EVALUATE
How to read an inbox placement test before buying a warmup tool
A placement test tells you where seed copies of one message landed, for one send, at one moment. Read the sample size and the provider split first, the authentication lines second, the folder split third. It never proves that a tool caused a change, and it never guarantees the next send. Treat a weak result as a diagnosis, not as a purchase order.
What a placement test actually measures
A placement test sends one message to a list of addresses owned by the testing provider, the seed list. Each seed reports which folder the copy reached: inbox, spam, or nothing at all. The tool then groups the seeds by provider and shows the split. That is the whole measurement. GlockApps built its Inbox Insight suite around exactly that reading, and MailReach bundles a lighter version of it next to its warmup service.
Three things follow from that. The unit is seed addresses, never your recipients. The scope is one message at one moment, not your sending in general. And the precision is limited by how many seeds each provider group holds, which is usually a few dozen at most.
Warmup, placement testing, authentication and list verification stay four separate jobs on this site. A placement test measures. It does not repair anything, and a warmup service does not measure anything.
A sample report, line by line
The report below is fictional and written for this guide. The figures are invented to show the shape of a real report and how to read it. They describe no product, no customer and no test we ran.
Inbox placement report, single run
Sample report · fictional data| Provider group | Seeds | Inbox | Spam | Missing | Split |
|---|---|---|---|---|---|
| Gmail | 42 | 29 (69.0%) | 11 (26.2%) | 2 (4.8%) | |
| Microsoft (Outlook, Office) | 31 | 9 (29.0%) | 14 (45.2%) | 8 (25.8%) | |
| Yahoo, AOL | 22 | 19 (86.4%) | 3 (13.6%) | 0 (0.0%) | |
| Other providers, self-hosted | 25 | 21 (84.0%) | 3 (12.0%) | 1 (4.0%) | |
| All seeds | 120 | 78 (65.0%) | 31 (25.8%) | 11 (9.2%) |
Fictional figures, written for this guide. No provider, tool or customer account produced them. "Missing" means the seed never reported a copy in any folder, which is not the same as a bounce in your sending platform.
Inbox Spam folder Missing
Start at the top, not at the big number. The header says the list holds 120 addresses and that 109 reported a result. It says the message was a single subject line and body, sent once. It says SPF, DKIM and DMARC passed on every seed that reported. Those lines decide how much the table below is worth.
Now the table. The overall inbox share is 65%, and it hides the only finding that matters here: the Microsoft group sits at 29% inbox with 45% in spam, while Yahoo and the smaller providers are above 84%. A single average across providers would have buried that.
Read the sample before you react. The Microsoft group holds 31 seeds. One seed moving changes that share by about 3 points, and the 95% interval around 29% runs from roughly 14% to 44%. The gap with Yahoo is wide enough to look at; the exact number is not precise enough to report to a client as a percentage.
What the report shows, and what it cannot prove
The report supports
- Where the seed copies of this message landed, provider group by provider group
- Whether SPF, DKIM and DMARC passed for this send, on the seeds that reported
- Which seeds returned nothing, as a separate count from spam
- A difference between providers that is large enough to survive the sample size
The report does not support
- An inbox rate for your real recipients, who behave differently from seeds
- Any claim that a tool, a warmup service or a setting caused a change
- A prediction about your next send, tomorrow, or a different message
- A verdict on list quality, engagement or complaints, which it never sees
- A guarantee of deliverability, which no tool and no test can give
The causal point deserves its own sentence. If you switch a warmup service on, wait three weeks and run a second test that comes out higher, the report shows two measurements. Your sending volume, your content, your list and the providers' own filters all moved in those three weeks. The report cannot separate them, so a before and after pair is an observation, not a proof, and it is not a reason to renew a subscription on its own.
Read the numbers in this order
- Sample and provider split. How many seeds in total, and how many in each provider group? Anything under about 20 seeds in a group gives you a direction at best.
- Authentication. SPF, DKIM and DMARC, on the seeds that reported. A failure here explains most of the rest, and it costs nothing to fix.
- Missing versus spam. Missing seeds are not spam placement. Cross-check them against the bounce log in your sending platform before calling them a reputation problem.
- Folder split per provider. Compare groups against each other, not against a target number. One provider far below the others is the finding.
- Repeat. Run the same seed list with the same message again before you act. Two runs that disagree mean your sample is too small to buy anything on.
