How to run a B2B SaaS onboarding email test
Onboarding emails decide whether a new B2B SaaS signup activates or churns silently. This guide shows how to set up a repeatable B2B SaaS onboarding email test using marginal, the email MCP server that lets AI agents generate variants, launch subject line A/B tests, and read back lift metrics.
Why onboarding emails deserve their own test cadence
Onboarding is a sequence, not a single send. Welcome emails, setup nudges, and feature-discovery messages each compete for attention in a busy work inbox, and the subject line is what gets them opened in the first place. Testing them as a system tells you which framing actually moves activation.
A structured B2B SaaS onboarding email test isolates one variable at a time so you can trust the result. Subject lines are the highest-leverage starting point because they directly control open rate, which gates every downstream click and in-product action.
- Welcome email: does a personal sender voice beat a product-led headline?
- Day-2 setup nudge: urgency framing vs. value framing
- Feature reveal: outcome-driven vs. capability-driven subjects
- Re-engagement: question vs. statement openers
Setting up the test with marginal's email MCP server
marginal is a hosted email MCP server, so your AI agent connects over the endpoint at https://marginal.sh/mcp using a Bearer API key. There is nothing to self-host. Once connected, the agent has four tools to drive the whole loop: generate_variants, launch_test, get_results, and recommend_next.
For an onboarding test, you describe the target email and audience, and the agent calls generate_variants to draft candidate subject lines. It then runs launch_test to ship a subject line A/B test across your onboarding cohort with managed sends and open/click tracking built in.
- generate_variants — draft subject line candidates for each onboarding step
- launch_test — start a managed subject line A/B test on the cohort
- get_results — pull open/click data and within-test lift metrics
- recommend_next — decide the winner or propose a follow-up variant
Reading lift and iterating with an AI agent
Where this approach pays off is the feedback loop. After a send, get_results returns within-test lift metrics so you can see how the challenger subject performed against the control without leaving your editor. Because the workflow lives inside your MCP client, AI agent email marketing becomes a tight cycle rather than a quarterly project.
Use recommend_next to keep momentum. The agent can suggest whether to promote a winner, retest a close call, or generate a fresh angle for the next onboarding step — turning a one-off experiment into an always-on optimization habit.
- Compare open and click lift per onboarding step
- Promote winners and roll them into the live sequence
- Queue the next subject line A/B test automatically
- Free tier covers 100 experiments/month to get started
Connect your client and start testing
marginal works with the agent tools developers already use, including Cursor, Claude Desktop, Claude Code, Codex, Windsurf, Cline, Continue, and Zed. Point any of them at the marginal MCP server, add your API key, and your agent can run a full B2B SaaS onboarding email test from a single prompt.
See the docs at https://marginal.sh/docs/ for tool schemas and example prompts to wire the workflow into your onboarding stack.
- 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