Every mailbox provider caps how much you can send. The caps are rarely documented precisely, they change without announcement, and they are applied more aggressively to accounts without an established history. Here is what is broadly true, and more importantly, how the enforcement actually behaves.
The approximate numbers
| Account type | Approximate daily ceiling | Notes |
|---|---|---|
| Consumer Gmail | ~500 recipients | Rolling 24 hours, not calendar day |
| Google Workspace | ~2,000 recipients | Lower over SMTP than via the web interface |
| Microsoft 365 | ~10,000 recipients | Also subject to a per-minute recipient rate |
| New accounts, any provider | Substantially lower | Throttled until a sending history exists |
Treat every figure above as an upper bound. A brand-new Workspace account will not send 2,000 messages on its first day without consequences, whatever the documentation implies.
Rolling windows, not midnight resets
A frequent misunderstanding: these are rolling 24-hour windows, not daily counters that clear at midnight. Sending 500 messages at 4pm does not free you up at 00:01 — it frees you up at 4pm the following day, gradually, as the earlier sends age out of the window.
This is why a campaign that "worked yesterday" can be throttled today when you run it a few hours earlier.
What enforcement looks like
Providers enforce with temporary rejections rather than permanent ones, which is genuinely good news — your mail is not lost and the addresses are not invalid.
- 421 — Service unavailable or throttling. Temporary. Stop and retry later.
- 429 — Too many requests. Temporary. Same response.
- 4xx generally — Transient. Retry with a backoff.
- 5xx — Permanent. The address or the policy is the problem; retrying will not help.
The critical behaviour is what you do next. Backing off on the first 421 is correct. Retrying immediately, repeatedly, is read as abusive and typically extends the throttle window — turning a twenty-minute pause into something much longer.
Handling it automatically
On Email.biz.pk a 421 or 429 puts that account into a 60-minute cool-down recorded as an expiring key, and the account is skipped in the sending rotation until it clears. The remaining accounts keep working, so the campaign slows rather than stopping. If every account ends up cooled down, the campaign pauses and alerts you rather than continuing to accumulate failures.
Scaling horizontally is the only real answer
You cannot negotiate a meaningfully higher limit, and you should not want to. The limit exists because volume-per-sender is the signal filters use. An account sitting at its ceiling is, by definition, sending at the rate that looks most like bulk mail.
The alternative is to add accounts. Ten mailboxes at a comfortable 150 messages a day each will deliver 1,500 messages more reliably than one mailbox pushed to 1,500 — because each of the ten looks like an ordinary person doing ordinary work.
Applying a randomised delay between sends compounds the effect. A 25 to 45 second gap holds each account near 80 to 144 messages an hour, a cadence no throttle will object to, and one that does not resemble a script.
Practical guidance
- Stay well under the published ceiling. Sending at 60% of a limit is dramatically safer than sending at 95%.
- Ramp new accounts over two to four weeks rather than starting at full volume.
- Never retry immediately after a 4xx. Back off, and let the account rest.
- Track which accounts get throttled. A single account hitting limits repeatedly while others do not usually indicates a reputation problem specific to it.
- Remember that recipients, not messages, are usually what is counted. One message to fifty people counts as fifty.