Inbox Placement Tests: What They Tell You and What They Miss

A 95 percent inbox placement score is the most dangerous number in cold email.

We run outbound for 50+ B2B companies, and a clean placement test has never once predicted whether a campaign would book a meeting.

Below is what the test actually measures, the 4 ways a seed list lies to you, how to read a placement report line by line, and the exact diagnosis path from a failed test to a fix.

What Is an Inbox Placement Test?

The confusion lives in the name. People hear inbox placement test and assume it predicts where their real mail will land in their real prospects' inboxes. It does not. It tells you where one message landed in a controlled set of test inboxes that the tool owns. That is a meaningful signal, because if your message cannot reach a neutral seed inbox, it certainly cannot reach a skeptical buyer. A pass is a floor, not a ceiling.

The right mental model is a pre-flight check. Before a plane leaves the gate, the crew confirms the systems work. That check does not promise a smooth flight or that the passengers will enjoy it, it only confirms the aircraft is safe to fly. An inbox placement test is the same. It confirms your sending path is airworthy, which is necessary and nowhere near sufficient. To read it properly you need the layer underneath it, which we lay out in what is email deliverability.

Inbox Placement Test

A diagnostic that sends a test message to seed inboxes across multiple mailbox providers and reports the folder each copy reached, producing a placement rate by provider. It isolates whether your domain reputation, authentication, and content are healthy enough to reach the inbox, separate from whether real recipients will engage. Because it runs on dedicated test addresses rather than real prospects, it reflects sending health but not the personalized filtering a live recipient triggers.

Seed List

The set of test mailboxes a placement tool owns across Gmail, Outlook, Yahoo, Apple, and sometimes corporate or regional providers. Seed addresses are not subscribers. They never open, click, or reply, they receive a small test send rather than your real campaign, and they build no engagement history, which is why a seed result is directional evidence about your sending path rather than a prediction about any individual buyer.

That second definition is the whole article in miniature. Every strength and every blind spot of a placement test traces back to the fact that the recipients are props, not people. Keep that in front of you and the tool becomes genuinely useful. Forget it and you end up rebuilding infrastructure that was never broken.

How Does an Inbox Placement Test Work?

The mechanics are simple, and the simplicity is the point. Here is the sequence every major tool follows, whether you run it through Instantly, GlockApps, Mailreach, or a standalone checker:

1. Generate a seed list and a tracking code. The tool hands you a batch of test inboxes spread across Gmail, Outlook, Yahoo, and sometimes corporate or regional providers, along with a unique string so it can identify your message later.
2. Send the real message. You send the exact email you plan to use, from the exact domain and mailboxes you plan to scale, with the code included. Sending a stripped down test version is the most common way operators invalidate their own test, because content is one of the three things the test is grading.
3. Let the tool scan the folders. It checks every seed inbox and logs where your message landed: primary, promotions or another tab, spam, or missing entirely. Missing is its own category and it is worse than spam, because the message was rejected or silently dropped rather than filed.
4. Read the per-provider breakdown. You get placement by provider plus an overall score, so a Gmail-strong, Outlook-weak result shows up as two different numbers instead of one misleading average.
5. Read the diagnostic panels. Most tools also report authentication results, blocklist status, and a spam filter score alongside placement. Those panels are where the actual fix lives, and most operators never scroll that far.

The per-provider split is the most useful part of the raw output. A blanket 90 percent placement rate hides the story, because Gmail and Microsoft run completely different filters and one can inbox you while the other quietly buries you. According to Instantly's inbox placement testing guide, the goal is never to predict every recipient's inbox, it is to confirm the sending path is healthy enough to scale. Reading the providers separately is how you turn a vague score into an actionable one.

One more mechanical point that trips people up. The test grades the sending identity you used, not your account in general. If you run 30 mailboxes across 8 domains, a test from 1 mailbox tells you about that mailbox and its domain. Pools drift apart, which is the reason multi-domain sending strategy and mailbox rotation both change what a single test is worth. Sample across the pool or accept that you only tested a slice of it.

What Is a Good Inbox Placement Rate?

Benchmarks are where the conversation gets honest, because most operators have never seen a number to compare theirs against. The widely cited figure comes from EmailToolTester's running deliverability test, which sends through 15 email services on a repeating schedule and publishes the folder breakdown. The headline is that roughly 1 in 6 legitimate emails never reaches the inbox.

Here is the provider breakdown from that same benchmark, which is the part worth bookmarking, because it tells you which provider to blame when your blended score drops:

Source: EmailToolTester deliverability test, last updated March 2026.

Two things fall out of that table. First, Microsoft is the hard one, and it is not close. A sender sitting at 89 percent on Google and 60 percent on Outlook has a Microsoft problem, not a general deliverability problem, and the fix path runs through Microsoft's Smart Network Data Services rather than through rewriting subject lines. Second, missing is a real and separate bucket. Between 3.8 and 7.5 percent of mail does not get filed anywhere, which is a rejection or a silent drop, and that number belongs in the same conversation as your bounce rate.

