Growth stacks don't crash. They go quiet.

A pixel dies. A webhook gets switched off. An automation starts failing every run. No error, no page — the numbers just rot until someone asks why last month looks wrong. RunMend reads each tool's own API and makes the silence loud, the day it starts.

read-only keys · no snippet · key never stored

every stream normalized to one signal · 7-day window · read from each tool's own API — no snippet on your site

detectors

What it catches.

enabled-zero-events

On, but nothing flows

The flagship silent failure: a stream is switched on and delivering zero events — dead CAPI pixels, dormant mail streams.

high-failure-rate

Trying, mostly failing

Events are attempted but largely erroring. Warns at your threshold, escalates to critical past it.

stale-stream

Used to deliver, went quiet

An active stream whose last event is older than it should ever be.

blind-monitor

The watcher is off

A webhook or status-only monitor is disabled — events are silently not arriving anywhere.

no snippet

Nothing to install. Nothing that can itself go dead.

Monitoring that depends on a snippet in your page or a webhook receiver in your stack is one more thing that can fail silently. RunMend polls each tool's own API with a read-only key, so your data path is untouched.

And if we ever can't reach a tool, that's loud too: a collection failure is itself a finding, never a blank.

sendgridmeta-capitwilioread-onlyRUNMEND</>no snippet
how it works

Connect, watch, hear.

01

Connect

Paste one read-only key per tool. Keys are encrypted; we store counts, statuses, and timestamps — never message bodies or recipients.

02

Watch

The same detectors run on a schedule across every tool. Findings dedupe, escalate, and resolve — no re-alert spam.

03

Hear

Slack or email, the day a stream goes quiet — not the day the monthly report looks wrong.

connectors

Six tools, one signal.

SendGriddelivery stats
Postmarkmessage streams
Twiliomessages + error codes
Meta CAPIevents per pixel
n8nworkflow executions
Makescenario runs

Zapier is missing on purpose — it has no public run-history API. We don't monitor what we can't measure.

honesty rules

We never fabricate signal.

Nothing invented. We never invent an endpoint, a field, or a count. If we haven't confirmed it exists, it doesn't exist.

Honest zeros. If a tool doesn't expose a signal, we report zero — with a note explaining why. An honest zero beats a guessed number.

Dead ends, stated plainly. GTM won't tell anyone a single tag stopped firing. Volume baselines don't exist on day one. We say so instead of pretending.

and when we can't collect, that is a finding too — collector_error, never a blank

open source

The engine is open. Check our work.

Every detector and connector ships in runmend-engine. Run it yourself from cron and read the same findings — or let us host the scheduling, history, and alerting on the identical engine.

pricing

Flat. Never per event.

Priced by connectors and check frequency — never by how many events your tools happen to move. Every account starts free.

$0$79$199$499

See the plans

free audit

Find out what's already quiet.

Paste one read-only key. Ten seconds later you'll see every silent failure in that tool. The key is used in memory once and never stored.

RunMend — silent-failure monitoring for your growth stack