When two reports are not comparable
Most placement figures that circulate compare runs that cannot be compared. A report is comparable with another only when the seed list, the provider mix, the message, the sending domain and the send window all hold still, and when both samples are of similar size.
| What changed | Run A | Run B |
|---|---|---|
| Seed addresses | 120 | 60 |
| Gmail seeds in the list | 42 | 12 |
| Message | Subject A, plain text, 2 links | Subject B, HTML, 7 links |
| Send window | Tuesday, 09:00 UTC | Saturday, 22:00 UTC |
| Sending domain | Main domain | New sending subdomain |
| Reported inbox share | 65% | 81% |
Not comparable. Four inputs moved at once and the second sample is half the size, so the 16 point gap measures the setup, not the sender.
Two more traps. Providers do not share one folder taxonomy: a copy in the Gmail Promotions tab is inbox for some tools and a separate category for others, so check how your tool counts it before you compare two vendors' numbers. And a percentage without its denominator is unusable, which is why the table above carries the seed count in its own column.
Turn the reading into a decision
Each reading points to one action, and two of the five point to no purchase at all.
| What the report shows | What it points to | Spend? |
|---|---|---|
| SPF, DKIM or DMARC failing on part of the seeds | A DNS and signing problem. Fix the records at your registrar and in your sending platform, then run the same test again. | No purchase needed |
| High missing share, and hard bounces in your sending platform log | A list problem, not a reputation problem. Verify the list before you send again. ZeroBounce sells verification credits for that job. | List verification, if the bounce log confirms it |
| Spam placement concentrated on one provider group, authentication passing, steady sending history | A content and reputation question at that provider. Repeat the test with the same seed list and a changed message before concluding anything. GlockApps is built for that repetition. | Placement testing, not warmup |
| A domain or mailbox with almost no sending history | The report has nothing to measure yet. A warmup service such as MailReach or Warmup Inbox builds an interaction history over weeks; the report cannot show that it will change your placement. | Warmup, with a stated expectation and a repeat test |
| Inbox share even across provider groups, authentication passing | Nothing in the report justifies a purchase. Keep a scheduled control and spend the budget on the list and the offer. | No purchase needed |
The sample report above lands on the third row. Authentication passes, the domain has been sending for over a year, and one provider group is far below the others. Nothing there is a warmup problem. The next step is a repeat test with the same seed list and a reworked message, not a subscription.
The control to run before you spend anything
This one is observable and free of charge in most plans. Run the same seed list twice, with the same message, 48 hours apart, and change nothing in between.
- If the two runs land within a few points of each other, your sample is stable enough to act on the gap you saw.
- If they disagree widely, the difference you were about to pay for is noise. Enlarge the seed list or stop reading the number that closely.
- In parallel, open the bounce log of your sending platform and your DMARC aggregate reports. Both are free, both cover your real recipients, and the placement test covers neither.
GlockApps publishes a free tier with two spam tests, which is enough for this repeat control on a single message before any paid plan enters the conversation. Whatever you conclude, cost it across every mailbox first: a per mailbox price looks small until it is multiplied by a whole team or a client roster.
Questions people also ask
My inbox rate rose after I switched on a warmup tool. Does that prove it worked?
No. It shows two measurements taken weeks apart, with your volume, your content, your list and the providers' filters all moving in between. A placement test has no way to isolate one cause. Keep the observation, drop the causal claim, and judge the tool on what it does that you can check instead.
How many seed addresses do I need for a usable result?
What matters is the count inside each provider group, not the total. Below roughly 20 seeds in a group, treat the percentage as a direction rather than a figure, and never quote it as a rate. The report should always show the denominator next to the share.
Do seed results predict what my real recipients see?
No. Seeds are addresses that never open, never reply and never buy, so the engagement signals that carry real placement are absent by construction. A placement test is a diagnostic on one message, not a forecast of your campaign.
Is a placement test the same thing as a warmup tool?
No, and mixing them wastes money. A placement test measures where a message lands. A warmup service sends automated interactions to build a sending history. Some products do both, MailReach among them, but the two jobs are still judged separately. The diagnostic tree sorts which one your situation calls for.
Do I need a paid plan just to read one report?
Not to start. GlockApps advertises a free tier with two spam tests, which covers a first reading and one repeat run. Move to a paid plan when you need tests on a schedule or several sending domains, and price it per test rather than per month.