Deliverability monitoring
You can only act on what you can see. This page is what is readable, and honest about what is not.
1. Your own outcome counts
The primary signal, and the only one that is real time.
curl -s "https://api.lockally.com/v1/messages/stats" \
-H "Authorization: Bearer $LOCKALLY_API_KEY"
Counts and rates by outcome over a window. This is what exists instead of open and click tracking, and it is the number the reputation engine is also watching.
For detail rather than totals, GET /v1/messages filters by status, and
GET /v1/messages/{id}/events gives a per-recipient timeline.
Computing a bounce rate correctly
The mistake is dividing by everything sent. Some messages end as
unconfirmed, meaning the outcome was never
learned, and counting those as successes flatters the number while counting
them as failures invents a problem.
Divide by decided messages only, which is what the platform does:
bounce rate = bounced / (bounced + delivered)
Anything unconfirmed, deferred or still queued is excluded from both
sides. A rate above 2% is worth acting on; the platform does not act until 8%.
2. Google Postmaster Tools
The closest thing to reading a provider’s own opinion of you: domain reputation, IP reputation, spam rate, authentication results and delivery errors, for Gmail.
Verify your sending domain there. It is free, it is the single most useful external signal available, and the spam rate it reports is the closest proxy you have for a complaint rate given no feedback loop currently reaches us. See Complaint handling.
It is per domain and you own the verification, so this works regardless of what the platform exposes.
3. DMARC aggregate reports
Who is sending as your domain, and whether it authenticates. Ingested for domains hosted here and readable over the API:
DOMAIN=yourdomain.com
curl -s https://api.lockally.com/v1/domains/$DOMAIN/dmarc \
-H "Authorization: Bearer $LOCKALLY_API_KEY"
This is the signal that catches a misconfigured third-party sender using your domain, which is a reputation problem you would otherwise never see.
What is not measured
Opens and clicks. Deliberate: no pixel, no link rewriting. Engagement is yours to measure in your own application.
Complaint rate. The ingestion path is built and currently receives nothing, so your complaint rate reads zero because it is unmeasured, not because it is zero. Postmaster Tools’ spam rate is the substitute.
Inbox placement. Nothing tells you whether a delivered message landed in the inbox or in spam. “Delivered” means the receiving server accepted it, and nothing more.
Blocklist status. Not checked and not surfaced.
Inbox placement, blocklist status and complaints stay unmeasured, but your
reputation state is readable: GET /v1/reputation reports your state, the
engine’s last measured bounce rate and the thresholds it acts on. See
Sender reputation.
A workable routine
Weekly, read GET /v1/reputation for the engine’s own view, then glance
at Postmaster Tools for spam rate and domain reputation.
On every campaign, check outcomes a few hours after sending rather than at the end.
On any spike, pull GET /v1/messages?status=bounced and look at the
recipient domains. Concentrated at one provider means reputation or
authentication; spread across many means your list.
Act on the warning email, not the throttle. The platform warns at 5% over an hour before it throttles at 8% over a day, and that gap is the window in which fixing it is cheap.