Corvane
How Corvane keeps message delivery visible during peak sending hours
Corvane is an SMS and email marketing platform used by thousands of retail brands. Its APIs send billions of messages a year, with most volume landing in a few short peak windows.
Industry
Marketing Technology
Location
New York
Employees
1000+
Identity Provider
Okta
Workloads
Messaging APIs, Carrier Gateways, Webhooks
Challenge
Corvane's delivery pipeline depends on dozens of internal APIs and carrier gateways. Slowdowns during peak campaigns were often reported by customers before the on-call team saw them.
Solution
The platform team put every public API and gateway behind Navy HTTP and keyword monitors running every 30 seconds from 18 regions, with Slack and PagerDuty alerts and a public status page for brands.
Results
Result 01
Median time to detect a delivery slowdown dropped from 11 minutes to under 1 minute
Result 02
Support tickets during incidents fell by 64% after the public status page launched
Result 03
Over 3,200 brands subscribed to status updates by email in the first quarter
Story
Peak hours without surprises at Corvane
A pipeline that cannot slow down
Corvane sends marketing messages for retail brands, and most of that volume lands in a few hours around launches and sales. A gateway that adds two seconds of latency at the wrong time can delay millions of messages. The team needed to know about it before brands did.
Monitoring every hop
The platform team added Navy monitors to every public API, every carrier gateway and the webhook endpoints brands rely on. Keyword monitors check that responses contain a valid delivery receipt, not just a 200 status. Because an outage is confirmed from 3 regions before an alert goes out, the on-call engineer only gets paged when something is actually wrong.
A status page brands trust
Corvane now runs a public status page on its own domain. When an incident starts, the timeline is created automatically and updates go out to subscribers by email. Brands check the page first, which has cut the volume of support tickets during incidents by almost two thirds.
Today every new service at Corvane ships with Navy monitors defined in the same pull request. The team reviews response time charts in its weekly reliability meeting and writes a postmortem in Navy for every customer-facing incident.
