Why contract review backlogs appear overnight

Your contract queue looks fine for months. Same headcount, same rough volume, same average time per file. Then, over two or three weeks, turnaround doubles and nobody can point to what changed. The team wasn't slacking off in July and didn't suddenly get slow in August. What changed is a number nobody was watching: how close the team was running to full capacity the whole time.

Queues don't back up in a straight line. They sit calm for a long stretch while a team absorbs more work, then blow up all at once once the team crosses a specific point, a pattern every contract review lead has lived through without ever putting math to it.

The part of the queue nobody sees coming

Picture a single reviewer as a cashier working through a line. At 50% utilization, basic queueing math says a new file waits, on average, about as long as one file takes to review before anyone gets to it. Push that reviewer to 75% utilization and the wait roughly triples, to about three files' worth of time. Push it to 90% and the wait is around nine files' worth, off a change in busyness that would look minor on any staffing spreadsheet. A rule of thumb from queueing theory holds that wait times go up quickly once utilization passes 80%, and that the same principle applies whether the server being watched is a computer or a person.

That curve is the reason a contract queue can look under control for a quarter and then fall apart in three weeks with no obvious trigger. The team was already running close to the edge, and the edge is where small changes in arrivals stop behaving like small changes.

Utilization, not headcount, is the number to watch

Plenty of teams track turnaround, and some now split it the way we've written about before: how long a file waits before anyone opens it against how long the review itself takes once someone does. Both of those matter, but they're both downstream of one input that almost nobody puts on a dashboard: what share of a reviewer's available hours is already spoken for by the files sitting in front of them.

You can get a rough version of that number without new software. Add up the hours your team spent actually reviewing files last month. Divide by the hours available to do that work, after meetings, onboarding, and everything else that isn't review. That ratio is your utilization. If it's sitting near 60%, a bad week barely dents your queue. If it's sitting near 85%, the same bad week turns into a backlog that takes a month to work off, because the team no longer has slack to absorb the next surge while still clearing the last one.

This is also why counting reviewers rarely tells you what you think it does. Two teams with identical headcount and identical average review times can have completely different queues, because one of them is running at 65% utilization and the other at 88%. The second team has roughly the same people as the first. What it's missing is slack, the room a queueing system needs to absorb demand that never arrives at a steady pace, and contract intake never does.

Where an AI first pass actually helps, and where teams waste it

An AI contract review workflow can cut the hours a reviewer spends per file, checking a document against your playbook before a person opens it and shortening the read that follows. That's a real reduction in the denominator of your utilization math. What it won't do is fix a queue by itself. If the hours saved just get filled with more intake at the same crowded utilization, the queue behaves exactly like before, because the ratio never moved. The benefit only shows up if the saved time actually lowers utilization, either by holding volume steady while capacity effectively grows, or by deliberately capping how much new work gets loaded onto a team running near the edge.

Software can shorten a review. Deciding how much new volume to accept in exchange for that time back is a separate call, and it stays with whoever owns the queue, made on purpose instead of by default.

Hiring isn't the lever most teams have

Rather than replacing lawyers, 63% of CLOs expect headcount to remain stable, with departments focusing on upskilling existing teams to support higher-value work through AI-driven efficiency, according to ACC's 2026 Chief Legal Officer Survey. For most teams, adding reviewers to buy back utilization isn't the plan for this year. Which means the number worth protecting is the one you already control: how much work you let stack up against the capacity you actually have, tracked the same way you'd track queue time and touch time.

Pull last month's review hours, divide by hours available, and look at where you're sitting. If it's creeping past 80%, that's the warning worth acting on, well before the queue itself tells you.

Try goHeather free if you want to see what an AI first pass does to the hours a file actually takes, so you can judge for yourself whether that time is buying you slack or just getting absorbed by more volume.

This is legal information, not legal advice; consult a lawyer for legal advice.