Backlog triage: digging out without burning the team

The oldest ticket in your queue is eleven days old, and you only know that because it just got reopened for the third time — the customer now CC'ing someone with "Director" in the signature, the thread long enough that nobody remembers the original problem. That is what a backlog actually is: not a tidy stack of work waiting its turn, but a pile that gets heavier and angrier the longer you look away.

Backlog is a loan, not a queue

The instinct is to treat a backlog as a volume problem — this many tickets behind, that many people, divide and grind. But backlogs don't sit still. Every day a ticket ages, it costs more to resolve: the customer loses the context, chases for an update, opens a duplicate, escalates. One neglected ticket quietly manufactures its own follow-ups — the chase email, the duplicate, the angry reply. You are not behind on a queue. You are paying interest on a loan.

And the interest rate is rising.

The tolerance for a ticket to sit is shrinking, so an aging pile doesn't merely stay bad — it degrades relative to what the customer now expects.

The humane math

Here is the trap. Faced with a mountain, most managers reach for the obvious lever: push utilisation. Cancel the 1:1s, kill the coaching hour, run everyone flat out until the number comes down. It works for about a week.

Occupancy — the share of logged-in time an agent spends actually handling contacts — is the number that governs this. Across 160,000-plus staffing calculations, the average maximum that professionals target is surprisingly modest:

And there is a reason they don't set it higher.

Push a team past that 90% line to clear a backlog and you won't clear a backlog — you'll trade a ticket debt for an attrition debt, and the second one is far more expensive. The people who actually know your product start updating their CVs, and the queue gets deeper because now you're short-staffed on top of everything else.

A method that digs out without digging a hole

Freeze and count by age, not volume. Stop reporting "412 open." Report the age distribution: how many over three days, over a week, over a month. The old tail is where the interest compounds, and it's usually a small, ugly slice of the total.

Work the tail first. Counter-intuitively, don't start with easy new tickets to make the number drop fast. Start with the oldest and the most-reopened, because those are the ones spawning duplicates and eroding trust. Killing one two-week-old reopener removes more future work than a handful of fresh closes.

Swarm the hard ones. Aging tickets are usually aging because they're stuck — they need a second brain, not a third escalation tier. When Ping Identity replaced tiered escalation with intelligent swarming, the backlog fell fast:

On the genuinely gnarly cases the effect was larger still —

Pulling the right people onto a stuck ticket at once beats passing it down a chain and waiting a day for each handoff.

Time-box the dig, then protect the floor. Run the cleanup as a defined sprint with an end date, not an open-ended death march. Ring-fence a couple of agents for the backlog while the rest hold the line on new volume, and rotate people out so nobody lives in the tail for two weeks straight.

What "done" looks like

Digging out isn't hitting zero. It's getting the age distribution back under control — no ticket older than your SLA quietly compounding in a corner — while occupancy stays somewhere a human can sustain. If you cleared the backlog but three good agents are interviewing elsewhere, you didn't win. You refinanced.

So track two numbers side by side through any triage push: the age of your oldest open ticket, and your occupancy. If the first is falling while the second climbs past the danger line, you aren't solving the backlog — you're moving it off the queue and onto your people.

Check your own occupancy and response-time targets against the benchmarks, and if the line between occupancy, utilisation and productivity still feels fuzzy, the glossary pins the definitions down.