Email metrics
The Metrics page shows how your workspace's email is doing: how much of it reached the inbox, how recipients engaged, and whether your bounce or complaint rates are drifting toward trouble. Every number aggregates everything the workspace sends.
Open it at Email → Metrics in the dashboard. It has two tabs: Outbound, for what you send, which is the one that opens first, and Inbound, for what your workspace receives. For the outcome of a single message, whether this address bounced, opened, or clicked, use the email log. To pull the same aggregates from code, use the stats API.
Read your metrics
Everything in this section is the Outbound tab. The range picker offers Last 24 hours, Last 7 days, Last 30 days, and Last 90 days, and every number on the page follows it. Thirty days is the default.
Rates attribute by event time, not send time. A click that arrives today for a message sent last week counts toward today, and a bounce counts when it arrives rather than when the message went out. The figures come from a rolling aggregation and trail live traffic by a few seconds, so reloading picks up anything newer.
Delivery rate and bounce rate share a denominator, delivery attempts: the messages a receiving server accepted plus the ones it refused. A late bounce that retracts an earlier acceptance moves that recipient out of the delivered count and into the bounced one.
The four health cards
Each card shows a rate for the selected range, the count behind it, the change against the previous equal-length period, and a Healthy or Risk label.
- Delivery rate: deliveries that stuck, over delivery attempts. Healthy above 95%. Below that, something is refusing your mail, and the bounce breakdown says what: dead addresses, an authentication gap, or receiving servers turning your sending IP away.
- Open rate: unique opens over delivered mail, one open per recipient. Mail privacy features fetch images automatically, and those fetches are left out of this rate; Apple Mail Privacy and open tracking covers why that still leaves opens a soft signal. Healthy above 35%.
- Bounce rate: every bounce, including late ones, over delivery attempts. The card shows your rate against a 5% limit and flips to Risk at 5% or above. Sustained high bounces are a list-hygiene problem and they cost you sending reputation.
- Complaint rate: spam reports over delivered mail. The limit is 0.1%, low because mailbox providers act on complaints quickly. Watch this one hardest.
A card reads Healthy up to its limit, so Healthy is not the same as good. The exact formula behind each rate is documented per field on the stats summary reference.
Delivery & engagement over time
The main chart plots Sent, Opened, and Bounced volume across the range, which is where a bad send, a list import gone wrong, or a campaign that landed well shows up. Buckets are hourly over 24 hours and daily over longer ranges.
Bounce rate and its causes
The bounce card breaks your bounces down by cause, so a rising rate points at a fix:
- Hard: a permanent failure, such as an invalid address or a domain that does not exist. These suppress the address automatically.
- Soft: a transient failure, such as a full mailbox or a server that was briefly unavailable. We retry these.
- Block: the receiving server refused your sending IP.
- Admin: an administrative refusal, such as relaying denied or a blocklisted domain.
- Undetermined: the server's response was too ambiguous to classify.
- Out-of-band: a late bounce that arrived after the receiving server had already accepted the message. It retracts that acceptance, so the recipient becomes bounced.
These are the same bounce_type values the event stream uses, and the bounce classification table maps them to the finer-grained codes.
Complaint rate
The complaint card charts your spam-complaint rate across the range. Since the limit is so low, watch the shape as much as the number: a steady climb well under 0.1% is an early warning that a segment or a campaign is annoying people.
Latency
The Latency table splits time-to-inbox into Processing (the time we spend accepting and queueing the message), Delivery (time at the receiving server), and Total, each at p50, p95, and p99, with a sparkline per stage. Read p95 and p99 for the slow tail: a fast median with a slow p99 usually means one provider is deferring your mail.
Breakdowns
The Breakdowns panel slices the same delivery and engagement numbers one dimension at a time, so you can compare senders and isolate a problem to its source. By domain, By IP, and By tag are always there, and By broadcast appears once you have sent a broadcast. Tags are the most flexible cut: tag a campaign, template, or experiment variant at send time and compare them directly.
Each tab lists sent, delivered, bounce, open, and complaint figures per row, plus a Status. The IP tab adds the pool each IP belongs to. Rows are ranked, filterable by name, and capped, so the panel says "Top 50 of 120 sending domains" when a range holds more senders than fit.
A row's Status is derived from its own delivery, bounce, and complaint rates, and the thresholds are stricter than the four cards use:
| Status | When a row gets it |
|---|---|
| Healthy | Complaints at or below 0.06%, bounces at or below 0.4%, delivery at or above 99% |
| Watching | One of those crossed, but complaints still at or below 0.1%, bounces at or below 0.6%, delivery at or above 98.4% |
| Throttled | Complaints above 0.1%, or bounces above 0.6%, or delivery below 98.4% |
Because the two scales differ, a single sender can read Throttled while all four cards read Healthy: the workspace-wide average absorbs one bad domain that the per-row view exposes. Treat a Throttled row as the sender to investigate first.
Inbound
The Inbound tab covers what your workspace receives, once you have set up a forward address or your own domain for receiving. It shows one number, Messages received for the range you picked, and a chart of that volume over time.
A message you receive has no delivery, bounce, or complaint outcome the way a message you send does, so this count stands on its own instead of joining the health cards above. Until you set up receiving, the tab reads zero.
Sandbox traffic counts
Sends to sandbox addresses on messagebird.dev move these numbers, including the bounce and complaint rates, because they run through the same pipeline as real mail. They never leave our systems, so they cannot touch your real sending reputation, but a heavy test run does change what a range containing it reports.
Advanced analytics
The Metrics page reads your own sending and receiving. Two more products see past it: Inbox Tracker monitors where your mail actually lands across mailbox providers, and Competitive Tracker benchmarks your program against other senders. Both are separate products; book a demo to see either one.
Next steps
- Stats API: pull every number on this page from code
- Open and click tracking: the instrumentation behind the engagement numbers
- Email log: the per-message view behind the aggregates
- Events and webhooks: the per-recipient event stream both are built from