SkyAlert built its own earthquake sensor network after a false alarm

SkyAlert launched in 2011 using Mexico's public Seismic Alert System (SAS). A false alarm confused users and exposed the risk of building a safety product on a system the company did not control. The 2017 earthquake made the failure visible at scale: SkyAlert's warning arrived about 12 seconds before the city alarm.

The turning decision was to build an independent network instead of continuing to depend on SAS. SkyAlert installed more than 120 internet-connected sensors in schools, hotels, and other buildings, covering about 80% of Mexico's seismic areas, and used its own alert logic, loud notifications, and configurable thresholds.

The decision converted reliability into adoption and revenue. The app had just over 100,000 active users before the 2017 quake and more than 6 million afterward; about 1 million paid for subscriptions. SkyAlert also sold alarm systems to buildings, which generated about 60% of its income.

Its alert arrived about 12 seconds before the public alarm in 2017; active users rose from over 100,000 to more than 6 million, including about 1 million subscribers.

The public system had produced false alarms, neighborhood outages, and failures to sound during earthquakes.

Owning sensors let SkyAlert control detection, coverage, and notification behavior rather than inherit SAS's gaps.

The independent network could serve users outside Mexico City, where a city loudspeaker system could not.

When a public dependency fails at the moment trust matters, own the critical sensing layer—even if the harder infrastructure becomes the business.

SkyAlert became Mexico's only legally permitted earthquake alert app and reached more than 6 million users. Its network still had limits: it required internet connectivity, and neither SkyAlert nor SAS could provide early warning for micro-earthquakes originating inside Mexico City. Users also reported missed or overly broad alerts, so the company continued to improve detection and follow-up messaging.