How can I use seed lists to test deliverability variations?

You've crafted two versions of a campaign and you're wondering which one will actually reach the inbox. Engagement metrics only tell part of the story. Seed list testing shows you what's happening before the open even happens.

A seed list is a collection of test accounts you control (or access through a service) at the major mailbox providers. Think a handful of addresses at Gmail, Outlook, Yahoo Mail, and others. You include them in your send, then check where the mail landed. Inbox, spam, or nowhere at all.

Here's how to use them to test deliverability variations specifically.

Step 1: Get your seeds

You can build a small DIY seed list by creating accounts yourself at each major provider. It's tedious but free. For more coverage, tools like GlockApps and Mailtrap provide managed seed networks that cover dozens of inbox providers and return placement reports automatically. The trade-off is cost versus scale. A DIY list of 10 addresses gives you directional data. A managed service gives you statistically cleaner results.

Step 2: Split your seeds across both variants

But if you're testing Variant A against Variant B, split your seed addresses evenly between the two groups. Send Variant A to its seed addresses and Variant B to its. Keep the send timing close together so filtering conditions are comparable. Sending one variant on a Tuesday and the other on a Friday introduces noise you don't want.

Step 3: Check placement, not just opens

Log into each seed account (or review your service's report) and note where each email landed. You're looking for inbox placement rate by provider. If Variant A lands in the inbox at Gmail but Variant B hits the spam folder, that's a signal worth digging into. Content differences, link choices, image-to-text ratio, and subject line phrasing can all influence how spam filters react to each version.

Step 4: Cross-reference with engagement

Seed data alone doesn't tell you which variant your real subscribers prefer. Combine it with your A/B engagement data. The ideal outcome is a variant that places well in the inbox AND generates stronger opens and clicks. If one variant has better placement but lower engagement, you've got a real trade-off to think through. And strong engagement itself feeds back into deliverability over time, so neither signal should be ignored.

A few honest limitations to keep in mind

Seed accounts don't behave like real recipients. They have no engagement history with your domain, no prior relationship, no previous opens or clicks. That means seeds tend to be slightly harder to impress than a warmed-up subscriber segment. If your seeds show inbox placement, that's a positive signal. If they show spam, take it seriously. But don't read seeds as a perfect 1-to-1 mirror of what your full list will experience (of course, nothing in deliverability ever is quite that clean).

Sample sizes are also small. If you have 10 seed addresses per variant and three hit spam, you're looking at 30% spam placement from a very small sample. Directionally useful, not statistically final.

And if your seed results are raising red flags and you're not sure what's causing the filtering, our SOS hotline is free and we're happy to take a look with you.