Email Deliverability vs. Inbox Placement: What's the Difference?

Here is a scenario that plays out in almost every outbound team at some point. The dashboard says 98.7% delivered. Bounces are under 1%. Nothing is technically broken. And yet the campaign that pulled a 4% reply rate two months ago is now pulling 0.6%, with a suspicious pattern: nobody at Gmail-hosted companies is replying anymore.

The dashboard isn't lying. It just isn't measuring the thing you care about. "Delivered" and "in the inbox" are two separate events, and the gap between them is where most cold email campaigns quietly die.

{"h2":"The short version"}

Deliverability (more precisely, delivery rate) measures whether the receiving mail server accepted your message instead of rejecting it. Inbox placement measures where that accepted message ended up — Primary inbox, Promotions, Updates, spam folder, or silently quarantined.

A message that lands in the spam folder counts as delivered. So does a message that gets routed to Promotions, or one that Microsoft drops into a quarantine your recipient will never check. From the sender's point of view, all three are indistinguishable from success.

{"table":{"headers":["","Delivery rate","Inbox placement rate"],"rows":[["What it measures","Was the message accepted by the receiving server?","Which folder did the message land in?"],["How it's measured","SMTP response codes and bounce handling","Seed mailboxes, recipient reports, or engagement inference"],["Who can see it","Every sending tool, automatically","Nobody, unless you deliberately test for it"],["Typical healthy number","97-99%","Highly variable — 40% to 90%+"],["What a bad number means","List hygiene or authentication problem","Reputation, content, or engagement problem"],["How fast it changes","Slowly","Overnight"]]}}

{"h2":"What delivery rate actually tells you"}

When your mailbox hands a message to Google's or Microsoft's inbound servers, you get an SMTP response. A 250 means accepted. A 550 means rejected — bad address, blocked sender, policy violation. A 4xx means try again later.

Your sending tool records the 250s as delivered and the 550s as hard bounces. That's it. The conversation ends the moment the receiving server says "got it." Everything that happens afterward — content scanning, reputation lookups, machine-learning classification, folder routing — happens inside the receiving system and generates no feedback to you at all.

So delivery rate is a genuinely useful metric, but for a narrow set of problems:

{"ul":["List quality. Bounce rates above 2-3% mean you're sending to dead addresses. Verify before you send.","Hard blocks. A sudden spike of 550s with policy language in the response means a provider has blocklisted your domain or IP outright.","Authentication failures. Some providers reject outright on a DMARC fail rather than filtering.","Rate limiting. A run of 421 or 4.7.x deferrals means you're pushing volume faster than the receiver wants to take it."]}

What it cannot tell you: whether a single human being ever saw the email.

{"h2":"Why placement is invisible by default"}

There is no header, no callback, no API that tells you "this message went to spam." Gmail does not report it. Microsoft does not report it. The only entities that know are the receiving system and the recipient.

This is not an oversight — it's deliberate. If senders could see exactly which messages were filtered, spammers could iterate against the filter in real time. So the filtering decision stays a black box, and legitimate senders inherit the same blindness.

That blindness has a nasty second-order effect: placement can collapse without any metric moving. Your delivery rate stays at 98%. Your bounce rate stays at 0.4%. Your open rate might even stay flat, because open pixels fire from spam folders and from security scanners that pre-fetch images. The first signal most teams get is a slow decline in replies, six weeks after the actual damage started.

{"h2":"The three ways to see placement"}

1. Seed mailboxes. You maintain a set of real mailboxes across Gmail, Outlook/Microsoft 365, Yahoo, and a few corporate hosts. You include them in a campaign send — or send the exact campaign content to them — and then check, per provider, which folder each one landed in. This is the only method that gives you a direct read on the folder.

2. Recipient feedback. Someone replies with "found this in spam." Accurate, honest, and far too late to be a control system.

3. Engagement inference. If your reply rate at Microsoft-hosted domains drops to a third of your Gmail rate over the same list, that's a placement signal even without seed data. Segment your reply rates by recipient MX provider and the pattern usually shows up before anything else does.

Most teams should run 1 and 3 together. Seeds tell you where you are today; segmented engagement tells you whether it's getting worse.

