KoltrixInbox placement testing: seed lists, panels and limits
Seed tests show where a message lands for test accounts, not for your users. What seed lists and panel data can tell you and how to use them honestly.
On this page(7 sections)
Delivery logs tell you a receiving server accepted your message. They do not tell you whether it landed in the inbox, a tab, or the spam folder.
Inbox placement testing tries to fill that gap, and it is useful as long as you understand exactly what it measures.
Accepted is not the same as inboxed
When a mailbox provider responds 250 OK, your job as a sender is done as far as SMTP is concerned. What happens next, inbox, spam folder, a category tab or silent filtering, is decided inside the provider and is not reported back to you. Most "deliverability" numbers in sending dashboards are really acceptance numbers.
Placement testing estimates the missing step. There are two main approaches: seed lists and panel data.
Seed list testing
A seed list is a set of test mailboxes at many providers, controlled by you or a testing service. You add the seeds to a campaign or send a test message to them, and software logs into each mailbox to check where the message landed.
What seeds are good at
- Catching hard failures. If a message lands in spam at every seed of one provider, something is badly wrong, often authentication, a blocklisted link or a content problem.
- Comparing variants. Send two versions of a message, or the same message from two configurations, and compare placement under identical conditions.
- Checking before a big send. A pre-flight seed test before a large campaign can catch a broken DKIM signature or a bad link domain.
- Covering providers you rarely see. Smaller regional providers that you cannot easily monitor otherwise.
Why seeds can mislead
Seed mailboxes are not real users. They do not open, click, reply, move messages or mark them as spam. Modern filtering at large providers relies heavily on engagement and per-recipient signals: how this specific person has interacted with your mail before. A seed account has no such history, so it sees a version of filtering that ignores the single most important input.
Other limitations:
- Small samples. A few seed accounts per provider is not a statistically meaningful view of millions of recipients.
- Known seeds. Some providers may recognize accounts that behave like test mailboxes, and testing services' seed lists are not secret.
- Per-recipient variation. The same message can go to the inbox for an engaged subscriber and to spam for an inactive one. A seed is neither.
- Tabs. Whether a message lands in a category tab is placement too, and seeds report it inconsistently.
A seed result of "80% inbox" does not mean 80% of your real recipients saw the message in their inbox. It means 80% of a handful of test accounts did.
Panel data
Some services estimate placement from panels of real users who have consented to share data about their mailboxes, through apps or browser extensions. Because these are real people with real engagement histories, panel data reflects actual filtering more closely than seeds.
Limitations:
- Coverage. Panels are biased toward the kinds of people who install such tools, and coverage of your specific recipients may be thin.
- Scale. Smaller senders may see few or no panel observations.
- Privacy and consent. Make sure any service you use is clear about how panel data is collected.
Better signals you already have
Placement tests are a supplement. Your own data is often more reliable:
Engagement by receiving provider
Break open and click rates down by provider group. If clicks from one large provider drop sharply while others hold steady, placement at that provider has probably worsened. Clicks are more reliable than opens, because some mail clients load images automatically and inflate opens.
Postmaster dashboards
Google Postmaster Tools gives you domain-level spam rates and authentication data for Gmail. Microsoft SNDS gives IP-level data for Outlook.com if you use dedicated IPs. These are first-party signals from the providers themselves.
Complaints and replies
A rising complaint rate is a placement problem in progress. A stream that normally gets replies and suddenly does not may be landing in spam.
Real test mailboxes
Keep a few personal mailboxes at major providers that subscribe to your mail and engage with it like a normal user. They are not statistically meaningful either, but they show what an engaged recipient experiences.
How to use placement tests well
| Use case | Good fit? | Why |
|---|---|---|
| Pre-flight check for broken authentication or links | Yes | Hard failures show up on any account |
| A/B comparison of two configurations | Yes, with care | Same conditions for both variants |
| Estimating campaign-wide inbox rate | Weak | Seeds lack engagement history |
| Monitoring trends over months | Moderate | Consistent method helps, absolute numbers do not |
| Proving compliance with bulk sender rules | No | Placement is not compliance |
When you run seed tests:
- Test the real message. Send it from production infrastructure with production authentication, not from a test environment.
- Repeat tests. Single results are noisy; look for consistent patterns.
- Look at headers on spam placements. The Authentication-Results header in a spam-foldered seed often reveals the cause.
- Do not chase single-seed results. One seed in spam at one provider is not a trend.
Checklist
- Treat delivery logs as acceptance data, not placement data.
- Use seed tests for pre-flight checks and controlled comparisons.
- Read the headers of any seed message that landed in spam.
- Track clicks and complaints by receiving provider.
- Use Postmaster Tools and, for dedicated IPs, SNDS.
- Keep a few engaged personal test mailboxes.
- Never present seed inbox rates as real-audience inbox rates.
Bottom line
Seed lists and panels estimate where mail lands, but neither sees your recipients' engagement history, which modern filters weigh heavily. Use placement tests to catch hard failures and compare variants, and rely on engagement by provider, complaint rates and postmaster dashboards for the real picture of how your audience receives your mail.
Start with Koltrix
Your domain, one inbox, and an API that sends.
A team inbox where AI sorts and drafts (nothing is sent without your click), plus the transactional API and SMTP relay your product sends with. 7 days free, no card.
