and the readout.
Because the test, tracking, and lift metrics live inside marginal, you don't wire up a separate analytics pipeline. The within-test lift tells you which subject line earned more opens and clicks for this specific audience.
- generate_variants — draft several subject line and copy angles for the upsell
- launch_test — split your student list and send the variants
- get_results — pull open/click tracking and within-test lift
- recommend_next — let the agent suggest the follow-up based on results
A subject line A/B test workflow for upsells
Start with the customer state. A student who just finished a beginner course is in a different mindset than one who lapsed three months ago, so the upsell angle and subject line should reflect that segment. Ask your agent to generate variants tuned to each scenario.
Keep each subject line A/B test focused on one variable at a time — outcome vs. urgency, or curiosity vs. specificity — so the lift you measure is interpretable. Then feed the winner into your next send.
- "You're 1 module from your certificate" vs. "Unlock the advanced track"
- Test discount framing against value framing on the same offer
- Compare a deadline-driven subject against an evergreen one
- Use recommend_next to plan the follow-up to non-openers
Connect your AI agent
AI agent email marketing with marginal works in the tools you already use — Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Zed, and OpenAI Codex. Point your client at the endpoint, authenticate with a Bearer API key, and ask it to run the upsell test.
The free tier covers 100 experiments per month, which is plenty to iterate on upsell messaging across course segments. See https://marginal.sh/mcp for setup and https://marginal.sh/docs/ for tool details.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key (marg_live_...)
- Registry: sh.marginal/mcp
- 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