Seed lists: what they are for and how to build one properly

A seed list is a set of mailboxes you own at every provider your real recipients use. You include them in a send — or send them the identical message — and then look at where each copy landed: inbox, a tab, Junk, or nowhere. That is the only direct observation of folder placement available to a sender, because no provider reports placement back to you. Your ESP's delivery rate cannot substitute: a message accepted with 250 OK and filed in Junk counts as delivered.

Build the list from your recipients, not from a template

The mistake is assembling seeds from whatever is easy — five Gmail accounts and one Outlook — and then reporting an average. The list should mirror the provider distribution of the people you actually mail.

1. Export your recipient domains and group them by MX record, not by domain name. Thousands of distinct company domains usually resolve to a handful of platforms: Google Workspace, Microsoft 365, and then a long tail of hosts.
2. Create or acquire mailboxes on each platform in that distribution, weighted roughly by share.
3. Include the consumer providers your mail reaches: Gmail, Outlook.com, Yahoo, iCloud, and the regional providers that matter in your markets.
4. Include at least one mailbox behind a security gateway if your buyers are enterprises, because the gateway, not the mailbox provider, is what filters that mail.
5. Keep the mailboxes looking like mailboxes: real names, some history, occasional human use. A mailbox that has only ever received test mail is an unusual recipient.

Send to it without distorting the result

• Send the production message: same template, same links, same footer, same From name. A stripped "clean" test measures a message you will not ship.
• Send from the mailbox that will do the sending, not a test account.
• Spread the send. A burst at one provider's seeds is itself a rate signal.
• Record provider, folder and Message-ID for every seed so you can read headers later.
• Repeat weekly. One test is a snapshot; the trend is the instrument.

Then read the headers of anything that landed badly. Authentication-Results tells you whether SPF and DKIM passed and aligned; the Received chain shows added hops; Microsoft often leaves behind X- headers that hint at the verdict.

What a seed list cannot do

It cannot tell you about individual recipients, because filtering is per-recipient and influenced by that person's history with you. It cannot produce statistically tight numbers from a dozen mailboxes — treat a 15-point week-on-week drop at one provider as a signal and a two-point wobble as noise. And at Gmail it has a specific blind spot worth knowing: the Promotions tab carries the INBOX label, so any measurement built on reading labels through the API scores a Promotions landing as a clean inbox hit. Check a tab by eye on at least one seed before you trust the number.

It also cannot replace the signals you still get directly. Google Postmaster Tools retired its domain and IP reputation charts, but spam rate, authentication results and delivery errors remain, and spam rate under 0.3% — ideally under 0.1% — is the threshold that matters for Gmail bulk senders. Microsoft SNDS gives per-IP data for IPs you own. Seeds complement those; they do not replace them.

What not to do

Do not use a seed list as a warm-up mechanism. Seeds exist to be measured. Sending volume at them, or having them open, click and reply to manufacture positive signals, is the practice Validity's Heatwave blocklist was built to catch — it launched on 3 September 2026 targeting synthetic reputation warming, lists domains rather than IPs, launched with over a million domains, and feeds live filtering, with lookup at lookup.validity.tools. Its thresholds are not public, so no tool, ours included, can promise exemption. Keep measurement and warm-up separate in your head and in your sending.

Do not seed only providers you find convenient. Do not reuse the same five addresses for a year without using them like mailboxes. Do not read a single message's placement as a verdict. Do not include seeds in a send that also has a high bounce rate and then blame filtering, when you are really measuring list quality. And do not let a vendor's composite "deliverability score" stand in for placement; it bundles unrelated checks into a number that moves for reasons unconnected to where your mail lands.

Reading the trend

The value shows up on the week something changes. You shipped a new template on Thursday; Outlook placement halves on Friday. You added a link shortener; three providers slip at once. That causal link between a change and a measurement is what seeds buy you, and it is worth more than any absolute number.

MailPilot runs a seed-inbox network for exactly this and reports real per-provider inbox placement, alongside SPF, DKIM and DMARC checks and per-mailbox reputation recorded daily — so placement data arrives continuously rather than only when you remember to run a test.