AI is freeing up CS hours (maybe?). Almost nobody's decided on purpose where they should go.

Ask a room of CS leaders what they'd do with ten hours a week back per CSM, and almost everyone gives the same answer. More accounts.

Give any organization spare capacity and the reflex is the same: turn it into more of the current job, faster, because that's the only version of value finance already has a column for. Call it the throughput reflex.

Here's the harder question underneath it.

If AI genuinely frees up hours, there are four honest places that time can go: fewer CSMs, more accounts per CSM, deeper work on the accounts already assigned, or protected time to explore something nobody's assigned yet.

Two of those already have a metric. Guess which two win by default.

Cutting heads is the fastest way to "capture" the efficiency, and it's the one finance reaches for first if nobody argues for something else before the savings show up on a spreadsheet. More accounts per CSM is the second-fastest, because headcount-per-book is a number everyone already tracks. Neither requires anyone to defend a new metric.

Deepening and exploring are different animals.

Every CSM already carries a private list of things that would move an account forward and never make the top of it: the account still single-threaded, the outcome nobody's revalidated in months, the use case everyone assumes exists and nobody's confirmed. Urgent is the only filter most teams actually run on, and none of that list is ever urgent, so none of it ever happens.

A recent thread in Uncommon turned this from a nice idea into an actual audit.

Have your team pull up last week's calendar and sort every recurring task into one of five buckets.

No hedging, no "it depends." Run it honestly and somewhere between forty and fifty percent of what a team currently protects as essential doesn't survive as "keep." It won't stay honest on its own. Someone has to hold the line in that room, or nearly everything quietly lands back in keep.

“Start” is the one nobody's built a column for. That's where the real customer work lives: the account succeeding without anyone knowing why, the outcome nobody's checked on in months, the relationship still stuck on one contact. None of that shows up as more accounts per CSM. It shows up, eventually, in whether the customer actually gets the outcome they signed up for.

So decide what goes there based on what actually moves the customer forward, not on what's easiest to defend in a conversation with finance.

Run the five-bucket sort with your own team, out loud, with the people doing the work, and judge what survives by customer outcome, not by how efficient CS looks on paper.

What's sitting in your team's start bucket right now, the customer work with no defender and no deadline?

Come debate it in Uncommon.