Operations Metrics

Storage fills slowly and then all at once. What counts towards it, where to see what is using it, and what happens on the day a workspace is full — nothing stored is touched.

Half the people an operations system needs to reach do not read email at work. Why every alert email now has a twin in the bell, addressed to a person rather than to the room.

Push notifications are the most valuable channel an operations system has and the fastest to ruin. A three-question test, a starting manifest, and how to set push one alert at a time.

A drawing of the escalation ladder our live chat climbs when a customer is waiting and nobody has answered: who is alerted at once, at 45 seconds, at two minutes, and what the customer is told at three.

Live chat with AWRA support: offered only when someone is online, a clock instead of a spinner, files and screenshots, a bar that follows you around the app, and an email when nobody picks up.

Switching off an email should not mean switching off the alert behind it. Why email and the in-app notice are separate switches on the same row, and which combination saves a stock-out.

Mandatory notices carry a duty optional ones do not: restraint. Set days, never two in a row, monthly after the first month, and an end — the schedule we hold our own account notices to.

Shared cloud or a server in the cupboard used to be the whole choice. A managed dedicated instance gives you a system of your own and leaves the patching, backups and upgrades with the vendor.

What competes for a system on payroll day and at the close, what a dedicated queue and scheduler change, and what the Dedicated and Dedicated server tiers each do and do not share.

A dedicated instance runs the same product as a shared workspace and costs more. When it is money well spent, when it is not, and four questions that decide which side you are on.

Five things have to be true before a notification reaches a phone, and when one fails the other four look broken too. Both ends of the chain now name the first one instead of listing all five.

A workflow can be built with a wait in the middle — do this, pause three days, then do that. The pause works. Everything before it runs, the run is marked paused, and the log names the date it will continue. Nothing in the product ever comes back to it.