How to Run a Question Mark Subject Line Test
Curiosity-driven subject lines often end in a question mark — but does the punctuation actually move open rates for your list? Instead of guessing, run a controlled question mark subject line test and let real opens settle the debate.
Why test the question mark at all?
A question mark signals an open loop — it invites the reader to resolve the curiosity by opening. But the effect is list-specific. What lifts opens for a B2B SaaS newsletter can flatten engagement on a transactional ecommerce blast. The only reliable answer comes from a subject line A/B test on your own audience.
The cleanest experiment isolates the punctuation. Hold the words constant and vary only whether the line ends in a question mark, so any difference in opens is attributable to the variant rather than the copy.
- Variant A: "Your invoice is ready."
- Variant B: "Your invoice is ready?"
- Keep word count, casing, and emoji identical between arms
- Measure within-test open lift, not historical averages
Running it through marginal's email MCP server
marginal is a hosted email MCP server that connects your AI agent directly to subject line testing. Once configured, your agent can generate question-mark and declarative variants, launch the split, and read back open and click metrics — all without leaving the chat.
The relevant MCP tools map cleanly onto a punctuation experiment:
- generate_variants — produce matched question-mark and statement versions
- launch_test — split your list and send both arms
- get_results — pull open/click tracking and within-test lift
- recommend_next — surface the winning variant and suggest a follow-up
Reading the results without fooling yourself
A higher open rate on the question-mark arm only counts if the gap is meaningful relative to your sample size. Small lifts on small segments are noise. marginal reports within-test lift so you compare the two arms against each other under the same send conditions, rather than against a stale baseline.
Treat one result as evidence, not gospel. The question mark may help on promotional sends and hurt on alerts. Repeat the test across campaign types before adopting a blanket rule, and let AI agent email marketing handle the repetition for you.
Where this fits in an agent workflow
Because marginal speaks MCP, the question mark subject line test becomes a single instruction in tools like Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, or Zed. Ask the agent to test punctuation, and it generates the variants, launches the send to https://marginal.sh/mcp, and waits for results.
The free tier covers 100 experiments per month, which is plenty to test the question mark across several campaigns and build a small body of evidence for your own list. See the docs at https://marginal.sh/docs/ to wire it up.
- Endpoint: https://marginal.sh/mcp
- Auth via Bearer API key (marg_live_...)
- Hosted — nothing to self-host or operate
- Free tier: 100 experiments/month
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