Email Marketing Deliverability
Deliverability Testing: Tools and Methods That Work
Seed lists, inbox placement tests, what each method cannot tell you, and how to test without polluting the result.
On this page
The short version
- Seed testing shows placement for a panel of accounts you control, not for your list. It is a trend indicator, not a measurement.
- The strongest available evidence is indirect: open rate broken down by receiving domain, compared against itself over time.
- Test before you need to. A baseline taken while everything is healthy is what makes a later reading interpretable.
Nobody can see where your mail lands for your actual subscribers. Mailbox providers do not report it, and no tool has access to a stranger's inbox. Every deliverability testing method is therefore a proxy, and knowing which proxy you are holding is most of the skill.
That is not a counsel of despair. The proxies are useful, they agree with each other when something is wrong, and combining two of them gives a reliable enough picture to act on.
The methods and what each cannot tell you
| Method | Shows | Cannot show |
|---|---|---|
| Seed list testing | Placement across accounts you control | Placement for real subscribers |
| Postmaster tools | The provider's own view of your domain | Anything about other providers |
| Open rate by receiving domain | Relative performance across providers | Absolute placement, and it is contaminated |
| DMARC aggregate reports | Who sends as you and whether they pass | Where the mail landed |
| Spam-score checkers | Mechanical faults in the message | Reputation, which dominates |
| Subscriber reports | Actual placement, for one person | Anything generalisable |
Why seed lists drift from reality
Seed accounts do not behave like people. They never open, never reply, never move a message out of spam, and have no history with you — and personal history is one of the largest inputs into placement.
The consequence is systematic rather than random: seed results tend to be pessimistic for senders with an engaged list and optimistic for senders whose list is disengaged, because the seed panel has no engagement either way.
That does not make them useless. It makes them a comparison against themselves. A panel that showed inbox placement last month and spam this month has detected a change, which is the question worth asking.
The reading that costs nothing
Break open rate down by receiving domain and watch each one against its own history. This is available in every platform and it is the most sensitive early indicator most senders have.
The absolute numbers are contaminated by privacy protection and by scanner opens, and that does not matter here, because the contamination is roughly constant per provider. What you are looking for is one provider's line diverging from the others — that pattern means something changed at that provider specifically, which is exactly what a placement problem looks like.
A drop across every provider at once is more likely to be a content, timing or audience issue than a deliverability one.
A testing routine that is worth the time
Take a baseline while things are healthy
Seed test one campaign, record open rate by receiving domain, note your postmaster reputation. Without this, a later reading has nothing to be compared against.
Check the cheap indicator every send
Open rate by domain takes a minute and catches divergence early.
Seed test on change
A new template, a new sending domain, a new provider, a significant volume increase. Not every campaign — the panel does not change between sends, so testing an unchanged setup repeatedly tells you nothing new.
Review postmaster data monthly
Spam rate and reputation at the major providers, looked at deliberately rather than when something is already wrong.
Investigate on divergence, not on a single bad number
One low open rate is noise. One provider's open rate falling while the others hold is a signal, and that is the point at which the seed test earns its cost.
Testing check
- A baseline exists from a period when everything was healthy
- Seed addresses are not mixed into the main list
- Open rate by receiving domain is reviewed every send
- Postmaster tools are connected and reviewed monthly
- Seed tests run on change, not on a fixed schedule
- Results are compared against your own history, not published benchmarks
Frequently asked questions
How many seed addresses do I need?
Enough to cover the providers your list actually uses, weighted the same way. A panel of a hundred accounts at providers your subscribers do not use is a precise measurement of nothing relevant.
Are paid deliverability platforms worth it?
They save time by consolidating seed testing, blocklist monitoring and reputation data in one place. The underlying data is largely available free; you are buying the aggregation and the alerting, which is a real product for a programme large enough to need watching daily.
Can I test before I have a list?
You can verify authentication, run a spam-score check for mechanical faults and seed test the template. What you cannot test is reputation, because you do not have one yet — which is what warmup is for.