Cross-Sell Email A/B Testing for DTC Brands
Cross-sell campaigns are where DTC brands quietly leave revenue on the table. This guide shows how to run a structured DTC brands cross-sell email test with marginal, the hosted email MCP server your AI agent can drive directly.
Why cross-sell deserves its own email A/B test
For DTC brands, the second purchase is usually more profitable than the first — you already paid to acquire the customer. A cross-sell email that recommends a complementary product (the case for the phone, the refills for the device) converts on intent that's already there. But generic 'you might also like' copy rarely beats a tested alternative.
Treating cross-sell as a dedicated email A/B test, rather than a one-off blast, lets you isolate what actually moves opens and clicks: the framing of the recommendation, the urgency, and especially the subject line.
- Higher margins than first-purchase acquisition campaigns
- Clear hypotheses to test: bundle framing vs. single add-on
- Subject line A/B test impact compounds across a large post-purchase audience
- Repeatable structure you can rerun per product category
How marginal fits a DTC cross-sell workflow
marginal is a hosted email MCP server that exposes a small set of tools to your AI agent: generate_variants, launch_test, get_results, and recommend_next. Your agent connects to the endpoint at https://marginal.sh/mcp with a Bearer API key — there's nothing to self-host.
That means an AI agent email marketing setup can take a cross-sell brief and produce a full experiment without you hand-writing every variant. The agent drafts subject line options, launches a managed send, and reports open and click tracking with within-test lift metrics.
- generate_variants — produce subject line and copy alternatives for the cross-sell offer
- launch_test — start a managed email A/B test split across your list
- get_results — pull opens, clicks, and lift between variants
- recommend_next — get a data-backed suggestion for the next iteration
Running the test step by step
Start by framing the cross-sell offer for a specific customer segment — for example, recent buyers of a flagship SKU. Ask your agent to call generate_variants with the offer context, then launch_test to split traffic. Once enough sends land, get_results surfaces which subject line A/B test variant won and by how much.
Because marginal tracks within-test lift, you compare variants on equal footing rather than against a stale historical baseline. Use recommend_next to fold the winning angle into the following campaign.
- Define the segment and the complementary product to recommend
- Generate 3–5 subject line variants with distinct angles
- Launch and let opens and clicks accumulate before judging
- Promote the winning variant and iterate on the runner-up's angle
Getting started for free
The free tier includes 100 experiments per month, which is plenty to validate a cross-sell program across several DTC product lines before scaling. marginal works with Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Codex, and Zed, so you can drive tests from whatever agent you already use.
Point your client at the marginal endpoint, authenticate with your API key, and check the docs at https://marginal.sh/docs/ for tool schemas and examples.
- 100 experiments/month on the free tier
- Works across major MCP-capable clients
- Endpoint: https://marginal.sh/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