Hold cold outbound to a higher bar than those averages. Those figures come from email services sending to opt-in lists with engagement history behind them. Cold email has none of that, so the same placement rate on a cold domain represents a thinner margin of safety. We treat anything under 90 percent on Google as a signal to stop scaling and look, and anything under 80 percent on Microsoft as a reason to check the domain before adding volume. Both thresholds are operating policy, not an industry standard, and the point is to have one rather than to copy ours.

How Do Seed Lists Lie to You?

This is the section most placement-test content skips, and it is the one that saves you the most time. A seed test is not wrong so much as structurally narrow, and the narrowness runs in 4 directions:

• Zero engagement history. Mailbox providers weigh how a specific recipient has treated your past mail. Seed addresses never open, never reply, and never move you out of spam, so they cannot reproduce the personalized filtering that decides a real buyer's folder. As Suped's breakdown of seedlist accuracy puts it, seed data is directional rather than exact, reliable for catching major failures and unreliable as a live truth source.
• The test send is not your campaign. You send a handful of messages to seeds. Your campaign sends thousands over days from a rotating pool at a ramping volume. Volume, pacing, and spread are themselves filtering signals, so a clean 25-address test says little about what happens at 3,000 sends a day. Twilio's guidance on seed list testing is explicit that seeds supplement live data rather than replace it.
• Seed addresses age and burn. The same seed inboxes get used by thousands of senders, including bad ones, and they accumulate a filtering history that has nothing to do with you. That is how a perfectly healthy domain draws a spam result on one provider in one tool and inboxes everywhere else. Trend over several tests, never panic on one.
• Filtering is personalized, so there is no single answer to grade. Two people at the same provider can receive the identical message into different folders based on their own history with you. A seed inbox is 1 draw from a distribution, and a placement score reports it as if it were the mean.

Put together, the practical rule is that a failed test is more trustworthy than a passed one. A red result usually means something real and binary broke, and those causes are in SPF, DKIM, and DMARC, in a blocklist listing, or in domain reputation. A green result only tells you nothing is obviously broken, which is a much weaker claim than operators hear it as.

How Do You Read an Inbox Placement Report?

Every serious placement tool reports more than a percentage, and the extra panels are the only part that points at a repair. Here is what to read, in the order that narrows the problem fastest:

That last row is underrated. Providers often slow you down before they filter you, so rising delivery latency on one provider is a leading indicator, and it is the kind of thing you only notice when you log tests over time instead of reading one in isolation.

Pair the report with the provider's own telemetry, because the provider knows things no third party tool can see. Google Postmaster Tools shows your actual user-reported spam rate and domain reputation. Microsoft's Smart Network Data Services does the equivalent for Outlook. Yahoo's sender best practices documents what it expects. A placement test plus provider telemetry covers both sides of the question, and neither one alone does. The full stack is in cold email deliverability monitoring.

How Often Should You Run an Inbox Placement Test?

Cadence is what turns the tool from a vanity check into an actual safeguard. A one-time test at launch tells you the campaign started healthy and nothing about whether it stayed that way. The point is to catch a reputation dip in the window where it is a 1 day fix, before it hardens into a burn that costs weeks and a domain. Here is the rhythm we hold across the book:

That last row is the single highest leverage use of the tool. A reply rate drop has 2 possible causes and they lead in opposite directions, so a 10 minute test that rules one of them out is worth more than a day of guessing. The benchmarks that tell you whether your reply rate actually dropped are in cold email reply rate benchmarks.

Keep the tests small. Seed addresses never engage, so testing constantly adds a block of permanently unengaged recipients to your sending history. A few dozen addresses on a weekly schedule is a rounding error. Daily tests across a large seed list on a young domain are not.

What Do You Do When an Inbox Placement Test Fails?

A failed test is good news, because it means something real and findable broke. Here is the order we work it, and the order matters because each step is cheaper than the one after it:

1. Confirm authentication passes and aligns. SPF, DKIM, and DMARC are pass or fail, and a failure explains a total spam result on its own. Check the records and the alignment between the From domain and the signing domain against the SPF, DKIM, and DMARC reference, and the syntax and lookup limits against the DNS records guide. DMARC.org documents how the alignment check is supposed to resolve.
2. Check the blocklists. A listing overrides everything downstream, so look before you touch anything else. Spamhaus and MXToolbox cover the lists that matter for B2B sending, and the standing process is in cold email blacklist monitoring.
3. Read the provider's own numbers. Google asks senders to keep the user-reported spam rate in Postmaster Tools below 0.10 percent and never to reach 0.30 percent, and its sender guidelines state that senders above 0.30 percent are ineligible for mitigation until they hold under it for 7 consecutive days. If your spam rate is the problem, no test result and no copy edit matters until it comes down.
4. Isolate content and borrowed reputation. Read the spam filter rules the report names. Then check the things that carry someone else's reputation into your message: a shared tracking domain, a link shortener, a redirect chain, an attachment. Swap 1 element at a time and retest.
5. Look at volume and pacing. A domain that passed last week and fails this week after a volume increase did not develop a content problem. Pull volume back, hold it flat, and retest. The sizing math is in cold email infrastructure setup and how to set up email domains for outbound.
6. Look at the list. Bounces and complaints are reputation inputs, so a dirty list damages placement with a perfect technical setup. Verification standards are in why email verification matters, and the bounce taxonomy is in bounce rate causes and fixes.
7. Decide between repair and retirement. Some domains come back and some do not. The recovery path, and the point where rotating onto a fresh domain beats nursing a burned one, is in how to recover a burned domain. Keeping a secondary domain warm is what makes that decision cheap instead of a 3 week outage.

