Testing Review Request Subject Lines in Continue
Continue brings AI agents into your editor, and with marginal's email MCP server it can also run live subject line experiments. This guide shows how to draft, test, and ship review request subject lines without leaving your workflow.
Why review requests need a subject line test
Review request emails live or die in the inbox. A customer who loved your product still has to notice the message before they leave a star rating, so the subject line carries most of the weight. Small wording changes — naming the product, asking a direct question, or adding the order context — can move opens enough to change your total review volume.
Rather than guessing, run a subject line A/B test on a real send. marginal handles the split, the delivery, and the tracking, while Continue lets you describe what you want in plain language and review the results inline.
- Review prompts are time-sensitive — opens early after delivery matter most
- Personalized vs. generic phrasing is a classic, high-impact variable to test
- Within-test lift metrics tell you which framing actually earns more opens
Connecting marginal to Continue
marginal is a hosted email MCP server, so there is nothing to deploy. Point Continue at the endpoint https://marginal.sh/mcp and authenticate with your Bearer API key (it starts with marg_live_). Once connected, Continue can call the four marginal tools: generate_variants, launch_test, get_results, and recommend_next.
Because the server is hosted, you don't manage infrastructure or wire up a separate provider. You bring an API key, and the AI agent email marketing loop runs through Continue.
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer API key in the marg_live_ format
- Registry name: sh.marginal/mcp
- Free tier covers 100 experiments per month for early testing
Running the test from your editor
Start by asking Continue to generate review request subject lines, then launch a split test against your customer list. A typical flow keeps each step explicit so you can sanity-check the agent's choices before anything sends.
- Use generate_variants to produce several review request subject lines — mix personalized, question-style, and product-named angles
- Call launch_test to send the variants across an even split with open and click tracking enabled
- Pull get_results to see opens, clicks, and within-test lift once data accumulates
- Ask recommend_next to surface the leading subject line and suggest a follow-up variation
Reading results and iterating
After the test runs, get_results gives you per-variant open and click rates plus the lift of each variant relative to the others. For review requests, watch opens closely since the click into a rating flow depends on a customer first opening the email.
Use recommend_next inside Continue to turn one win into a habit. Promote the winning subject line, then spin up the next subject line A/B test with a fresh angle — different urgency, a new incentive mention, or tighter personalization — and keep compounding gains over each send.
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