Skip to content
Flow metrics & forecasting

Aging Work in Progress: The Metric That Catches Stuck Tickets Early

By Sebastiaan Jansen · · 8 min read

Aging work in progress is the time each unfinished ticket has already spent in an active state. It's a snapshot of open work, sorted by how long each item has been moving or sitting since it started. Teams use it to catch stuck tickets while there's still time to unblock them, rather than finding the delay weeks later in a cycle-time average, after the ticket has closed.

A healthy-looking board can still hide an item that has been in progress for three weeks. Aging puts that item at the top.

What aging work in progress measures

Aging work in progress answers one question: of the tickets open right now, how old is each one?

The clock starts when a ticket enters the first active stage you care about, often "In progress," and stops at done. Most teams count working days. Whatever you pick, stick with one rule, including how you treat weekends.

Sort oldest first, talk about the top few, and write down why each one is still open. The list is not a score for the person whose name is on the ticket.

Two other clocks get confused with this one:

  • Cycle time is the age of a ticket that has already finished, so you only know it after the fact. Cycle time is what you want for baselines and forecasts. Aging is what you want for work still on the board.
  • Lead time usually starts earlier, when the request arrives, and includes the wait before anyone picks it up. Aging covers only the active part of that span, and only for tickets that are still open.

If Ready is a long column, decide whether age starts there or at In progress. Starting at Ready mixes queue wait with build time. Whichever you choose, write it next to the chart.

Why the average hides stuck work

A team can report a median cycle time of four days while a customer is still waiting. The median only describes finished tickets, so a stuck ticket shows up in it after it closes, often weeks too late.

That's the trap with stale tickets. A status meeting that only asks for the average will sound fine in the same week an integration ticket hits 18 working days open.

Blocked work is easy to spot when the flag stays current. But a ticket waiting on review, a design answer, or a flaky test often has no flag at all. Age catches it anyway.

Throughput tells you how much finished, and cycle time tells you how long it took. Reading throughput and cycle time together stops you celebrating a short average while old items pile up. Aging names the ones already past that usual duration.

How to calculate it

You need the date active work started, today's date, the current stage, and a working-day calendar if weekends don't count.

The formula

For each ticket still in an active stage:

Age = today minus the start of active work, in whatever unit you chose. Working days are the usual pick.

Sort descending, and that list is the report. If a dashboard insists on one number, use the age of the oldest ticket, or the count of tickets older than a threshold taken from your own history.

Then compare the 85th percentile of open age with the same percentile of finished cycle time. If p85 cycle time is 8 working days and five open tickets are past day 8, those five are late compared with your own history. Treat that as an illustration, not a benchmark.

A worked example

On Wednesday 7 October the team counts working days from entry into "In progress." Six tickets are active:

Ticket Entered in progress Working days open Stage now Note on the ticket
PAY-41 Mon 15 Sep 16 In review Reviewer out since last week
API-12 Mon 22 Sep 11 In progress Waiting on a partner sandbox
WEB-7 Wed 30 Sep 5 In progress Building
WEB-9 Fri 2 Oct 3 In review Review requested Monday
MOB-3 Mon 5 Oct 2 In progress Building
WEB-11 Wed 7 Oct 0 In progress Started today

Recent finished work has a p50 of 4 working days and a p85 of 8. Anything older than p85 gets a comment the same day. Anything older than twice p85 gets a decision: finish it, split it, or send it back.

PAY-41 and API-12 cross the line. WEB-7, at 5 days, doesn't, so the standup covers two tickets. Someone else reviews PAY-41 today, or it goes back to Ready. API-12 stays visible as blocked work with a chase date, and sits outside the active WIP limit if you have a blocked state. Set your lines from the last 20 to 50 finished tickets of the same type.

Aging, flow, and blocked work

Aging is a point-in-time cut, so it helps to read it alongside a few related views.

A cumulative flow diagram shows how many tickets sit in each stage. A widening band means work is arriving faster than it leaves, and aging tells you which tickets make up that bulge. Check both once a week.

WIP limits cap how many tickets are in progress, not how old they are. If you're at the limit and every ticket is under three days, the team is flowing. If you're at the limit and half the tickets are past p85, it's stuck. Finish or unblock the oldest before you pull new work.

