InboxRatio
Check your domain

Glossary term

Seed list: the controlled mailboxes behind placement data

What a seed list is

A seed list is a set of mailboxes that a tester controls, distributed across the mailbox providers that matter (Gmail, Outlook, Yahoo and others) and added as recipients to a mailing precisely so the tester can open each mailbox afterward and record where the message landed: inbox, a tab like Promotions, the spam folder, or nowhere at all.

The technique exists because the metric that matters is otherwise unobservable. Delivery data from your sending platform tells you a message was accepted — the receiving server said 250 OK during the SMTP transaction. It does not tell you what happened next, because filtering into spam happens after acceptance and no provider reports folder placement back to senders per message. Inbox placement is invisible from the sending side; seed mailboxes make a sample of it visible from the receiving side.

Seed lists are the instrument behind most published deliverability testing, including ours. InboxRatio's own protocol is a seed-list design — the short version is on the how we test page, and our test cycles are announced as they complete; until a cycle publishes, no seed-derived numbers appear on this site.

How a seed list differs from your real audience

A seed list is a measurement instrument, and the measurement is honest only if you stay clear about what the instrument is not: it is not your subscribers, and the difference is systematic rather than random.

Engagement history. Your real recipients have relationships with your mail — opens, replies, deletes-without-reading, prior complaints. Modern filters, Gmail above all, weigh that per-recipient history heavily. A seed mailbox has none of it, so seeds experience your mail the way a brand-new, perfectly neutral subscriber would.

Personalization. Providers filter per mailbox, not per message. The same campaign can land in Primary for an engaged reader, Promotions for a cold one, and spam for someone who complained once. A seed list returns one placement per seed, which samples this distribution rather than resolving it.

Panel data, the main alternative methodology, flips the trade: it observes real users' mailboxes (via consented inbox access) and thus captures engagement effects, at the cost of coverage you do not control and audiences skewed toward panel joiners. Neither method is the truth; they are different instruments with different error bars.

The right reading of seed data follows: it isolates the sender-and-content component of placement (authentication, sender reputation, content filtering) under controlled conditions, which is exactly what a comparative test between services wants.

How seed testing works

The mechanics are simple to describe and demanding to run well.

Provisioning. The tester creates mailboxes across the target providers, enough per provider for a placement rate to be meaningful rather than anecdotal, and keeps them in a defined state — the aim is neutrality, so seeds are not trained toward or against any sender.

Sending. The seed addresses are included as ordinary recipients of the mailing under test. For comparative testing (our case), the same campaign archetype goes to the same seed set through each service being compared, holding content and timing constant so the service is the only variable. This is an inbox placement test in the strict sense.

Reading. After a settling window, each mailbox is checked and every message is recorded in one of a small set of states: inbox, categorized tab, spam, missing, or bounced at send time. Missing matters — silent loss after acceptance is a real filtering outcome that no other state captures.

Aggregating. Counts become rates per provider and overall. In our protocol those rates feed a published formula, described on our methodology page, in which tab placement is scored as its own state rather than collapsed into inbox or spam.

One rule governs everything around those steps: the protocol is public, the infrastructure is confidential. Publishing seed addresses, sending domains or timing patterns would let a tested service recognize test traffic and treat it specially, which would quietly convert the instrument into a press release. This is why our methodology page describes the design but never the deployment details.

Seed lists and your deliverability

For a sender running their own program, seed thinking earns its place in three ways.

It separates acceptance from placement. The most common deliverability self-deception is reading a platform's "delivered" number as inbox reach. A handful of your own cross-provider mailboxes on every major campaign gives you the placement dimension your dashboard lacks, and catches a spam-folder slide weeks before open-rate trends make it undeniable.

It isolates variables. Placement changed — was it the new template, the new dedicated IP, the list source? Seeds receive your mail without engagement history, so a shift in seed placement points at sender-level causes: authentication breakage (verify with the email authentication checker), domain reputation movement, or content. Provider dashboards like Postmaster Tools then corroborate from the receiver's side.

It grounds vendor claims. Any service's deliverability marketing can be checked against controlled seed data — which is the entire premise of this site's testing program. Comparative claims without a controlled protocol behind them are adjectives.

Limitations and failure modes

Reading seeds as audience truth. The permanent caveat: seed placement is not your subscribers' placement. Engaged audiences will outperform your seed numbers; burned lists will underperform them. Seeds measure the mail; your audience measures the relationship.

Trained seeds. Mailboxes that accumulate history stop being neutral instruments — a seed that keeps opening your mail drifts toward inbox, one buried in unread mail drifts away. Disciplined protocols manage mailbox state deliberately; casual ones decay without noticing.

Recognizable test traffic. Seed sets whose addresses leak, or that are shared across the industry, become known to platforms and filters. Mail to known seeds can be handled unrepresentatively — in either direction. This is the confidentiality rationale in one sentence.

Sample-size theater. Three Gmail mailboxes cannot distinguish 70% placement from 100%. Rates need denominators; a protocol that publishes percentages without disclosing its design deserves suspicion. Ours gates publication on minimum seed counts, per the methodology page.

Snapshot bias. One send on one day is one draw from a noisy process. Filters update, reputations move, greylisting and deferrals delay individual deliveries past a hasty reading window. Trend lines and repeated cycles beat single snapshots.

Forgetting the send side. Seeds measure nothing if the mail never leaves cleanly: a hard bounce at the gateway or a quota block is a send failure, not a placement result. Record bounced-at-send as its own state, as our protocol does.

Related terms

Inbox placement test, inbox placement, Promotions tab, sender reputation, domain reputation, Postmaster Tools, deferral, greylisting.

Frequently asked questions

What is a seed list in email testing? A set of tester-controlled mailboxes across major providers, added to a send so the tester can check each one and record actual folder placement — the data that sending-side delivery metrics cannot show.

How is a seed list different from a test list? A test list is usually colleagues checking rendering and links. A seed list is a measurement instrument: provider coverage chosen for representativeness, mailbox state managed for neutrality, results recorded as placement rates.

How many seed addresses do I need? Enough per provider that a rate means something — single digits per provider is anecdote, not measurement. Our own protocol sets a minimum seed count as a publication gate; the principle applies at any scale.

Are seed test results the same as my real inbox rate? No. Seeds lack the engagement history that personalizes filtering for your actual audience, so they isolate the sender-and-content component. Treat seed rates as a controlled baseline, not a forecast.

Does InboxRatio use seed lists? Yes — our Deliverability Scores come from controlled seed-list cycles, protocol summarized on our methodology page. Cycles are announced as they complete; a service we have not yet tested shows "Not yet tested," never an estimate.

Why don't you publish your seed addresses or domains? Because a service that can recognize test traffic can game it. The protocol is public and criticizable; the deployment details stay confidential to keep the measurement honest.

If you run any recurring volume, stand up a modest cross-provider seed row of your own and copy it on every major send. The day your Gmail seeds move from inbox to spam in unison, you will know hours after the send — and the deliverability troubleshooter plus your authentication records are the first places to look.

Sources