Google Postmaster Tools: read your Gmail reputation, and what an agent checks

What Google Postmaster Tools is

Google Postmaster Tools is a free Google service that shows your compliance status, spam rate and authentication results as Gmail sees them. The older v1 domain and IP reputation dashboards are being retired, so the spam rate is now the figure to read. Once a person reads that spam rate in Postmaster, an AI agent can put it next to the send data and stop the next campaign before it crosses Gmail's line.

An agent sending over MCP runs on a schedule, retries, and keeps sending unless something tells it to stop. Postmaster is one of the few places Gmail tells you, in numbers, that it is time to stop.

What changes when an AI agent watches Google Postmaster

Postmaster data is not live. It arrives with a delay, and a domain that sends little mail to Gmail users shows nothing at all. An agent should read it as one signal next to the bounces and complaints from its own sends, not as a live gauge. When the spam rate climbs toward 0.3%, the right move is to hold the next send and find the cause, never to raise volume to dilute the rate.

The panels to read have changed too. With the v1 domain and IP reputation dashboards on their way out, do not chase a Gmail domain reputation score. Read compliance status, spam rate and the delivery verdict in Postmaster Tools v2 instead.

Gmail sends no report for each complaint and computes its spam rate on mail delivered to the inbox, so a blended complaint rate across all providers understates the Gmail figure when Gmail is a big share of the list.

Postmaster stays a dashboard that a person reads. In the workflow on this page, the person reads the spam rate and compliance status there and gives the two figures to the agent. The agent, working through Claude, ChatGPT or Cursor over MCP, sets them next to the health of its own sends from nitro_get_insights and can hold a campaign before a dip turns into a placement problem.

What Google Postmaster Tools shows

What Postmaster does not show is the folder your mail landed in, so pair it with inbox placement testing, start from the inbox placement hub, and read it alongside the wider email deliverability picture rather than as a verdict on its own.

Postmaster figures are an input to your deliverability rate rather than a measure of it, which is why the two often disagree, and what email deliverability actually is covers the four layers Postmaster only partly sees.

George Hartley reads Postmaster through a specific lens: for Gmail, domain reputation matters more than IP reputation, and IP reputation has a 120-day memory. Dedicated IPs only pay off past 1 million emails per month, because below that a shared pool usually beats a cold dedicated IP. The same logic puts marketing and transactional mail on separate subdomains by default, so each stream builds its own record at Gmail.

One practical step when a Gmail message lands in spam: open Show original and copy the raw email. The raw headers carry the authentication verdicts Postmaster summarizes, and an inbox test shows where the next one lands.

The spam rate is the Postmaster number that matters

The spam rate. Google's email sender guidelines ask senders to keep it below 0.10% and never let it reach 0.30%, and Postmaster is where you watch it. Cross that line and Gmail throttles or blocks you, so the goal is headroom, not just staying under. Pair it with a sender score for a cross-provider view.

George Hartley's Email Marketing Bible sets a stop rule for this number. While the complaint rate is at or above 0.1% or Postmaster shows a high spam rate, broad sends stop, and only recovery sends to people who clicked in the last 30 days continue. A not-compliant verdict in Postmaster blocks every send until it is fixed.

Worked example: an agent checks Gmail reputation before a weekly newsletter

The prompt: "Check our Gmail health before Thursday's newsletter goes out, and tell me if it should wait."

Before. Left unguided, an agent schedules the newsletter at the usual time for the whole list, even if the last two issues drew complaints, because nothing it reads tells it to stop. If opens look soft, it may widen the audience to make up the numbers, which pushes the Gmail spam rate further toward the line.

After. With the Nitrosend MCP tools, the agent works through a fixed sequence:

1. nitro_get_status confirms the account is ready and names the sender identity the newsletter will go out under.
2. nitro_get_insights pulls the trend for the last few issues, with bounce and complaint movement and benchmarks to compare it against.
3. nitro_manage_domains runs a diagnose on the newsletter's sending domain to confirm SPF, DKIM and DMARC alignment, the same results Postmaster summarizes for Gmail.
4. The person pastes this week's Postmaster spam rate and compliance status into the chat, and the agent sets them next to the send data.
5. nitro_compose_campaign holds this week's issue as a broadcast draft, which never delivers on its own.
6. nitro_review_delivery checks the draft issue for spam signals and preflight problems before anyone commits to sending it. It approves nothing.
7. nitro_send_test_message mails one copy of the issue to the account's test recipients, so a person reads it the way a subscriber would.
8. The agent recommends one of three paths: send to the full list, send only to recent clickers, or hold. A person makes the call and approves the campaign with nitro_control_delivery.
9. After the send, nitro_get_insights reports how the issue landed, which becomes next week's starting point.

Before, a rising Gmail spam rate shows up after the newsletter has made it worse. After, it decides whether the newsletter goes at all.

Where it goes wrong when an agent reads Postmaster

• No data read as a clean bill. A low-volume domain, such as a new domain, shows empty Postmaster charts, and an agent can report the blank as no problems. Empty means Gmail has not published data, so the agent should fall back on bounces and complaints from the sends themselves.
• IP reputation chased on a shared IP. On a shared pool the IP reputation belongs to every sender on it, and Gmail weighs the domain more heavily anyway. An agent that changes send times or batch sizes to move an IP number is working on the wrong lever.
• The root domain verified while mail is sent from a subdomain. An agent that sets up and checks the root domain, then sends the newsletter from a marketing subdomain, can report clean SPF and DKIM results for a domain the mail never used. The diagnose in nitro_manage_domains and the Postmaster figures both need to be read for the exact domain in the From address.
• A one-day spam-rate spike acted on without the trend. A single day can jump on a small Gmail sample. An agent that cuts the list or rewrites the schedule on one reading overreacts. Several days of trend, checked against complaints from the sends, should drive the decision.
• Volume raised to dilute the rate. Spam rate is a ratio, so an agent told to bring it down may try to send more mail to lower the number. Gmail counts the rate on inbox-delivered mail, and a sudden volume jump is a warning sign of its own. On the Free plan the allowance also caps this, since a new account starts at 100 recipients per rolling 24 hours and rises with sender standing up to 5,000, as the pricing page sets out.

How Nitrosend uses this signal

Nitrosend moves the checks that Postmaster reports after the fact to before the send. nitro_manage_domains adds, verifies and diagnoses sending domains, so SPF, DKIM and DMARC are in place before Gmail sees the first message. nitro_review_delivery runs spam analysis and preflight on a draft without sending it. Campaigns and flows stay drafts until a person approves them with nitro_control_delivery, and nitro_get_insights reports trends, benchmarks and recommendations after each send.

Splitting the streams helps here, since transactional email rarely generates complaints and should not inherit a marketing domain's spam rate. Sends made through the email API or the transactional email API go out from the same verified domains, so one set of DNS checks covers both.

FAQ

Go deeper