MailStrike logoMailStrike.ai

Troubleshooting

When warming isn't warming.

Most warmup problems look the same on the dashboard: a flat-or-falling line where there should be an upward curve. Here are the five real causes, in roughly the order we see them.

8 min read·Updated May 2026

Warming is mostly hands-off. The benefit is also the bug: when something goes wrong, you may not notice for days because there's nothing for you to actively be looking at. The dashboard is the first place to check, and the patterns are surprisingly recognisable once you've seen each one a couple of times.

Below are the five issues we see most often, in roughly the order of frequency, with what to look for and how to fix each.

1. Score stuck below 70 with daily activity flat

This is the classic “why isn't this working” pattern. The score sits at 64 or 67, the daily-sent chart looks normal, and nothing seems to be moving.

Cause, 80% of the time:infrastructure. SPF is missing the warming sender, DKIM isn't actually signing, or DMARC is misconfigured and receiving servers are rejecting outright. The score is low because warming emails aren't even reaching the persona inbox to be opened.

How to verify: run the domain through the Email Auth Checker. Look specifically for the SPF section — if it says permerroror shows more than 10 DNS lookups, that's your problem. SPF flattening or removing unused providers fixes it.

The single most common SPF mistake is having “include:” statements that recursively pull in more SPF records than you realize. Three top-level includes can easily add up to 14 lookups, which makes the entire record fail with no fallback.

2. Daily volume not ramping up

You set a Standard ramp (5 → 30 over 4 weeks), but on day 14 you're still sending 5 a day.

Cause: nearly always a provider-side rate limit. Google Workspace caps new OAuth-connected accounts at a low daily quota until the account has been active for a couple of weeks. Microsoft 365 throttles similarly. The warming engine sends what it can; if the provider rejects with a 4xx rate-limit, it backs off.

Fix:two parts. First, check your sending provider's admin panel for sending limits — for Google, this is under Reports → Email. Second, send a few real emails from that account day-to-day (replies to your own threads count), which signals to the provider that the account is genuinely in use, not abandoned.

3. Mailbox went into “Disconnected” on its own

A mailbox was happily warming, then a few days later you check the dashboard and it's greyed out as Disconnected. You didn't do anything.

Cause: OAuth token revoked. This happens for boring reasons — the user changed their account password, an admin re-ran a security review, or the org enabled stricter app permissions and the OAuth grant was invalidated. For SMTP-connected mailboxes, an app password being rotated does the same thing.

Fix:click Reconnect on the mailbox row, run through OAuth again. History is preserved; warming resumes from the same ramp day. Don't add it as a brand-new mailbox — you'll lose the warming history.

4. The DKIM warning won't go away

The yellow banner about a 1024-bit DKIM key is persistent for a reason: rotating to 2048 is a real fix, not cosmetic. Major providers' 2024 sender guidelines specifically call out 2048-bit as the recommended minimum for senders over 5,000/day. You're likely to scale past that threshold during the year.

Fix in Google Workspace: Admin → Apps → Gmail → Authenticate email → Generate new record → choose 2048-bit. Publish the new TXT record at the selector Google gives you (usually google._domainkey), wait 24 hours for DNS propagation, then click Start authentication.

Fix in Microsoft 365: the new 2048-bit option lives under Security → Email authentication → DKIM. Same flow — generate, publish to DNS, wait, enable.

Rotating a DKIM key won't hurt your reputation. The new key takes over for new mail; older mail still validates against the old key. Brief overlap, no breakage.

5. Score is high but real campaigns still spam

The hardest one, and the most counterintuitive. Warming reads as healthy. You launch a real campaign. Inbox rate is 60%.

Causes, in order:

  • Sender-name mismatch. Warming sends as “Alex Park”; the campaign sends as “Acme Outreach Team.” To the filter, that's a new sender with no history.
  • List quality. Warming uses curated personas with high engagement. A purchased B2B list has high bounce rates and low opens — completely different signal. The score doesn't carry across.
  • Volume jump. Warming was 25/day. Campaign launched at 400/day. That 16× jump alone is enough to trip provider heuristics, regardless of reputation.
  • Content shift. Warming used natural-language LLM replies. Campaign uses a templated subject line repeated 400 times. Identical subjects are a notorious spam signal.

Fix: match the campaign sending profile to the warming profile as closely as possible. Same from-name, same domain, gradual volume increase (3× the warming volume on day 1, 5× by week 2). Vary subject lines per recipient at least minimally.

What to do when you're stuck

If the score is dropping and none of the above match: don't panic, and don't increase volume to compensate. The instinct is to push harder; the right move is the opposite. Pause warming on one mailbox, leave the others running, and look at what changed in the 7 days before the drop started. The 90-day placement chart on the dashboard makes the inflection point easy to spot.

And if you genuinely can't figure it out — that's what support is for. Email support@mailstrike.aiwith the affected domain and a screenshot of your dashboard. We've seen most of the weird ones.

Stop landing in spam.

MailStrike warms your domain using AI-driven personas that open, reply, and engage like real recipients. The fastest path to inbox.