KPI Rules allow business processes to be triggered automatically when KPIs change, enabling the complex and subtle behaviour of IoT device estates to be managed effectively. Read this first for a brief refresher on KPIs.
Many IoT platforms and services offer the capability to trigger an action when a device gets into a specific state. The action might be to notify someone, or to raise a ticket in a ticketing system (via an API call). For example, if a battery goes flat then schedule a service person to replace it. Or if a fridge gets too hot then send an urgent text message to the owner.
These simple rules can be useful, but usually:
a) The rules don’t have any concept of time passing, and therefore of previous state - they just act instantaneously. So they can't easily detect change, and they can't easily measure how long a condition has been true
b) They analyse each device individually - with no concept of how that device plays its part in the larger device estate. So they can't act on performance metrics at a higher level, such as "per-site" or "per-customer" or even those measured across the whole device estate.
DevicePilot's new KPI Rules do have these capabilities, which is useful because IoT device behaviour is often a little more subtle than the examples above. For example, a customer was using DevicePilot for fridge monitoring and recently found that the fridge’s regular defrost cycle (which melts ice by raising the condenser temperature a lot, but only for a short time) was triggering notifications erroneously, as soon as the sensor temperature went above the threshold, which is a limitation of simple instantaneous device-state rules.
How to use KPI Rules
DevicePilot KPI Rules are triggered from changes in KPIs, which are measurements over time. To continue the above example, we can define a KPI to measure “the average temperature over 15 minutes of every fridge” and then we can define a KPI Rule saying “if that average temperature is 5 degrees above normal for more than 5% of the time [in any 15 minutes], then send a notification”. Now an alert won’t be sent for defrosting (because the condition isn’t true for long-enough), but will be sent for a compressor failure, power failure, or when the door is left open - which is what we want.
DevicePilot users will appreciate that this is just the beginning of the power of KPI Rules, because KPIs can be grouped not just by time but also by any other property, too. So a KPI can measure the overall behaviour of groups of devices, not just of each device in isolation. For example, if a KPI measures “average availability of EV chargers on each site” then a KPI Rule can take action when less than 10% of chargers are available on any particular site. The notification can of course then tell you which site, and offer a hyperlink to take you straight into seeing the situation first-hand in DevicePilot.
Another example is a business delivering a consumables to many sites (e.g. coffee). The business can define a KPI which measures “sites likely to run out today” and then schedule a KPI Rule to automatically identify all such sites daily and log them as today’s work schedule for the service team, via a service such as Jungleworks' Tookan which integrates seamlessly with DevicePilot to achieve this.
This can even be extended to look at all the devices in a device estate, to spot major infrastructural problems. For example if a KPI defines the “uptime of the entire device estate over the last 15 minutes” and it typically runs at 95% +/- 2%, then a rule which triggers if it falls below 90% will quickly warn of the start of any major incident - whether that is caused by a cloud ingestion hiccup, a power outage across a large part of the device estate, or some mass user behaviour-change, perhaps in response to an emergency, or even something you've never thought of and so won't be explicitly watching for (a so-called "Black Swan" event).
Remember that you can trigger all sorts of business processes with Rules, including notifying your team directly in Slack.
Most businesses using IoT live or die by the quality of service that they deliver. KPI Rules enable IoT businesses to measure the amount and quality of their service delivery directly, respond quickly to changes and thereby increase up-time and customer satisfaction - without the cost of a vast operations team to manually monitoring everything.