Reputation is plural
Mailbox providers build their own views of domains, IPs, sending streams, and recipient interactions. Gmail's view is not Outlook's view, and neither is reducible to a public number you can carry between providers. Commercial sender scores can be useful proxies, but they are not the score a mailbox provider necessarily uses to place your next campaign.
Think in layers: the visible From domain, DKIM signing domain, return-path domain, tracking domains, sending IP, platform account, and sometimes the individual mailbox all accumulate evidence. A provider may treat these layers differently. That is why a reputation investigation needs several sources rather than one green badge.
Evidence you can observe
Use Google Postmaster Tools for Gmail aggregate data when volume is sufficient, Microsoft SNDS for authorized Outlook.com IP data, complaint feedback loops, DMARC reports, blocklist checks, provider delivery errors, hard-bounce trends, and repeated inbox placement tests. Opens and clicks can help with audience behavior, but privacy features and measurement limitations make them imperfect reputation diagnostics.
Distinguish fact from inference. A blocklist entry for the actual sending IP is a fact. A drop in Gmail domain-reputation label is a fact within Google's system. 'Our subject line damaged reputation' is an inference until a controlled comparison and broader evidence support it.
What usually damages reputation
Recipients marking mail as spam is one of the clearest negative signals. Invalid or old data can create bounces and trap risk. Sudden volume, inconsistent cadence, a compromised account, unclear consent, deceptive content, and mixing risky marketing traffic with critical transactional mail can all contribute. Broken authentication makes identity less trustworthy but is not the only cause.
The common thread is unwanted or unsafe behavior, not aesthetic imperfection. A beautiful template sent to people who do not expect it is still a bad reputation strategy. Conversely, a plain operational message to a genuinely engaged audience can perform well without elaborate optimization.
A recovery plan
Contain compromise and stop harmful sources first. Suppress complaints, hard bounces, unsubscribes, and addresses without a defensible origin. Restore SPF, DKIM, and DMARC alignment on the production path. Reduce volume to the recipients most likely to expect the message, then increase only when provider responses and placement remain stable.
Do not rotate domains to outrun history. If the same acquisition and sending behavior continues, the new identity will accumulate the same evidence. Recovery is usually the result of consistent wanted mail over time, not one delisting form or warmup sequence.
Monitor movement, not mythology
Create a baseline by provider, sender, and message stream. Run comparable placement tests before high-value campaigns and after infrastructure, audience, or tracking-domain changes. Alert on meaningful movement and preserve individual reports so a trend can be traced back to the message and configuration that produced it.
A stable average can hide one provider getting worse while another improves. Review provider and segment breakdowns, not just an overall score. When reputation evidence and placement disagree, keep investigating; they measure related but different parts of the system.
Questions readers ask
What is a good sender reputation score?
There is no universal provider-independent score. Use provider dashboards, complaint and bounce evidence, blocklists, delivery errors, and placement trends together.
Can I reset sender reputation by changing domains?
A new domain has no positive history and can inherit the same bad behavior. Fix the source of complaints, invalid data, compromise, or volume instability first.
How long does reputation recovery take?
There is no fixed period. It depends on the severity, provider, sending volume, recipient response, and whether the underlying behavior has genuinely changed.