Building a Seed Account List for Inbox Placement Testing

What a seed list is for

A seed list is a set of mailboxes you control across the providers your real recipients use. You send a campaign to them alongside the real send and observe where each copy landed: inbox, promotions, spam, or nowhere at all. It is the cheapest way to notice a placement problem before your engagement metrics reveal it a week later.

The value comes from spread, not size. Twenty accounts covering five providers tells you more than two hundred Gmail accounts, because placement decisions are made per provider and the differences between them are exactly what you are trying to see.

Match the mix to your actual recipient base

Pull the provider distribution from your own list rather than copying a generic split. If seventy percent of your recipients are on Google and you weight your seed list evenly across five providers, most of your signal is coming from providers that barely matter to you.

MailMaker covers Gmail, Yahoo, Outlook, Proton, and iCloud, which maps to the bulk of most consumer lists. Gmail and Yahoo cost one credit each, Outlook and Proton two, iCloud three, so weighting toward your dominant provider is also usually the cheaper choice.

• Derive the split from your own recipient data
• Weight toward your dominant provider
• Keep at least a few accounts on every provider you send to
• Re-derive the split when your list composition shifts

New accounts skew the result, and you should account for it

This is the caveat most deliverability guides skip. A freshly created mailbox has no engagement history, no address book, and no prior interaction with your domain. Providers weigh all of those when deciding placement, so a brand new seed account is not a neutral observer.

In practice a new seed list tends to read more pessimistically than your real audience for warm senders, and more optimistically for cold ones. Use the list to detect changes over time rather than to estimate an absolute inbox rate, and give the accounts some age and interaction before treating their placement as meaningful.

Running the test so the result means something

Send to the seed list at the same time and through the same infrastructure as the real campaign. A separate test send from a different IP or with a different authentication setup measures the test send, not the campaign, and teams discover this after acting on the wrong number.

Record placement per provider per send and track the trend. A single reading tells you very little. A provider that moves from inbox to promotions across three consecutive campaigns is a signal worth acting on.

• Same send time, same infrastructure, same authentication
• Record per provider, not as a blended average
• Compare against your own history, not an industry benchmark
• Investigate trends, not single readings

A stale seed list fails quietly

Seed accounts get suspended, providers change their filtering, and your recipient mix drifts. None of these announce themselves. A seed list assembled two years ago and never revisited will keep producing numbers long after those numbers stopped describing anything.

Put a refresh on the calendar. Replace suspended accounts, re-derive the provider split from current recipient data, and retire accounts that have been sitting unused long enough that their behaviour no longer resembles a real recipient.