The spreadsheet says week three is 15,000 recipients. Gmail started deferring at 6,000. The useful move is to throw the spreadsheet’s next row away until you know why 6,000 was too many, or too cold, or pointed at a domain that is not authenticated yet.
Warmup fails when the next slice of mail is worse news than the last slice. Sending the slice after that does not average the bad news out.
Last verified: 26 September 2026. Ramp mechanics stay on the Klaviyo warmup guide. This page does not publish a new day-by-day schedule.
Stop, then name what broke in the IP warmup
| What broke | What it usually means | What to do |
|---|---|---|
| 4xx replies climbing as volume climbs | A throttle or greylist. See deferrals. | Hold volume. Let retries run. |
| 5xx mailbox unknown on the new slice | The new slice is dirtier than the warm one | Suppress hard bounces. Do not add a wider slice. |
| Accepted, then spam, one provider | Placement, not a bounce | Use provider-specific diagnosis. |
| 550 5.7.515 | Outlook is rejecting the From domain | Stop and fix authentication. The Outlook guide is the procedure. |
How Klaviyo warms shared and dedicated IPs explains what the ramp is for. The warming guide is the operational checklist. Neither one is a promise that a calendar date repairs a complaint rate. If you want the troubleshooting checklist rather than the explanation, the failed warmup fix page has it.
Why more mail is the wrong repair
Filters learn from recipients who got the message. If those recipients ignore it, delete it, or mark it spam, the next thousand recipients teach the same lesson faster. That is true on a dedicated IP, where the history is yours, and it is true on a shared IP, where you can also inherit the pool’s current mood. A larger send during a failed warmup adds complaints and bounces faster than it adds good history; it does not dilute them. The one exception is volume to people who already engage, which is why the next paragraph is about that slice and nothing else.
The engaged slice is the only mail that can add a positive mark while you are in trouble: recent buyers, recent openers, people who expect the message. A win-back, a purchased file, or a “just in case” segment is how a failed warmup becomes a reputation problem with no end date. Providers do not publish how many good days erase a bad spike.
Resume on evidence, not on the original plan
- Keep SPF, DKIM, and DMARC aligned on the domain in the From line. A warmup failure is a bad time to rotate domains. If you do need a new domain, that is a planned change, with its own ramp, after this incident is contained.
- Send the smaller engaged slice. Check placement at Gmail, Yahoo, AOL, and Outlook. InboxEagle’s seeds cover those four. Also read the bounce log, since Outlook authentication rejections happen at SMTP time and never reach a folder.
- Check Postmaster spam rate and delivery errors before you add the next band of subscribers. A deferral wave can be over while the spam rate is still the thing that will block mitigation at Gmail.
- Add one band. Wait for that band’s result. The original week-three target is a guess you already tested. When you rebuild the plan, the email warmup planner lets you restart the schedule from the volume that last held instead of from day one.
Seasonal pressure does not change the rule. BFCM preparation and volume spikes assume the ramp is healthy. A broken ramp during a sale is a reason to mail buyers you already have, not the full list you hoped to have by Friday.
InboxEagle
Rebuild the ramp from the last volume that held
Plan the next warmup schedule starting from the send size that still placed, not the row the old spreadsheet expected. Then test each step before you add recipients.
Sources
- Email sender guidelines FAQ, Google. Spam-rate mitigation eligibility. Not a warmup calendar.
- Sender best practices, Yahoo. Bulk complaint-rate requirement. No instruction to increase volume through a failure.
- Understanding bounced emails, Klaviyo. How temporary and permanent failures are recorded while a ramp is running.