Metrics in the Age of the Smartwatch — Librato Blog

Metrics in the Age of the Smartwatch


By all indications, the smartwatch is about to go mainstream. Now that Pebble and Android Wear have nabbed the early adopter market, Apple Watch, with its array of fashion-forward designs, promises to be the wearable for everyone else. If your company provides an app, the smartwatch could be an important new distribution and branding channel. A host of players, including Uber, Instagram, CNN and eBay, have already announced Apple Watch apps, even though it’s still unclear exactly what those apps will do.

One thing is for certain: diving into the smartwatch app market now, if done right, could pay major dividends in the future. The hard part for developers is that smartwatch apps, even more than mobile apps before them, must be succinct, uber-responsive, and intelligent. User tolerance for latency and irrelevance is already low —on a smartwatch, patience will wear even thinner. If an app pops up on a wrist, and it isn’t useful immediately, there’s no incentive for the user to keep it.  

Knowing that the app must be fully functional in seconds puts app developers under the gun to figure out how and where to best allocate resources. Smartwatch apps rely on smartphones to do the heavy processing, but it’s up to developers to decide which computations happen in the watch, which ones happen on the smartphone and which end up the cloud. In some ways, it’s the biggest development challenge yet—maximizing utility on the tiniest available interface, with almost no room for error.

Developers must ensure that all the moving parts of a smartwatch app, from sensors to back-end SaaS, are orchestrated seamlessly at all times. The cycle in which an engineer observes an abnormal behavior, orients herself towards the solution and fixes it becomes compressed with smartwatch apps. If the app is too slow or doesn’t work as advertised, users will simply delete it. What they’ll retain is a memory of their frustrating experience with the brand.

The solution? More than the smartphone or laptop before it, the smartwatch issues a mandate for developers to instrument their apps during the development process. If you are already using a service like Librato, this is not a tremendous leap in best practices. It’s something we at Librato refer to as metrics-driven development, and the smartwatch is giving that practice urgent status.

With smartwatch development, there simply is no room for figuring it out retroactively—you’ve already lost when that happens. Monitoring must happen on a continuous basis, and that’s easiest to achieve when you integrate metrics as a standard part of the engineering process. Everyone on your team can consult the same source of truth, and problems are automatically surfaced in an intelligent way that enables fast, actionable responses.

Traditionally, monitoring services rely on agents on servers to communicate metric data over a network, but smartwatch devices add an additional layer of complexity as the agents typically cannot be installed on a device. Using simple http calls with our Rest API, Librato can circumvent this restriction, and give meaningful metrics to developers in real time.