Renew where the work changed.

The first question before an AI renewal is what result would justify renewing. Agree it before anyone pulls a number, and the date becomes a decision rather than a deadline.

Then ask four things of the tool: which workflows changed because of it, who owns each one, what measure moved, and what check shows the output is still good.

Usage figures from the provider are a fair place to start. Ask how they were produced: who was measured, against what baseline, and by whom. Then set them beside your own.

Know what it costs to run, too. KPMG found that organizations with full visibility into their AI operating costs are likelier to report established ROI: 15 percent, against 3 percent of those without it.

A renewal date is a term you can negotiate: a short extension, a smaller term, or an option. Renew where the work changed, and decide the rest on the evidence, with both sides priced.

Constructed example

At Orrin & Vale, 41 percent of seats are active weekly, and no Content draft yet runs through a written check. Their plan gives each AI investment a goal, an owner, and the result it must show by renewal.

So: before the renewal date, write down the result that would justify renewing and who will show it. Renew where the work changed, and set the date you decide to keep, fix, or stop the rest.

Sources checked October 2026