Blocked work needs its own state. If you only add a label and the ticket stays in "In progress," its age keeps climbing while nobody touches it, and WIP looks full. A blocked column, or a flag you can filter on, keeps parked work out of active age. Someone still has to review that column, or it turns into a second backlog.

Monte Carlo forecasting runs history forward to estimate a finish date. If you leave out tickets that have already burned two weeks, the forecast describes a cleaner board than the one you have. Check them before you publish a date. Any date other teams will plan around should come from finished history.

A weekly review that surfaces stale tickets

One 25-minute review a week is enough. Between reviews, standups only mention tickets that are already flagged.

  1. List active tickets by working-day age, oldest first. Leave out done items and the backlog, unless you track time in backlog separately.
  2. Mark p50 and p85 cycle time for that ticket type, and flag anything older than p85.
  3. Write one sentence for each flagged ticket: waiting on a person, waiting on a decision, too big, or actually moving. "Still working on it" doesn't count.
  4. Change something about the ticket this week: review it, split it, chase the blocker with a date, or move it back so WIP matches what people are really focused on.
  5. Give every blocked item an owner and a chase date. A flag with no date is just a stale ticket with a sticker on it.

Keep ticket types separate, because a bug and a large feature shouldn't share one p85. If the same ticket shows up three reviews in a row, either the slice is wrong or nobody owns the blocker.

What to do when a ticket is already old

Age tells you where to look. The one-sentence reason tells you what to do about it.

Waiting on review. Cap the wait based on your p50, and let older reviews jump the queue. If the reviewer is away, reassign the review or send the ticket back to Ready.

Waiting on an answer. Name the question, who you're asking, and when you'll ask again. If the answer is days away, pull the ticket out of active WIP.

Too big. Ship the smallest slice and return the rest. Only reset age if the remainder counts as new work.

Actually moving. Say what will be true by the next review. A specific comment is fine. A blank one isn't.

Blocked outside the team. Add an owner, a chase date, and a stage that doesn't pretend anyone is coding. If the wait outlasts the reason the ticket exists, cancel it.

The goal for next week is fewer tickets past your p85. Closing easy items to pull the average down doesn't count.

Where the metric misleads

Moving a ticket to the backlog and pulling it back the next day hides the wait. Keep the original start date through that round trip.

Don't mix calendar days and working days, and don't start some clocks at creation and others at "In progress." Write the rule down once, next to your cycle-time definition.

A p85 based on six finished tickets will flag the board more or less at random. Wait until you have a few dozen completed items of that type, and review the raw list until then. The oldest ticket is still worth a conversation.

Don't rank people by age. Tickets go stale because of reviews, dependencies, and oversized slices far more often than because of one slow engineer. Share the list with the whole team.

A spreadsheet works if you store start dates. Falrow records stage changes and shows cycle time next to open work, so the p85 line and the aging list fit in one review, but any tracker with a start date can run the same meeting. Cycle time, cumulative flow, and forecasts give you the history, and aging checks what hasn't finished yet. Teams that publish dates, including launch dates marketing can plan against, need that check first.

Every review should end with an action on each of the oldest tickets.

FAQ

What is aging work in progress?

Aging work in progress is how long each open ticket has spent in an active stage: today minus the start of active work, usually counted in working days and listed oldest first. Cycle time only appears once a ticket is done, so a calm average can sit right next to a ticket that has been active for weeks.

How is aging different from cycle time?

Cycle time measures completed tickets, from the start of active work to done. Aging measures tickets that aren't done yet, and it keeps climbing until they finish. Use cycle time for baselines and aging for the current board. Compare the two on the same ticket type with the same start rule, or you'll flag the wrong items.

How often should a team review aging work?

Once a week is enough for most teams. Age moves in days, so a daily pass mostly adds noise. Mark anything older than your recent p85 cycle time and write a next action for it, and keep standups to those tickets.

Does a high age always mean a ticket is blocked?

No. A high age means the ticket has been active longer than recent finished work. The cause might be review, a pending decision, a large slice, or work that's still underway. Blocked work is the subset with a named reason it stopped. Whatever the cause, put a date, a split, or a reassignment on each case.

More on flow metrics & forecasting