Measuring mailbox health across five client domains without mixing reputation dashboards

When an agency runs five client sending domains, the tempting move is one mega-dashboard. Average the spam rates, color the week green, and miss that Client C is on fire while Client A is fine. Reputation is earned per domain. Mixing clients in one blended chart hides the only signal that matters for pause decisions.

Google Postmaster Tools lists domains separately and exposes spam rate, domain/IP reputation, authentication, and delivery errors per domain. The API v2 path even supports querying stats across domains, which is useful for ops automation. Useful is not the same as blended. Pull many domains if you want. Still decide and alert per domain, tagged to one client workspace.

The unit of health is the domain inside a client workspace

For each sending domain, track at least:

• Spam / complaint rate (Postmaster spam rate where volume qualifies)
• Domain reputation and IP reputation bands
• SPF / DKIM / DMARC pass rates and compliance status
• Bounce mix (hard address failures vs policy blocks)
• Seed-inbox placement samples for major providers
• Daily send volume vs the domain's warm/cap plan

Google notes that some Postmaster statistics only appear when a domain sends enough daily volume to Gmail users (commonly cited threshold around 50 users/day). Thin domains can look "empty" without being healthy. Pair Postmaster with seed tests and your own bounce taxonomy so quiet domains are not mistaken for clean domains.

Why blended agency dashboards lie

• One bad domain dilutes into four healthy ones and never trips a threshold.
• One huge domain dominates the average and makes smaller clients look worse than they are.
• Complaint spikes need a pause on that domain, not a pep talk about "agency-wide deliverability."
• Client reporting must show their domains only. Buyers will ask.

Keep a roll-up for staffing ("how many domains need eyes today"). Never use the roll-up as the pause authority.

A practical board layout for five domains

Build one row per domain:

Status values that ops can act on:

• Healthy — within thresholds, campaigns free to run under caps
• Watch — one metric drifting; tighten volume or content
• Pause domain — complaint, reputation, or policy-block threshold hit
• Warming — not yet eligible for full cold volume

Color the status cell, not the whole agency.

Alert rules that respect isolation

Set thresholds per domain, with client-specific overrides when history demands it:

• Spam rate rising toward provider danger zones → page the domain owner, consider pause
• Domain reputation drop High → Medium/Low → investigate before adding volume
• Auth fail rate up → freeze new campaigns on that domain until DNS is fixed
• Hard-bounce spike concentrated in one list → pause the segment, not necessarily sibling domains
• Seed placement collapse on one provider → slow that stream and check content/auth

Wire alerts with client_id and domain in the title. An alert that only says "spam rate high" is how the wrong workspace gets paused.

Operating rhythm across many clients

Daily (weekday): scan the board for Watch/Pause, clear overnight Postmaster and bounce anomalies, confirm no cross-client filter mistakes.

Twice weekly: seed-inbox pass on active domains, compare to last week for the same domain only.

Weekly client note: send each client their domain rows, not the agency mosaic. Include actions taken.

On domain change: new domain starts in Warming with its own row. Do not inherit the previous domain's green status.

Tooling without mixing identity

Whether you use Postmaster UI, the Postmaster API batch query, Microsoft SNDS (where applicable), seed tools, or an aggregator, map every metric series to:

• client workspace id
• sending domain
• mailbox ids
• campaign ids (for content-scoped issues)

Aggregators that show "all domains" are fine for navigation. Your pause policy must still key off the domain row. Client-isolated workspaces in GTM exist for the same reason: Client A's reputation events should not teach you to throttle Client B.

What "good" looks like after a month

You can answer, without opening five random tabs:

• Which of the five domains are Healthy vs Watch vs Pause today
• Which client owns each domain
• What action was taken the last time a domain left Healthy
• Whether any dashboard view ever averages spam rates across clients (it should not)

Mailbox health at agency scale is a portfolio of domain-level truths. Measure many domains. Decide one domain at a time.

Sources: Google Postmaster Tools dashboards, Postmaster Tools API v2 migration, Google email sender guidelines, InboxEagle Postmaster monitoring, GTM product knowledge.