Incident: Reputation Tanking Overnight

Detecting a Reputation Drop

Reputation drops manifest through several observable signals, often in combination:

Primary indicators:

• Google Postmaster Tools shows domain reputation declining from “High” to “Medium,” “Low,” or “Bad”
• Microsoft SNDS shows IP status changing from green to yellow or red
• Inbox placement drops across seed tests
• Delivery rate declines (more 4xx and 5xx responses)

Secondary indicators:

• Open rates decline without changes to content or list
• Click rates drop proportionally with open rates
• Spam folder placement increases in seed tests
• Gmail Promotions → Spam transition in seed accounts
• Bounce rates increase, particularly 5.7.1 policy rejections

Important: Open rate declines alone are not conclusive evidence of a reputation drop. Privacy features (Apple Mail Privacy Protection, proxy-based opens) can distort open rate data. Always cross-reference with Postmaster Tools, SNDS, or placement tests.

Root Cause Analysis

A reputation drop is a symptom, not a cause. The first step in recovery is identifying what caused the decline.

Checklist of Common Causes

List quality issues:

• Did you recently mail a new list segment (purchased, old, unverified)?
• Has your bounce rate spiked (> 2% for any recent campaign)?
• Have you stopped processing bounces correctly?
• Did you merge in addresses from a different system without validation?

Complaint issues:

• Has your complaint rate exceeded 0.1% for any recent campaign?
• Did you change your From name, subject line style, or content type?
• Did you increase sending frequency recently?
• Is your unsubscribe process broken or difficult?

Infrastructure issues:

• Did you change ESPs or sending infrastructure?
• Were there changes to DKIM keys, SPF records, or DMARC policy?
• Did you migrate to a new IP without proper warmup?
• Is a shared IP affected by another sender’s behavior?

Content/engagement issues:

• Has engagement declined over the past 30 days?
• Did a recent campaign perform significantly worse than average?
• Are you still sending to subscribers who haven’t engaged in 180+ days?

External issues:

• Were you added to a blocklist (check Spamhaus, Barracuda, SpamCop)?
• Did a provider change their filtering algorithms?
• Is there a broader industry-wide filtering change?

The Investigation Process

1. Timeline correlation: Plot your reputation metrics (Postmaster Tools, SNDS) alongside your sending activity. When exactly did the decline start? What happened in your sending program on or within 48 hours before that date?

2. Provider isolation: Is the reputation drop at one provider or all providers? Gmail-specific drops often correlate with complaint rates visible in Postmaster Tools. Microsoft-specific drops correlate with SNDS metrics.

3. Campaign isolation: Did one specific campaign trigger the decline? Compare per-campaign bounce, complaint, and engagement metrics to find the outlier.

4. Segment isolation: Did one segment of your list cause the problem? If you mailed a re-engagement segment, a cold segment, or a new acquisition channel, isolate their metrics.

The Recovery Plan

Phase 1: Stop the Bleeding (Days 1–3)

Immediate actions:

1. Reduce volume by 50–75%. Send only to your most engaged subscribers (opened or clicked in the last 30 days).
2. Pause any problematic segments. If you identified the segment that caused the issue, stop sending to it entirely.
3. Fix authentication issues. If SPF, DKIM, or DMARC alignment is broken, fix it immediately.
4. Request blocklist removal. If you’re listed on Spamhaus, Barracuda, or others, fix the root cause and submit delisting requests.
5. Verify bounce processing. Ensure all hard bounces from the problematic period are suppressed.
6. Review FBL processing. Ensure all complaints are being received and addresses are being suppressed.

Phase 2: Rebuild Engagement (Days 4–14)

Focused sending strategy:

• Send exclusively to subscribers who engaged in the last 30 days
• Choose your best-performing content (highest engagement from past campaigns)
• Send at your standard frequency — don’t send extra to “make up” for the pause
• Monitor engagement metrics daily — open rate, click rate, complaint rate

What you’re doing: You’re showing the mailbox providers that people who receive your email want it. High engagement rates from a smaller, targeted list signal that your mail is valuable. This is the fastest way to rebuild reputation.

