
In this article
A team can have a sensible roadmap, an orderly backlog, and a weekly priority meeting—and still feel as if none of its important work moves. The usual explanation is that the team has too much to do. Often it has too much started.
This distinction matters because a ranked list answers one question: what should we do before other work? It does not answer a more operational question: what will we decline to begin while the team is already carrying unfinished work?
Busy team leads are often expected to solve both problems with one artifact. A backlog becomes a promise, a capacity plan becomes a schedule, and a board full of active cards is treated as proof of momentum. Yet these tools make different commitments. Priority ranking clarifies sequence. Capacity allocation protects categories of work. Work-in-progress limits control how much unfinished work the system is allowed to hold at once.
They are complementary, but they are not interchangeable. Choosing the wrong one can create a familiar pattern: every request is acknowledged as important, every initiative has an owner, and the team quietly learns that finishing is less visible than starting.
Three tools, three different promises
It helps to separate the methods before asking a team to adopt any of them. A priority list is a judgment about relative value. Capacity allocation is a decision about how much attention different obligations may consume. A work-in-progress, or WIP, limit is a boundary on concurrent commitments.
The distinction may sound semantic until a customer escalation lands on the same day as a planned release, a hiring interview, and an internal request for analysis. A priority list can tell you that the escalation comes first. It cannot tell you whether the release work already underway should pause, whether another person can take the escalation, or whether the team has exhausted the portion of its week reserved for reactive work. Those are capacity and WIP questions.
| Approach | Primary question | What it improves | What it cannot settle alone |
|---|---|---|---|
| Priority ranking | What matters most next? | Sequence and tradeoffs in the backlog | How much can be active simultaneously |
| Capacity allocation | How much team attention belongs to each type of work? | Protection for planned, operational, and developmental work | Whether too many items are already underway |
| WIP limits | How many items may be in progress at one time? | Flow, handoffs, and the visibility of blocked work | Which incoming request deserves priority |
The practical implication is modest but important: do not announce a new priority system when the actual problem is uncontrolled intake. And do not impose a WIP limit when the real tension is that support work has no acknowledged place in the plan.
Priority ranking is necessary, but it is only a queue
Ranking work is valuable because it makes tradeoffs explicit before a team spends effort. Without it, the loudest request, the most senior stakeholder, or the newest message can become a de facto scheduling system. A clear order gives people permission to ask whether a new request belongs above the work already promised.
But a ranked backlog is a queue, not a capacity constraint. It can hold fifty sensible requests without telling the team to begin more than three. This is where leaders sometimes mistake clarity for control. The backlog looks clean, so they assume the work system is healthy. Meanwhile, people pull items from the top whenever they find an open hour, a useful conversation occurs, or someone wants to demonstrate responsiveness.
That pattern is especially expensive for work that needs context to accumulate: product discovery, technical design, a difficult hiring decision, a piece of analysis with several dependencies. Research on task switching has found that alternation carries a time cost, and that the cost rises with more complex rules; cues can reduce it, but they do not turn switching into no-cost work. The underlying task-switching research is not a prescription for knowledge-work calendars, but it supports a basic management observation: changing priorities has a real operating cost even when each individual request is legitimate.
A priority list works best when the team can reliably distinguish “next” from “now.” If everything near the top is immediately started, the list has stopped ordering work and begun licensing parallelism.
A priority list can make the next choice clearer without making the current workload smaller.
Capacity allocation makes tradeoffs visible
Capacity allocation begins from a condition most teams already know but do not always name: not all work arrives in the same shape. A product team may owe time to planned delivery, customer or production issues, discovery, maintenance, mentoring, and coordination. A recruiting team may need room for live hiring, candidate communication, process improvement, and interview calibration. Treating all of it as one backlog usually makes recurring obligations look like interruptions.
The useful move is to reserve attention for categories that are predictably necessary, while keeping the allocations provisional. This is not a promise that every category will receive the same amount of time every week. It is a way to see which kind of work expands when the team is under pressure and what will be displaced as a result.
Google’s SRE guidance offers a specialized but useful example: it separates repetitive, reactive operational toil from longer-term engineering work and uses an explicit ceiling for operational load. That SRE model should not be copied as a universal percentage for every function. Its transferable idea is that recurring reactive work needs a visible limit; otherwise it tends to consume the capacity meant to reduce future demand.
For a team lead, capacity allocation changes a difficult conversation. Instead of telling a stakeholder, “We cannot help,” you can say, “We can take this on, but it uses the capacity we had reserved for reliability work this week. Which commitment should move?” That is not defensive language. It is a clearer representation of the choice.
Illustrative weekly allocation
The example below is not a benchmark or a recommended split. It simply shows how a seven-person team might make room for work that would otherwise be treated as an exception.
When every request is allowed to start, waiting does not disappear; it becomes harder to locate.
- Planned delivery40 percent of team capacity
- Customer and operational work25 percent of team capacity
- Maintenance and reliability20 percent of team capacity
- Discovery and planning15 percent of team capacity
Usable sample wording for a new request
When the request fits reserved capacity
We can take this in the support capacity we set aside for this week. I will confirm the owner and expected next step by tomorrow.
When the reserve is already used
This is important, but our reactive capacity is already committed. We can either move the maintenance review scheduled for Thursday or place this first for next week. Which outcome is more important?
Use the wording to make the displaced work explicit rather than treating it as an unspoken cost.
