Most of the Workforce Management coverage so far is aimed at high-volume contact centres. If you run a five-person service desk, it's easy to assume none of it applies to you.
I'd argue the opposite. Small teams get hurt by the same problems — work sitting unassigned, one person quietly absorbing everything, nobody knowing who's actually on today — they just can't afford a dedicated person to manage it manually. What small teams can't afford is a six-week configuration project to fix it.
So this is a practical setup order for a small organisation, plus the decisions that are worth making before you flip the toggle.
First, check it applies to you
Workforce Management is available on Jira Service Management Premium and Enterprise plans, and it's included — no separate add-on or licence. If you're on Standard, stop here. atlassian
Then ask an honest question: is routing actually your problem?
If your team of four all watch the same queue and pick up their own work without drama, WFM solves nothing and adds configuration you have to maintain. The signals that it will help are specific:
- Requests sit unassigned for hours because everyone assumes someone else saw it
- The same one or two people take everything, and you find out they're drowning only when they tell you
- You genuinely don't know who's on shift without asking
- Someone is currently maintaining a rota in a spreadsheet
One of those is enough. None of them, and you're better off leaving it off.
Step 0: the Atlassian Team is the real prerequisite
This is the step people skip and then get stuck on. Workforce Management routes at the team level — every agent who can be routed to belongs to an Atlassian Team connected to your service space, so you need that team set up first. atlassian
In a small organisation this is usually fine, but it does force a decision you may have been avoiding: who is actually on the support team? Not who has an agent licence, not who occasionally helps out — who should receive work automatically. Write that list down before you touch any settings.
Step 1: pick one space and leave the rest alone
Schedules and capacity are space-scoped and don't carry across spaces. That's a constraint in a large org, but for a small one it's a gift: start with the single space that hurts most, learn there, and don't try to model the whole organisation on day one. atlassian
The toggle lives in your service space under Features, and once it's on, a Workforce entry appears in the sidebar for admins and agents. atlassian
Step 2: set capacity with evidence, not a guess
Team capacity tells Workforce Management what counts as a full workload — how many open work items an agent can handle. atlassian
The temptation in a small team is to pick a round number. Don't. Two better options:
If you have history: look at concurrent open items per agent over the last month. Find the level at which things started slipping — first responses getting slower, requests going stale. That's roughly your ceiling.
If you don't: start deliberately high and tighten. A limit set too low in a small team is worse than no limit at all, because when everyone is "at capacity" the work has nowhere to go and you've reinvented the unassigned queue with extra steps.
Then watch it for a week — the capacity utilisation view (admin-only) tells you whether your limits match how the team actually works. Treat the first number as a hypothesis. atlassian
Step 3: keep shifts coarse
Shifts are required for automatic routing to work. And this is where small organisations most often over-engineer. atlassian
You will be tempted to model reality precisely: the part-timer who does Tuesday and Thursday mornings, the person who leaves at 3pm on Fridays, the one who works from the other time zone every third week. Resist it. A schedule that mirrors every individual arrangement is a schedule nobody maintains, and an out-of-date schedule routes work to people who aren't there — which is worse than not routing at all.
Model the coverage pattern, not the people. Core hours, who covers them, done. The day-to-day exceptions are handled by agents marking themselves available or unavailable, which costs them one click and is far more likely to stay current than a rota you edit centrally.
One practical note from the GA announcement: schedules must have an end date. Decide now whether you're building quarterly schedules or rolling monthly ones, and put the renewal in someone's calendar. atlassian
Step 4: use recommendations before automation
This is the part I'd push hardest for small teams. Alongside automatic assignment, WFM shows up to three recommended agents when someone is manually choosing an assignee from a queue or work item. jimiwikman
Run in that mode first. The human still assigns, but they see what the system would have chosen. Two weeks of that tells you whether your capacity limits and shifts are sane — and it builds trust, because the team sees the recommendations be right before the system starts acting on its own. If you go straight to automatic and the first week's assignments are wrong, you'll spend a month arguing about it instead.
Step 5: name an owner, in writing
Every WFM implementation I expect to see fail will fail the same way: the schedule decays. Someone changes hours, someone leaves, a new person joins and nobody adds them to the Atlassian Team, and six months later the routing is quietly wrong.
In a large organisation there's a team lead whose job this partly is. In a small one, it's nobody's job unless you say so. Pick the person, tell them it's theirs, and put a recurring fifteen minutes in their calendar. This is less work than the spreadsheet they're replacing — but it isn't zero work, and pretending otherwise is how it rots.
What to deliberately skip for now
Skills-based routing, a skills catalogue, configurable routing algorithms, reporting and leave management are in the next wave, with an EAP cohort open. atlassian
Worth knowing about, worth not waiting for. In a five-person team, skills live in people's heads and the overlap is usually large enough that availability and capacity get you most of the benefit. Ship the simple version now; revisit skills when the team is big enough that routing to the wrong person actually costs you something.
What good looks like after 30 days
- Nobody asks "who's on today?" in chat
- Nothing sits unassigned overnight
- You can answer "is anyone overloaded right now?" by looking rather than guessing
- The spreadsheet is gone and nobody has rebuilt it somewhere else
If you get three of those four, it was worth the afternoon.
Thanks a lot.