Metrics to track during Phase 2:

Phase 3: Gradual Volume Increase (Days 15–45)

Once reputation metrics show improvement (Postmaster Tools moves from Bad/Low to Medium, SNDS shows green):

Week 3 (Days 15–21): Expand to subscribers who engaged in the last 60 days. This approximately doubles your volume from Phase 2.

Week 4 (Days 22–28): Expand to subscribers who engaged in the last 90 days.

Week 5 (Days 29–35): Expand to subscribers who engaged in the last 120 days.

Week 6+ (Days 36–45): Expand to subscribers who engaged in the last 180 days.

At each stage, verify:

• Complaint rate remains below 0.05%
• Open rate remains above 20%
• Postmaster Tools reputation stays stable or improves
• No new blocklist listings
• Bounce rate stays below 1%

If any metric degrades when expanding, stop and hold at the current engagement window until metrics stabilize.

Phase 4: Sustained Operations (Day 45+)

After full volume recovery:

• Implement a permanent engagement-based segmentation strategy
• Never mail subscribers inactive for 180+ days without a re-engagement campaign first
• Set up automated monitoring with alerts for all key metrics
• Establish a regular cadence for list hygiene (monthly validation, quarterly sunset review)

Provider-Specific Recovery Notes

Gmail Recovery

Google Postmaster Tools reputation levels and recovery expectations:

• High → Medium: Can recover in 1–2 weeks with good practices
• High → Low: Recovery takes 2–4 weeks of careful sending
• High → Bad: Recovery takes 4–8 weeks, sometimes longer
• Bad → Low: First milestone; keep current strategy
• Low → Medium: Second milestone; can begin expanding volume
• Medium → High: Final recovery; may take additional weeks to stabilize

Gmail-specific actions:

• DMARC alignment is especially important for Gmail reputation
• Gmail weighs user engagement heavily — send to your most engaged Gmail subscribers first
• Monitor the “Spam rate” metric in Postmaster Tools — target < 0.1%, ideal < 0.05%

Microsoft Recovery

Microsoft uses SNDS status and Junk Mail Reporting Program (JMRP) data:

• Green → Yellow: Moderate concern. Review SNDS data for trap hits and complaints.
• Green → Red: Serious. Likely seeing delivery failures and junk placement.
• Recovery timeline: Microsoft reputation can recover faster than Gmail (1–2 weeks with good practices), but the metrics are less transparent.

Microsoft-specific actions:

• Register for JMRP (feedback loop) to receive complaint data
• Check SNDS trap hit data — any trap hits indicate list quality problems
• Microsoft is sensitive to IP reputation — if on a shared IP, contact your ESP

Yahoo Recovery

Yahoo is particularly sensitive to complaint rates:

• Yahoo blocks aggressively once complaint rates exceed their threshold
• Recovery requires complaint rates to stay very low for an extended period
• Yahoo doesn’t offer detailed reputation tools — you’re largely flying blind
• CFL (Complaint Feedback Loop) data through Yahoo is essential for monitoring

Metrics Dashboard for Recovery

Track these metrics daily during active recovery:

Common Recovery Mistakes

Mistake 1: Trying to “make up” volume After pausing or reducing sends, some senders immediately blast their full list to compensate. This makes the problem worse.

Mistake 2: Only fixing the symptom Removing a blocklist listing without fixing the root cause results in re-listing within days.

Mistake 3: Expanding too fast Moving from engaged-only sends to full-list sends in a single jump doesn’t give the reputation model time to adjust.

Mistake 4: Not monitoring during recovery Recovery isn’t “set and forget.” Daily monitoring catches problems early and prevents setbacks.

Mistake 5: Blaming the provider “Gmail is broken” is almost never the correct diagnosis. The reputation model is reflecting your sender behavior. Fix the behavior, and the model will reflect the improvement — but it takes time.

Mistake 6: Changing too many things at once If you simultaneously change your content, sending infrastructure, authentication, and list segmentation, you won’t know which change helped or hurt. Make one change at a time and measure the impact.

Reputation recovery is a patience game. The damage happens fast (hours to days), but recovery is slow (weeks to months). The most important principle: be systematic, be patient, and let the data guide every decision. Shortcuts don’t work — providers have seen them all.