The reason the order is fixed is that steps 1 and 2 take about 10 minutes and resolve a large share of failures outright. Operators who start at step 4, rewriting copy, routinely spend a week on a problem that was a missing DKIM record. A broader version of this tree, written from the symptom side, is in how to keep cold email out of the spam folder.

What Does an Inbox Placement Test Never Tell You?

This is where most operators get fooled, so it is worth being blunt. The test passes, they assume the campaign is healthy, it books nothing, they retest, see green again, and conclude deliverability is a mystery. It is not a mystery. They are reading a tool that was never measuring the thing that broke. The 3 blind spots, in the order they cause the most damage:

• Seed inboxes are not buyers. Providers decide placement partly on how a specific recipient has treated your past mail, and a seed inbox has none of that history. The test reflects your reputation and your content, never the human on the other end.
• It says nothing about list quality or relevance. A flawless score on a stale, mistargeted list still books nothing. Reaching the inbox of the wrong person at a 95 percent rate is a 95 percent waste. The fix is upstream, in how to define an ICP for cold email and what an ICP actually is.
• It is a snapshot of a moving target. Reputation shifts daily as engagement, volume, and provider algorithms change. A test that was clean last month is evidence about last month, and the green score you are trusting is already stale.

The throughline is that a placement test measures the pipe, not the water running through it. It confirms mail can physically reach an inbox, which is real and necessary. It is silent on whether you are sending the right message to the right person often enough for providers to keep trusting you, and that silence is exactly where campaigns die. Engagement shows up in positive reply rate and open rate long before a seed test flinches, and the lever that moves it is personalization at scale, not another infrastructure pass.

Read those 3 numbers as one rule. The first is the bar the test is good at measuring. The second is the number the provider is actually grading you on, and no test result survives crossing it. The third is why the test will never tell you whether the campaign works. Hold all 3 and the tool becomes a sharp instrument instead of a false comfort.

How We Use Placement Tests on a Live Invite Campaign

Our sending is podcast invites, not pitches, and that changes where the risk sits. An invite asks a senior operator to come on a show, so the copy reads like a compliment rather than an offer, which tends to keep content signals clean. What it does not change is the infrastructure underneath, and invite volume is high enough that a quiet reputation slide costs real recordings. The invite-specific version of this layer is in podcast invite deliverability and domains and warmup for podcast invites.

The practitioner move is to put the test in its lane and keep it there. A green score is permission to keep sending, never proof the campaign is healthy. When the test fails, the diagnosis is technical and the playbook above runs top to bottom. When the test passes and bookings are flat, the diagnosis is the opposite and we stop looking at infrastructure entirely, because every hour in the DNS panel is an hour not spent on the list. The failure modes specific to invite sending are in podcast invites and the spam folder.

That split is what keeps a quarter from disappearing. The most common deliverability mistake we see is an operator with a clean placement test rebuilding their domain setup over and over while the real problem, a bad list, sits untouched. The test would have told them the pipe was fine if they had read it for what it measures. Our commitment to clients is 30 recorded conversations with their ideal buyers in 90 days or their money back, and holding that means knowing within an hour which half of the system to fix, the layer that reaches the inbox or the layer that earns the reply.

The Practitioner Takeaway

An inbox placement test is a smoke detector, not a scoreboard. It does one job well, confirming your sending path is healthy enough that mail physically reaches an inbox, and it catches broken authentication, blocklist listings, and content problems faster than anything else available. Run it at the end of warmup, before every volume step up, after every change, and weekly while a campaign is live. Read the per-provider split rather than the blended score, and read the authentication and blocklist panels before the percentage, because those are the panels that name the fix.

What it cannot do is tell you whether the campaign will work. Seed inboxes carry no engagement history, the test send is not your campaign, seed addresses age badly, and filtering is personalized per recipient, so a green score is a weaker claim than it sounds. A perfect placement rate on the wrong list still books nothing. Treat the test as the gate it is, pair it with provider telemetry on one side and reply rate on the other, and you always know whether to repair the pipe or rewrite the message. The senders who scale cleanly are not the ones with the highest placement score. They are the ones who know exactly what that score is and is not telling them.