Six instances. One missed renewal and one quiet leak cost more than a year of the tool that stops both.

Finding 02: what each hour of the lapse cost

A few hours cost cents. A few days cost dollars. Nothing stops the cost at day 23.

The lapsed db.r6g.large pair costs $0.15 an hour and $3.60 a day. Autopilot costs $1,788 a year. The same lapse, left alone for a year, costs $1,314. A renewal missed by one afternoon is cheap. A renewal missed by one month is not.

This team lost $83 in 23 days. The same pair costs $1,314 over a year if nobody acts. Each extra lapsed term adds its own hourly premium.

Here is the same renewal with the free tools, at this fleet size. Reserver does not send messages yet. Email, Slack, and SMS notifications are on the roadmap. Renewal autopilot is also on the roadmap.

1. Any day of the term: on Watch, the Reservations list shows the days left on every term.
2. 90 and 7 days before the end date: the free Expiry Check makes a calendar file with reminders at these dates. You add the file to your own calendar.
3. 30 days before: the days-left figure turns amber. The term joins the dashboard count of reservations expiring in 30 days. The calendar file has a reminder here too.
4. End date: the on-demand rate starts, and the costs above start with it.

Finding 03: the instances that were never in the plan

The database got reserved. The rest of the fleet didn't.

Three m6g.large app-tier instances and one cache.r6g.large Redis node have run at on-demand rates since launch. That is 18 months so far. Nobody decided against reserving them. They never came up at a review, because the team held no review.

This needed no mistake. The team only lacked a process. Reserver's purchase recommendations list all four instances from the first scan, in their first month on the fleet. An age rule ("running longer than 30 days and uncovered, so reserve it") is on the roadmap. On a fleet this size, an early scan lets you act before the cost starts.