Lockally

      Use your API key in the examples

      Replaces lk_live_… in the code samples on this site so you can copy and run them. It is stored in this browser only and is never sent anywhere, including to Ask AI.

      It will be visible on the page. Use a test key if you are sharing your screen or taking screenshots.

      Sender reputation

      Your workspace has an automated reputation engine watching it. There is no human reading a dashboard, so the engine acts on its own, in two graduated steps, and tells you by email.

      What it measures

      A rolling 24 hour hard-bounce rate per workspace, over messages that have a decided outcome. Deferrals are not counted: a message still being retried has not failed yet.

      It also computes a complaint rate on the same window. That half is built and currently reads zero, because no feedback loop delivers complaints to us yet. See Complaint handling.

      What it does

      State Trigger Effect
      Warning Above 5% over the last hour, with enough volume An email. No enforcement. At most one per cooldown period.
      Throttled Above 8% over 24 hours, minimum 50 decided messages Your per-minute rate cap is cut. The original is saved.
      Suspended Tripping again while already throttled Outbound is halted.
      Restored Healthy again, with volume, while throttled The original rate cap comes back.

      The floor matters: below 50 decided messages in the window nothing trips, so a handful of bad addresses on a quiet workspace does not throttle you. It also means a small sender can sit above the threshold for a while before the engine has enough evidence to act.

      Why the thresholds are where they are

      Eight percent is high as an industry number and deliberately so: it is the point at which the platform stops you to protect everyone else sharing the sending IP, not the point at which your deliverability is healthy. A bounce rate above about 2% is already a problem for you long before it is a problem for the engine.

      Treat the warning at 5% as the real signal and the trip at 8% as the backstop.

      The shared-IP consequence

      Reputation accrues to the IP, and today that IP is shared by every workspace. So the engine is not only protecting you from yourself: it is protecting the other senders on the same address from you. That is why the response to a high bounce rate is to cut your sending rather than to warn and wait.

      It also watches for the inverse. If several workspaces trip at the same time, that points at the pool rather than at any one sender, and it is logged as a possible pool-level problem rather than treated as four independent faults.

      Getting back

      Stop sending to the list that caused it. The engine recovers on measured behaviour, not on a request, so continuing to send the same list keeps you throttled.

      Clean the list. Every hard bounce is already on your suppression list, which is the record of what went wrong. Export it and remove those addresses at source, or you will keep re-submitting them and the suppression list will keep absorbing them without your bounce rate improving.

      Let volume accumulate. Restoration needs healthy volume in the window, so a workspace that stops sending entirely stays throttled until it sends again.

      A suspension is not self-healing in the same way. Contact support.

      Reading your own standing

      curl -s https://api.lockally.com/v1/reputation \
        -H "Authorization: Bearer $LOCKALLY_API_KEY"
      {
        "state": "healthy",
        "bounce_rate": 0.012,
        "decided_count": 1840,
        "last_evaluated_at": "2026-10-20T18:05:11Z",
        "complaint_rate": null,
        "complaint_rate_note": "Not measured: no provider feedback loop currently delivers complaint reports, so this is unknown rather than zero.",
        "rate_cap_per_min": 120,
        "thresholds": {
          "bounce_rate_warn": 0.05,
          "bounce_rate_trip": 0.08,
          "min_decided_for_trip": 50,
          "throttled_rate_cap": 1,
          "window_hours": 24,
          "warn_window_hours": 1
        }
      }

      Scope send:read. state is healthy or throttled; when throttled, original_rate_cap is what your cap will return to.

      The rate and count are as of the engine’s last evaluation, not computed live, so last_evaluated_at tells you how fresh they are. A workspace the engine has never evaluated reports healthy with nulls, which is the normal state for one that has not sent enough to be decided on.

      complaint_rate is deliberately null rather than 0. Nothing measures it today, and a zero would read as “nobody complained”.

      For the underlying counts rather than the engine’s verdict, use GET /v1/messages/stats. See Deliverability monitoring.