What is inbox placement?

Providers deliberately withhold this. Telling senders exactly which messages were filtered would hand spammers a feedback loop for tuning past the filter, so the data is never published per campaign.

Seed testing approximates it by sending to controlled accounts across Gmail, Outlook, Yahoo and others, then recording the folder. It is genuinely useful for spotting a sudden change, and it does not reflect the personalized part of filtering.

The personalization is significant. Gmail decides partly on how this recipient has treated your mail before, so a subscriber who always opens sees the inbox while a dormant one does not, from the same send.

Proxy signals fill the gap. Open rates by provider, reply rates and the Gmail dashboard together show a real drop before your revenue does.

Say you keep twelve test mailboxes: four at Gmail, four at Outlook.com and four at Yahoo. After Tuesday's campaign you log in to each and write down the folder. Gmail shows three in Promotions and one in Primary. Yahoo shows four in the inbox. All four Outlook.com accounts have the message in Junk.

A four-out-of-four result at one provider and clean results elsewhere is about as clear as seed testing gets. It tells you the problem is specific to Microsoft and probably recent, and it tells you not to waste the week rewriting content that Gmail and Yahoo were happy with.

It doesn't tell you what share of real Outlook subscribers saw Junk. Seeds have no history with you, so they show the verdict a stranger's mailbox would get, which is the pessimistic case.

Microsoft writes part of its decision into the message, and you can read it. Open one of those junked test emails, view the message source, and look for the X-Forefront-Antispam-Report header. Inside it is a field labeled SCL, the spam confidence level. Low values such as 0 or 1 mean the filter judged the message clean. A value of 5 or higher is what sends it to Junk.

A second header, Authentication-Results, shows how SPF, DKIM and DMARC were evaluated on arrival. When the SCL is high and any of those three read fail, fix the authentication before anything else.

Gmail doesn't expose a comparable score. What you can do there is check the banner at the top of a message in the spam folder, which often gives a short reason, such as similarity to messages other users reported. It's vague, but it does separate a reputation problem from an authentication one.