Testing Your B2B Software Welcome Series with marginal
A new B2B software signup is your warmest moment — and the most under-optimized email sequence most teams ship. This guide shows how to run a B2B software welcome series email test with marginal, the hosted email MCP server, so your AI agent can experiment on subject lines and measure real lift instead of guessing.
Why welcome series tests matter for B2B software
In B2B software, the welcome series sets the tone for activation. The first few emails decide whether a trial user logs back in, connects a data source, or invites a teammate. Small subject line changes — clarity over cleverness, value over greeting — often move open rates more than any later campaign.
Because the welcome series fires automatically on signup, it's an ideal place for an ongoing email A/B test: high, steady volume and a clear downstream goal. Rather than rewriting the whole flow at once, you test one variable per email and let the data compound.
- Welcome opens correlate strongly with day-7 activation in B2B software.
- Each welcome email is a repeatable test surface, not a one-off blast.
- Subject line A/B tests are the fastest lever before touching body copy or timing.
Mapping a welcome series to marginal experiments
marginal exposes four MCP tools — generate_variants, launch_test, get_results, and recommend_next — that map cleanly onto a multi-email welcome flow. Treat each email in the series as its own experiment, then chain results into the next decision.
A typical B2B software welcome series test might look like this across the first three sends.
- Email 1 (instant): test a benefit-led subject vs. a plain 'Welcome to X' subject line A/B test.
- Email 2 (day 1): test whether naming a key integration or use case lifts opens.
- Email 3 (day 3): test social proof framing vs. a direct activation nudge.
- Use get_results to read open/click tracking and within-test lift before promoting a winner.
- Call recommend_next to let the agent pick the follow-up variant automatically.
Running it through an AI agent
marginal is a hosted email MCP server, so any MCP-capable client can drive the whole loop. Point Cursor, Claude Code, Claude Desktop, Windsurf, Cline, Continue, Zed, or OpenAI Codex at https://marginal.sh/mcp with a Bearer API key and your AI agent email marketing workflow can generate variants, launch the test, and read lift without you leaving the editor.
Because there's nothing to self-host, you skip infrastructure entirely. The agent calls generate_variants for fresh subject line ideas tuned to a B2B software audience, launches the managed send, and reports back which welcome variant won.
- Endpoint: https://marginal.sh/mcp · Auth: Bearer key (marg_live_...).
- Free tier covers 100 experiments/month — enough to test a full welcome series repeatedly.
- Managed sends and tracking are built in; no separate delivery setup required.
Reading lift and iterating
Once a welcome email test has accumulated sends, get_results returns within-test lift metrics so you can compare variants on equal footing instead of comparing across weeks of changing traffic. For B2B software, watch click-through on the activation CTA, not just opens — an attention-grabbing subject that doesn't drive logins isn't a win.
When a variant clears your bar, promote it and let recommend_next propose the next thing to test in the series. Over a few cycles, your welcome flow becomes a continuously tuned funnel rather than a static template. See https://marginal.sh/docs/ for tool details.
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