Enrichment Desk

Explainer / Outreach and rules

What email-sending agents must do for Gmail, Yahoo and Outlook

Gmail, Yahoo and Outlook sender rules for AI agents: domain authentication, bulk thresholds, unsubscribe requirements and the controls to check before sending.

Enrichment Desk · · 5 min readShare

An email-sending agent can cost the company its domain. Yahoo ties reputation to the sending IP and DKIM domain, so unwanted outreach can damage more than one campaign. Before enabling an agent, configure domain authentication and make permission, suppression and sending limits controls it cannot bypass. Passing those controls does not promise a place in the inbox.

Which sender rules apply to an email agent?

The recipient service and sending domain determine which requirements apply. An agent does not acquire a separate allowance simply because it has its own API key or campaign name.

As of October 8, 2026, the providers document the following requirements. The authentication column describes bulk senders; Gmail and Yahoo set rules for every sender too, below.

Microsoft's two threshold descriptions differ. Treat 5,000 as the point to have bulk authentication in place, not as a safe daily allowance.

For Gmail, the primary-domain count includes subdomains. A separate agent subdomain therefore does not create a fresh threshold. Count mail across the company's sending systems, rather than trusting the agent's own activity log to describe the domain's volume.

Below bulk volume, Gmail and Yahoo still require SPF or DKIM, valid forward and reverse DNS, and spam rates below 0.3%. Gmail also requires TLS and RFC 5322 formatting; Yahoo names RFCs 5321 and 5322. These baseline requirements apply to every sender, not only to bulk campaigns.

What must the sending service authenticate?

It must authenticate the domain the recipient sees, not merely a vendor's unrelated domain. Microsoft's troubleshooting page explains the distinction: publishing a DMARC record is one step; passing DMARC requires SPF or DKIM to align with the message's From domain.

Ask the sending service which domain it uses for the return path and DKIM signature. Then inspect a message sent through that service. A dashboard saying DNS is configured cannot tell you whether this particular sending path produced aligned authentication.

The minimum policy is also easy to misread. A p=none policy is permitted; it does not instruct receivers to quarantine or reject messages that fail DMARC. Authentication configuration and protection against forged mail are related decisions, not interchangeable ones.

Where are the complaint targets and penalty lines?

For Gmail, aim below 0.1% user-reported spam and avoid reaching 0.3%. Since June 2024, a bulk sender above 0.3% has been ineligible for mitigation; Google's FAQ requires seven consecutive days below 0.3% to restore eligibility. The target and penalty line are different: a rate below the penalty line is not a promise of good placement.

Yahoo requires less than 0.3% and calculates that rate against inbox-delivered mail. Dividing complaints by every attempted send can understate the provider's rate. Yahoo says noncompliant mail can be rejected or placed in spam.

For Microsoft bulk authentication failures, the concrete consequence is 550 5.7.515 rejection. Its policy page also describes junk placement and possible later rejection. Preserve the actual response in the sending log; do not let the agent turn a permanent failure into an endless retry loop.

Which decisions must stay outside the agent's prompt?

Put permission and suppression in the sending system. Our recommended boundary is simple: the agent may propose the message and recipient; a separate gate decides whether that combination may be sent now.

An address found or verified by an agent does not establish permission. Google advises against purchased lists and mail to people who did not subscribe. Lawful use of contact data remains the buyer's responsibility.

Recheck suppression immediately before sending, including queued follow-ups. An opt-out that arrives after drafting must still stop the scheduled message. Google's current enforcement table lists unsubscribes taking more than 48 hours as grounds for losing delivery support or mitigation; Yahoo's two-day requirement appears in the table above.

Do not let the agent decide that a personalized sales message is transactional merely because it addresses one recipient. Google and Yahoo exclude transactional messages from their promotional one-click requirement, but that exception is not a general exemption for agent-written mail.

What should you check before enabling sends?

Use mailboxes you own to check the complete sending path. Inspect SPF, DKIM and DMARC results in the received headers, then exercise the unsubscribe mechanism and confirm it blocks the next queued marketing message.

For Gmail's one-click implementation, check both List-Unsubscribe and List-Unsubscribe-Post, using the documented HTTPS POST arrangement. A link at the bottom of the message is not a substitute. Check complaint reporting too: Yahoo's feedback loop depends on enrollment of the DKIM domain, which the sending service may handle for you.

These checks establish that the controls work, not that future messages will land in the inbox. Before widening the agent's sending permission, make one suppressed recipient remain suppressed through a scheduled follow-up. The company owns the domain whose reputation is exposed; the agent should never own the decision to ignore an opt-out.

Frequently asked questions

Does every email agent need bulk-sender authentication?

Bulk rules depend on the sender's classification, not whether software wrote the message. Configuring SPF, DKIM and DMARC before increasing volume avoids making authentication an emergency at the threshold.

Does a new subdomain reset Gmail's bulk threshold?

No. Gmail combines subdomains under the primary domain for its volume count, and bulk status does not expire once assigned.

Does Gmail accept a mailto link for one-click unsubscribe?

No: its FAQ requires the RFC 8058 implementation. Yahoo accepts mailto, although it recommends POST.

Does passing authentication guarantee inbox delivery?

No. Yahoo explicitly makes no inbox-placement guarantee; authentication does not establish that recipients want the message.

Read this article as Markdown Open in ChatGPT Open in Claude