Feed breakage alerts
MartechFlow tells you when a feed breaks: failed refreshes, held publishes, count drops, and high error rates, shown in-app and emailed to your admins.
The promise: we tell you when it breaks
You should not have to babysit your feeds. When something goes wrong, MartechFlow raises an alert on the affected feed so you see it without logging in to check. Alerts appear in-app as a banner on the feed and roll up to a dashboard list, and the actionable ones also email your owners and admins.
Alerts are de-duplicated per feed and per problem type while still open, so a feed that fails every hour does not spam you. The alert is refreshed in place, and you are emailed once per problem, not once per run.
The alert types
Each alert type targets a specific failure mode, with a severity of critical (something is broken now) or warning (something looks off and is worth a look).
- Refresh failed (critical): a feed refresh or an export build failed. The alert carries the error reason.
- Publish held (critical): a new build looked unsafe, so MartechFlow kept serving the last good version instead of publishing it. Reasons include a build dropping to 0 products, a large product-count loss, a missing required column, no products passing validation, or a market export missing its base currency.
- Count drop (warning): the product count fell compared to the previous build, but not enough to hold the publish. A heads-up to check whether a category went missing.
- High error rate (warning): a large share of products in the built feed have validation errors and may be disapproved by the channel.
- Export stale and Refresh missing: reserved alert types for an export not rebuilding or a scheduled refresh not running.
The publish guardrail
Held publishes are a feature, not a failure. MartechFlow protects your live channels from a bad build. If a rebuild collapses (0 products, or a drop past the safety threshold from a healthy baseline), it does not overwrite the working feed; it holds the new build and keeps the last good version live at your pull URL.
The guardrail only kicks in once a feed has a healthy baseline to protect, so a brand-new feed's first build is not held just for being small. When a later run looks healthy again, the held alert resolves automatically.
Where alerts show up and how to clear them
Open alerts appear as a banner on the feed and in the cross-feed alerts list, scoped to the projects you can access. Critical alerts use a red tone, warnings amber.
Most alerts resolve themselves once the next run is healthy. For example, a refresh-failed alert clears after a successful refresh, and a count-drop alert clears when the count recovers. You can also dismiss an alert manually from its banner if you have edit access to the feed.
Getting alerts by email and webhook
Critical and warning alerts that are actionable (refresh failed, publish held, high error rate, plus the reserved export-stale and refresh-missing types) email your org owners and admins the first time the problem is raised, when email delivery is configured.
Count-drop alerts are deliberately in-app only and not emailed, because they fire on any decrease and would otherwise email on normal catalog fluctuation.
If you set an alert webhook URL, MartechFlow also POSTs a small JSON payload (type, severity, message, feed id) to it when an alert is raised, so you can route alerts into Slack or your own systems. The webhook is best-effort and time-bounded, so a slow endpoint never stalls your feeds.