Why starting everything makes the queue less honest
WIP limits address a different failure. They ask a team to define what counts as started, what counts as finished, and how much work may sit between those two points. The current Kanban Guide treats this shared definition of workflow—including how WIP is controlled—as a core practice. The important word is shared. A personal intention to focus is fragile when the group has no agreement about when new work may enter.
WIP limits can feel uncomfortable because they make waiting visible. A request may be approved and important but remain unstarted until an existing item is completed or deliberately stopped. That is often read as bureaucracy, particularly in teams that have earned trust by being helpful. Yet invisible waiting does not disappear when every task is marked in progress. It migrates into delayed reviews, half-written documents, unanswered dependencies, and work that belongs to someone who has changed focus three times.
Queueing theory gives this an unintuitive but useful frame. Little’s Law relates average items in a system, average completion rate, and average time in the system. Little’s account of the law and its applications is not a reason to force a neat formula onto uneven creative work. It is a reminder that, when throughput does not rise, adding more active work generally makes the time work spends in the system longer. A board with many active cards may describe a queue, not a productive team.
A WIP limit is therefore less about squeezing people and more about making finishing a team event. If the design-review column is full, the next response is not automatically to start a new design. It may be to review, unblock, narrow, pair, or renegotiate the items already held there.
- New requests accepted12 work items
- Items started9 work items
- Items completed5 work items
- Items still active4 work items
The cost hidden inside a full board
Not every active item makes the same demand. Some are waiting for an external answer. Some need a short review. Some require concentrated, sequential work from the same person. A single numerical limit cannot capture all of this. Team leads need judgment about the work’s age, risk, dependencies, and reversibility.
Still, the discipline of a limit exposes an important difference between being busy and being committed. When five projects each need the same specialist for a small but critical decision, the specialist is not merely overloaded. The system has created five competing handoffs. When a manager assigns a sixth project because it has a clear owner, ownership is being confused with capacity.
The chart below is illustrative: it shows how the number of active streams can increase coordination exposure even if the nominal workload has not changed. The point is not that every work stream has the same cost. It is that concurrent commitments multiply the places where work can wait for context, review, or approval.
Capacity is not demonstrated by how much a team can accept in one week, but by what it can bring to completion without private heroics.
- Two active streams4 coordination touchpoints
- Three active streams7 coordination touchpoints
- Four active streams11 coordination touchpoints
- Five active streams16 coordination touchpoints
Choose the constraint that matches the failure
There is no universal winner among these approaches. A stable team with a sprawling backlog may need stronger prioritization. A team repeatedly sacrificing maintenance, documentation, or candidate care to urgent delivery may need a capacity allocation. A team with too many half-finished initiatives may need WIP limits before it needs another planning ritual.
Many teams need all three, applied lightly and in order. First, clarify the sequence of requests. Then decide which kinds of work require protected capacity. Finally, limit what may be actively underway. The order matters because a WIP limit without priorities can turn the team into a gatekeeper for whoever argues hardest, while a capacity allocation without WIP control can still leave every category fragmented.
Scrum makes a related distinction through the Sprint Goal: its purpose is to create coherence and focus around a valuable objective rather than leave the team pursuing separate initiatives. The Scrum Guide’s treatment of the Sprint Goal is useful even for teams that do not use Scrum. A goal is not a substitute for capacity management, but it supplies a reason to protect some work from being displaced by the merely available.
A visible exception is easier to manage than a quiet pattern of broken commitments.
- WIP limit5 work items
- Active work before review5 work items
- New item waiting to start1 work items
- Completed item freeing capacity1 work items
Match the intervention to the pattern you can observe
- The team debates what matters most but begins little work
Improve priority ranking and make the next few items unmistakable before redesigning the board.
- Recurring operational work repeatedly erases planned improvement work
Allocate capacity by work type and revisit the allocation when the demand pattern changes.
- Many items are active, blocked, or awaiting the same reviews
Set a WIP boundary at the relevant stage and make unblocking existing work the default response.
- A sudden external event changes the week
Use the exception path, state what will move, and review whether the event revealed a missing capacity category.
A policy must make room for uncertainty
The strongest capacity policies do not pretend that interruptions can be eliminated. They state what happens when interruption arrives. A customer incident may legitimately break a WIP limit. An executive decision may change priorities. A candidate may need a faster response than the normal process allows. The policy is useful because it makes the exception visible and asks what work will be paused, reassigned, or removed.
That final step is where leadership becomes more than queue administration. A team lead cannot make demand disappear by naming it. They can prevent every new request from becoming private overtime, a silent delay, or an additional project that nobody has explicitly agreed to carry.
The aim is not an immaculate board. It is a work system honest enough to show what the team has started, what it can finish, and what must wait.
- Reserved reactive capacity15 percent of weekly capacity
- Request using reserve10 percent of weekly capacity
- Reserve remaining5 percent of weekly capacity
- Planned work displaced0 percent of weekly capacity
A four-week introduction without a major process overhaul
- Week 1: Observe
Record what the team starts, what it finishes, and where work waits. Do not set limits yet.
- Week 2: Define
Agree on the start and finish points for a small slice of the workflow, plus the work categories that deserve visibility.
- Week 3: Try one boundary
Use one modest WIP limit or capacity reservation for a week and log every exception.
- Week 4: Review
Discuss whether the boundary exposed a bottleneck, created an unintended delay, or protected work that normally disappears.
Treat the first policy as a test of the workflow, not a verdict on individual effort.
Career coach focused on helping women navigate the tech industry. I’ve worked in product and engineering roles before shifting to coaching full-time. Now I write about imposter syndrome, salary negotiation, and how to grow without burning out.