Midnight is not a send window

There is a particular kind of founder who connects an inbox on Friday, uploads 4,000 rows, and lets the tool “run over the weekend.” By Monday the mailbox looks busy. By Wednesday Gmail is teaching them about spam folders. The tool did what it was asked. The founder asked it to impersonate a night-shift SDR who does not exist, and then acted surprised when the people receiving mail at 1:14am reported it.

LeadHootz will not send a step that sits outside the campaign’s day-of-week window. Held is not failed. Held is the product refusing the request. That refusal is unpopular with people who bought “set and forget.” It is the only feature standing between a fresh domain and a reputation you will spend a quarter trying to get back. Cloning a sequence takes an afternoon. A domain that Gmail has decided is unwanted takes longer than your runway wants to hear.

The number Gmail actually published

This is not a vendor’s “best practice” blog. Since February 2024, Gmail has required senders to keep the user-reported spam rate in Postmaster Tools below 0.30%. They tell you to stay under 0.10%, and they are explicit that rates above 0.10% already hurt inbox placement. From June 2024, bulk senders above 0.30% are ineligible for mitigation — meaning when you are in the hole, they will not help you out until you have spent seven consecutive days back under the line. Bulk, in their definition, is about 5,000 messages a day to personal Gmail accounts from the same primary domain. A “weekend run” across a couple of mailboxes is how a small company wanders into a rule written for publishers. The page is Gmail’s sender guidelines. Read the spam-rate section before you read another thread about subject lines.

Look at the scale. You can have a dashboard full of opens and still be one irritated percent of a percent away from a domain problem. The 20% bar is not a Gmail limit. It is a reminder that the number founders celebrate and the number Gmail punishes are not in the same sport. Complaint rate is not open rate. A single campaign that feels like spam to a few dozen people, on a mailbox that only sends a few hundred messages, is enough to move 0.10% to 0.30%. Volume makes that easier, not safer. “We’ll warm it later” is how you arrive at the threshold with no warm left.

Timezone is the clock, not a preference

Account timezone is where daily caps reset. New campaigns inherit it. Override it per campaign when you sell into a different market — London hours for UK buyers, not your laptop’s clock in another country. If timezone is empty, you are guessing, and guessing is how 2am mail happens while you sleep and feel productive. The buyer does not experience your productivity. They experience a timestamp. If you would be embarrassed to have them notice it, the agent is not allowed to send it. That is the whole rule. Everything else is implementation.

Steps outside the window are held, not sent, and not marked failed. Failed is a bounce or a disconnected mailbox. Held is a decision. Live Activity should show the hold. If your ops review treats holds as errors and “fixes” them by widening the window to all night, you have automated the embarrassment.

Caps are a reputation budget

Set the daily cap to a number a human could actually send from that mailbox and still read the replies. Inbox sending is personal on purpose. Twilio SendGrid is for after you have authenticated the domain — SPF, DKIM, and DMARC, which Gmail also requires once you are a bulk sender, with a DMARC policy of at least p=none — and you have genuinely outgrown a real inbox. It is not a shortcut around warming, and it is not a place to hide a from-name that is almost you. A shared IP you have never heard of will not love you more than Gmail loves a person who mails at midnight.

Raising the cap because the queue is deep is the same move as the weekend blast, spread across weekdays. The queue is not a stakeholder. The people in it are. A fresh buying signal emailed at 2am is still a bad email. The signal can wait until their morning. If it cannot, it was not a signal. It was your anxiety.

What to set before the first real send

1. Timezone in Settings, then confirm the campaign inherited it. Quick start shows where it lives.
2. Weekday windows in the buyer’s hours. Not yours, if you are not in their afternoon.
3. A daily cap you would defend out loud. If the number needs a justification longer than a sentence, it is too high.
4. Stop-on-reply, and a named human in Unibox. A window that delivers mail nobody answers is just a polite way to annoy people on schedule.
5. SendGrid only after the domain is authenticated and the inbox cap is the actual bottleneck. The send-window note and the readiness checklist are the same argument with boxes to tick.

If you would not want the prospect to screenshot the timestamp and send it to their group chat, do not let the job queue send it. Windows exist so you never have to make that call at 2am because the worker was hungry. Gmail has already published the price of ignoring them. It is not a theory. It is 0.30%, and then they stop taking your calls.