In this Article
Email deliverability testing promises to tell you whether a campaign will land in the inbox before you send it. What the tools measure is real, but it is narrower than the promise, and mistaking the sample for the population leads teams to optimise the wrong thing.
This guide explains how placement tests work, the five factors that actually decide the inbox, why seed results and real results diverge, and the order in which fixes pay off.
Key Facts
- A seed list is a sample, not the truth. Seed accounts have no engagement history, and engagement is the strongest placement factor at the major providers.
- Authentication is table stakes. Without SPF, DKIM and a DMARC policy, nothing else in a testing tool’s report matters.
- Placement is per-recipient, not per-message. The same message reaches one person’s inbox and another’s spam folder on the same day at the same provider.
- Your own postmaster data beats any test. The major providers publish free reputation dashboards for domains that send to them.
- Proxies do not affect deliverability. Reputation lives on the sending IP and domain, and a proxy changes neither.
How does a deliverability test work?
You send the campaign to a set of seed mailboxes the vendor controls across the major providers, then the vendor checks where each copy landed and reports inbox, spam or missing.
That answers one useful question: whether the message itself, with its authentication, links and content, is being rejected outright or filtered on its own merits. It cannot answer the more important question, which is how your actual recipients’ mailboxes will treat it, because those mailboxes have years of history with your sending domain and the seed accounts have none.
Treat a placement test as a smoke test for the message and the setup, and treat your own delivery telemetry as the measurement of the programme.
What actually decides inbox placement?
Five factors, roughly in order of weight. We call it the 5-factor placement model.
| Factor | What providers look at | How fast it moves |
|---|---|---|
| 1. Recipient engagement | Opens, replies, moves out of spam, deletions without reading | Weeks; the dominant factor and the slowest to repair |
| 2. Complaint rate | Spam button presses per thousand delivered | Immediately punishing above a very low threshold |
| 3. Authentication | SPF, DKIM alignment, DMARC policy | One-time configuration; binary |
| 4. Sending pattern | Volume consistency, ramp rate, bounce handling | Days; sudden changes read as compromise |
| 5. Content and links | Link reputation, attachment types, image-to-text balance | Per message, and the factor tools over-weight |
Most testing tools report heavily on factor five because it is the one visible in a single message. It is also the least important of the five for an established sender, which is how teams end up rewriting subject lines when the real problem is a list they should have stopped mailing.
Why do seed results and real results diverge?
Three reasons worth knowing before you act on a report.
No engagement history. Seed accounts never open, click or reply, so they look like the most disengaged segment of your list. At providers where engagement dominates, that biases results pessimistically.
Per-recipient filtering. The major providers decide per mailbox using that user’s own behaviour. A single inbox-or-spam verdict for a provider is an average over a handful of accounts, and the variance is large.
Seed detection. Widely used seed addresses can be recognised, and filtering applied to them is not necessarily what a normal mailbox gets.
None of this makes testing useless. It makes it a diagnostic for setup rather than a forecast of results, and reading it that way prevents the classic mistake of chasing a seed score while the list quietly degrades.
What should you fix, and in what order?
| Symptom | Fix this when | Do not bother when |
|---|---|---|
| Messages rejected or bulk-foldered everywhere | Authentication: SPF, DKIM alignment, DMARC | Already configured and passing; look further down |
| High bounce rate | List hygiene and immediate suppression of hard bounces | Bounces are already low single digits |
| Placement falling at one provider only | That provider’s postmaster dashboard | You have not yet registered the domain there |
| Gradual decline across providers | Engagement: stop mailing inactive segments | You are unwilling to shrink the list, in which case nothing else will work |
| New domain, sudden volume | A slower warm-up | Volume has been stable for months |
The uncomfortable fix is usually the fourth row. Removing people who have not engaged in six months reduces reach on paper and improves it in practice, because placement for everyone else rises.
Where do proxies fit, and where do they not?
They do not fit the sending side at all, and it is worth saying plainly because the two topics get marketed together.
Deliverability is a reputation system attached to your sending IP and your sending domain. A proxy changes the source address of an HTTP request; it does not touch either, cannot improve placement, and routing mail through rotating addresses would look like exactly the pattern spam filters are built to catch.
Where collection infrastructure does appear in this workflow is earlier and separate: gathering public business information, checking that a company page still lists a contact, or verifying how a landing page renders in different markets. That is ordinary web collection, and DataImpulse residential covers it at $1 per GB across 195 countries.
The rules that govern the rest of it are consent rules rather than technical ones: the GDPR, ePrivacy and CAN-SPAM decide who you may contact, and no amount of deliverability engineering substitutes for that. General information, not legal advice. Related: bulk email verification.
Frequently Asked Questions
What is email deliverability testing?
Sending a campaign to a set of seed mailboxes a vendor controls across major providers, then reporting where each copy landed. It diagnoses the message and the sending setup well, and predicts real-recipient placement poorly, because seed accounts have no engagement history.
Why does my seed test look worse than my real results?
Seed accounts never open, click or reply, so they resemble your most disengaged segment. At providers where engagement is the dominant factor, that biases the test pessimistically. The reverse can also happen when seeds are recognised and treated differently.
What matters most for inbox placement?
Recipient engagement, then complaint rate, then authentication, then sending pattern, then content. Most testing tools report mainly on content because it is what a single message reveals, which is also why teams rewrite subject lines when the real problem is the list.
Do proxies improve email deliverability?
No. Reputation is attached to the sending IP and the sending domain, and a proxy changes neither. Routing mail through rotating addresses would resemble the pattern spam filters exist to detect. Proxies belong to public web data collection, not to sending.
What should I fix first?
Authentication, because it is binary and cheap: SPF, DKIM alignment and a DMARC policy. Then hard-bounce suppression, then the providers’ own postmaster dashboards, and only then engagement pruning, which is the most effective and the least popular fix.
The collection side, done cleanly
Checking how a landing page renders per market, or gathering public business data, is ordinary web collection. DataImpulse residential proxies cover 195 countries at $1 per GB with country and city targeting. Create an account and test one market.
Related: bulk email verification · data enrichment · proxies for web scraping.
Last updated: September 17, 2026.
State/City/Zip/ASN Targeting 
