Skip to content
Resources · Blog

Cold Email Deliverability: A Pre-Send Checklist 

Google, Yahoo, and Microsoft now set the floor for cold email. A pre-send checklist for DNS records, unsubscribe headers, mailbox limits, bounces, and warm-up.

November 10, 2026 5 min read
Cold EmailDeliverabilityEmail Authentication

A cold email campaign is ready to send when four things are true: the sending domain authenticates with SPF, DKIM, and an aligned DMARC record; every message carries a working one-click unsubscribe; each mailbox stays inside a conservative daily volume; and you have decided in advance which bounce rate stops the campaign. This checklist explains each item and how to monitor it once you are live. The rules come from the mailbox providers themselves. Google and Yahoo began enforcing bulk-sender requirements in February 2024, and Microsoft started enforcing its own for Outlook.com, Hotmail, and Live addresses on May 5, 2025, for senders above 5,000 messages a day. None of this guarantees inbox placement, which still depends on your reputation, your content, and your recipients. It does remove the failures that sit entirely within your control.

1. Authenticate the Domain You Send From

SPF lists the servers allowed to send for your domain. DKIM adds a cryptographic signature that receivers check against a public key published in your DNS. DMARC tells receivers what to do when those checks fail, and it requires alignment: the domain in your From address must match the domain that passed SPF or DKIM. Google's sender guidelines require SPF and DKIM for bulk senders, accept a DMARC policy of p=none as the minimum, ask for a DKIM key of at least 1024 bits with 2048 recommended, and require TLS and valid forward and reverse DNS. Yahoo's requirements are close, with relaxed alignment allowed. A few details catch teams out. An SPF record that needs more than 10 DNS lookups fails even when its syntax is correct, so audit the includes before adding another tool. The include also has to match the service that sends your mail: a record written for Google Workspace does nothing for mail relayed through Amazon SES or Mailgun. Finally, know who signs DKIM. Gmail, Microsoft 365, and the large sending services sign with keys you configure in their consoles; if you run your own SMTP server, the signing has to happen there or in the tool that sends through it. Many senders also keep cold outreach off their primary domain and send from secondary domains. That is vendor guidance rather than a provider rule, but it limits the damage if a campaign goes wrong.

2. Make Unsubscribing One Click, and Honor It Fast

Google requires marketing and subscribed messages from bulk senders to support one-click unsubscribe as defined in RFC 8058: an HTTPS List-Unsubscribe header plus List-Unsubscribe-Post: List-Unsubscribe=One-Click. Google and Yahoo both expect unsubscribes to be honored within two days, and Yahoo also asks for a visible link in the body. In the United States, CAN-SPAM adds accurate headers, a physical postal address, and a working opt-out to every commercial message. Treat legal summaries, including this one, as general information rather than legal advice; the FTC, the ICO, and the CRTC publish the primary guidance. The operational trap is timing. Many sequences check a contact's status when the contact is enrolled, then wait days before the next step. If someone unsubscribes or bounces in that window, the queued email still goes out unless the tool checks suppression again at the moment of sending. Ask your vendor where that check happens, then test it: unsubscribe a test contact while a step is pending and confirm nothing arrives.

3. Cap Volume Per Mailbox, Then Warm Up Slowly

Google Workspace allows 2,000 messages per mailbox per day, but that ceiling is beside the point for cold outreach. Practical guidance from cold email vendors is 25 to 50 cold sends per mailbox per day; Lemlist suggests 30 to 40, and Apollo's knowledge base cites 50 to 100 on a rolling 24-hour window. Reputation builds per mailbox and per domain, so set the limit on each mailbox and spread a campaign across several mailboxes rather than raising the daily total on one. Send windows in the recipient's time zone and a random delay between messages keep the pattern closer to a person working through an inbox. New mailboxes need a ramp. Vendor guidance ranges from about three weeks of warm-up to Lemlist's roughly 40 days plus 14 days of stabilization. Most tools warm up through a shared network that exchanges mail with other customers' inboxes. That is fast and usually included in the price, but users have reported open rates falling 30 to 40% after joining a shared pool. Those reports were relayed by competing vendors, so read them with care. The alternative is warming up only between mailboxes you control, which is slower and needs at least two mailboxes, but keeps your reputation tied to your own sending.

4. Clean the List and Set a Bounce Threshold Before Launch

Hard bounces are the fastest signal that a list is bad, and providers treat them that way. Verify addresses before the first send: correct syntax, a working MX record, and flags for disposable domains and role addresses such as info@ or sales@. Third-party tests, many of them published by competing vendors, have reported double-digit bounce rates on contact exports labeled verified, so verify again yourself rather than trusting a source's label. Then decide which bounce rate stops a campaign, and write it down before launch, because tools differ widely. Snov.io pauses at 15% of the last 100 sends. MailTig pauses a campaign once its bounce rate reaches 5% after 20 sends. Marketing platforms are stricter still: Brevo can suspend an account at a 2% hard-bounce rate on recent campaigns, and MailerLite at 5% bounces or 0.2% complaints. A threshold set in advance turns a judgment call made under pressure into an automatic pause with a reason attached.

5. Watch Complaints and Replies, Not Opens

Google asks senders to keep the spam rate shown in Postmaster Tools below 0.3%, ideally below 0.1%, and to avoid ever reaching 0.3%. Yahoo uses the same 0.3% ceiling, computed only on mail delivered to the inbox. Complaints are the metric that ends sending programs, so check them weekly and per domain. Open rates are a weak guide. Apple Mail Privacy Protection loads tracking pixels automatically, and one estimate puts the resulting inflation of open rates at 15 to 40%, which is why reviewers increasingly judge tools by replies and placement tests. Track replies, bounces, and unsubscribes per mailbox, not only per campaign, so one damaged mailbox shows up before it drags the rest of the pool down. A mailbox whose OAuth token has expired or whose IMAP login fails should raise an alert, because a disconnected mailbox that keeps getting scheduled is a common source of silent failure.

The Checklist, in Order

Before launch: publish SPF, DKIM, and DMARC for the sending domain and confirm they resolve; check the SPF lookup count and the include for the service that sends your mail; confirm who signs DKIM; add one-click unsubscribe headers, a visible unsubscribe link, and your postal address; set a daily limit on every mailbox; warm up new mailboxes; verify the list; and choose a bounce threshold. After launch: watch complaints, bounces, and replies per mailbox, and confirm suppression holds for anyone who opts out mid-sequence. We built most of these checks into MailTig. Its preflight lists blockers and warnings covering connected mailboxes, DKIM mode, the postal address, audience, credits, and schedule, and it will not start a campaign while a blocker remains. It rechecks suppression immediately before every send, and its warm-up runs only between your own mailboxes. The MailTig case study explains the research behind those choices. A passing check proves your records resolve, not that you will reach the inbox, so keep watching the numbers after launch.

Have a project that needs this kind of thinking applied to it?