{"h2":"How to read seed test results honestly"}

Seed tests are directional, not precise. A few caveats worth knowing before you make decisions on them:

{"ul":["Gmail filtering is personalized. Placement depends partly on the relationship between sender and recipient — prior replies, contacts list, other people at the same domain engaging. A seed mailbox has no relationship with you, so it tends to represent the harsh end of the range.","Consumer Gmail behaves differently from Google Workspace. So does consumer Outlook.com versus Microsoft 365 with Defender. If your list is B2B, weight your seeds toward business hosting.","Seed mailboxes that only ever receive test mail drift. A mailbox with zero outbound and zero real reading habits starts looking artificial. Rotate them and give them some genuine activity.","One test is a data point, not a trend. Placement fluctuates. Run the same content weekly and watch the direction.","Test the content, not just the sender. The same mailbox can inbox one sequence and spam another. If you change your copy, links, or attachments, retest."]}

{"h2":"Different problems, different fixes"}

Once you separate the two metrics, diagnosis gets a lot faster.

High bounces, unknown placement: fix the list first. Verification, not warm-up, is the lever.

Delivery fine, placement poor at every provider: usually content or authentication. Check SPF, DKIM, and DMARC alignment — the plain-English setup guide covers what actually needs to be right. Then look at link domains, image-to-text ratio, and spam-trigger phrasing. Run the draft through a spam scorer before you send it to 500 people.

Delivery fine, placement poor at one provider only: that's reputation, and it's provider-specific. Microsoft and Google score you independently. A domain that's fine at Gmail and quarantined at Microsoft usually means complaints or spam-trap hits on Microsoft-hosted lists.

Placement good at low volume, bad above a threshold: you're outrunning your reputation. Reduce per-mailbox daily volume or add mailboxes rather than pushing existing ones harder.

Placement declining week over week with no change to content: engagement quality. Low reply rates, high delete-without-read, rising complaint rate. This is the hardest one to fix because the answer is usually "your targeting is wrong," not "your setup is wrong." Our breakdown of 15 reasons cold emails go to spam walks through the specific causes in order of how often they're the culprit.

{"h2":"Stop optimizing the metric you can already see"}

Delivery rate is easy to measure, which is exactly why so many teams over-index on it. It's the streetlight, not the keys. A 99% delivery rate with 45% inbox placement is a worse outcome than 96% delivery with 85% placement, and only one of those numbers shows up on a default dashboard.

The practical shift is small: treat delivery rate as a hygiene check you glance at weekly, and treat placement as the number you actually manage. Measure it deliberately, segment it by provider, and react to direction rather than to any single test.

{"h2":"How Sendvanta handles this"}

Sendvanta reports both, separately, because they mean different things. Seed-mailbox inbox-placement tests send your real campaign content to a spread of provider mailboxes and report the folder it landed in per provider — Primary, Promotions, or spam — so placement stops being a guess. Drafts get SpamAssassin-scored before they go out, so content problems surface before a campaign runs.

On the engagement side, opens and clicks are classified against bot and security-scanner behavior, so a Defender link-scan doesn't get counted as interest that inflates your numbers and hides a placement problem. And deliverability health scoring runs across mailbox, domain, DKIM, and IP layers — when risk thresholds are crossed, sending slows or pauses automatically instead of quietly burning the domain.

You bring your own mailboxes and your own leads; nothing routes through shared IPs. The free plan is $0 forever with 1,000 active leads, 3,000 emails a month, and up to 10 of your own SMTP mailboxes — enough to run placement tests against your real content before you commit to anything. Paid plans start at $29/mo. More detail on what's included.

{"faq":[{"q":"Is a 99% delivery rate good?","a":"It's normal, not good. Anything above 97% just means your list is clean and you aren't hard-blocked. It says nothing about whether messages reached the inbox rather than the spam folder."},{"q":"Can I tell from open rates whether I'm in spam?","a":"Not reliably. Open pixels fire from spam folders and from corporate security scanners that pre-fetch images, so open rates can stay high while placement collapses. Reply rate segmented by recipient mail provider is a far better proxy."},{"q":"How often should I run inbox placement tests?","a":"Weekly for active campaigns, and again any time you change copy, links, sending domain, or volume. Placement moves faster than most people expect."},{"q":"Does landing in Promotions count as inbox placement?","a":"It's better than spam and worse than Primary. For B2B cold email to business mailboxes, Promotions is mostly a consumer-Gmail issue, but if your seeds land there consistently your content is reading as marketing rather than a one-to-one message."},{"q":"If placement is bad, should I warm up more or fix content first?","a":"Test content first — it's faster and free. Run the draft through a spam scorer and seed-test the exact copy. If clean content still lands in spam, then it's a reputation problem and warm-up plus lower volume is the right response."}]}