The commit, the term, what they've spent so far and what they spend now. See whether they'll use it, fall short, or run over.
Take what they've spent, add today's monthly spend for every month that's left, and compare it with the commit. That's where the account lands if nothing changes.
On a consumption plan, a commit nobody uses doesn't retire much quota. It also makes the renewal harder, because the customer remembers paying for capacity they never ran.
The number that matters is the monthly spend it takes from here. If it's well above today's pace, you need new workloads with dates attached. A more hopeful forecast won't get you there.
If they're running ahead of the commit, that's good news, but don't sit on it. Overage is spend nobody forecast, so start talking about the next commit before the renewal forces the conversation.
I spent six years leading federal partner sales teams at AWS, after plenty of years carrying a number myself. QuotaBird is mostly stuff I wish we'd had back then. Most of it's free. If the problem's messier than a little tool can handle, we can talk it through.