Skip to main content

Why Basic Kanban Breaks and How Trello, Jira, and Linear Compare for Growing Teams

Four flow controls decide when a simple board stops working, and which tool fits

Why Basic Kanban Breaks and How Trello, Jira, and Linear Compare for Growing Teams
Topic Software
Published
Read Time 19 min

It stops being right when a card cannot move until someone outside the team approves it, when two projects compete for the same reviewer, and when nobody can say how long work takes.

Process complexity triggers that break, whatever the headcount. A six-person agency running two client accounts with an external legal review carries more flow complexity than a thirty-person team shipping one product. Basic boards struggle when they are asked to show dependencies, external waits, shared ownership and several workstreams at once, while offering no cap on work in progress and no measurement of elapsed time.

The repair is structural. Add flow control, a second dimension, explicit links and time data. More columns alone only add places for work to sit.

Quick Take

  • Small boards fail on complexity, so a five-person team with linear work can stay on a simple column board for a long time.
  • Test every candidate against four specifications: a WIP cap with a pull rule, a second dimension, dependency or parent-child structure, and cycle-time reporting.
  • Trello fits small simple boards; Jira fits dependency-heavy software work; Linear fits software teams that want board and list parity with sub-issues, provided the analytics plan fits.
  • Confirm current prices and test whether a limit warns or blocks before you commit a team to a rebuild.

What a Kanban board does and when three columns are enough

Kanban is a management method applied to an existing process. It does not replace how a team already works. It visualises knowledge work and its flow, limits how much work is in progress, and can be applied to a single team or across several. At larger scope, a set of boards at different levels of detail, each with its own limits, is still ordinary Kanban practice.

A board carries a small set of parts. Columns represent stages of the workflow. Cards represent work items. WIP limits cap how many cards a column holds. Swimlanes are horizontal rows that group work by type, priority or team. Commitment and delivery points mark where work is pulled into active effort and where it is finished. Three columns cover simple, short-cycle work; a software team usually needs Backlog, Ready, Development, Code Review, Testing and Done, because review and testing are where delays actually happen.

That simplicity is the product. Trello presents work as boards, lists and cards that move left to right, with real-time updates, Power-Ups and access from a browser or a phone. For a team of four tracking a shared to-do list, that is enough: the whole board reads in one glance, nobody needs training, and nothing has to be configured before the first card goes up.

Why basic boards stop reflecting the work

Boards lose their explanatory value when the cards can no longer describe reality. That happens in a handful of recognisable ways, and most teams hit several at once.

  • Multi-step work is compressed into one column. A card in “In Progress” may be in design, in code review, or waiting for a test environment. The board shows movement while the delay stays invisible.
  • External review dependencies sit off the board. Legal, security, finance or a client approver holds the card. The column shows a queue nobody on the team can clear.
  • Shared ownership blurs accountability. Two names on a card, no single finisher. The card drifts because each owner believes the other is on it.
  • Concurrent projects share one surface. Two or three streams land in the same “Done” column, so finished volume looks healthy while neither project moves.
  • Boards grow wider than a screen. Status columns multiply, horizontal scrolling starts, and card detail gets compressed until the board cannot be read at a glance.
  • Stale cards accumulate. A card that has not moved in months still renders as active work, and the board inflates into a history archive.
  • Status exists without flow control. Nothing caps the in-progress column, so starting work stays cheaper than finishing it.

Simple boards work well within their design limits. They lose explanatory value when asked to show dependencies, external waits, shared ownership and several workstreams at once.

Four specifications that separate a task list from a Kanban engine

One distinction matters more than the rest, and teams get it wrong constantly. A work-in-progress limit can mean three different things: a team policy written in plain language, a visual constraint that colours a column when a count is exceeded, or a hard block that refuses a status transition. Only the third enforces itself. The first two depend on people.

Cap work in progress and write the pull rule

A WIP limit sets a maximum amount of work in a given status. It makes bottlenecks visible before they become a delivery problem, and it pushes the team toward finishing rather than starting. Set the number after watching the average amount of work sitting in that status over several cycles, and expect the first value to be wrong.

Many teams set the in-progress cap below headcount, which keeps room for code reviews and pairing. When the cap is consistently breached, that is a signal to examine the process or adjust the limit deliberately, not to raise it every morning so the red clears.

Now the enforcement question. Jira documents column constraints that set minimum or maximum work-item counts and trigger red or yellow visual indicators. Those values do not change the number of work items displayed in the column, so a full column still shows every card, and the documentation describes a warning rather than a blocked transition. For Linear, the board-layout documentation covers lane grouping and board configuration; a team that needs a hard cap should test whether its plan enforces one. Trello teams generally need a team rule or automation to cap work, and should confirm any native control in a trial. In each case, the pull discipline has to be owned by a person or added through automation.

Add a second dimension with swimlanes

A lane groups cards horizontally, so one board can separate work without splitting into five boards. Common groupings are team, urgency, client and project phase. Vendor practice guidance consistently points the same way: split the board by urgency when deadlines are tight, by department when several functions share a project, and by phase when the work is long.

Linear supports this directly. Board layout is available in most views, and a board can display an extra dimension as lanes or rows, grouped by team, cycle, assignee, lead, initiative or project health among others. Lanes collapse, which keeps a large board readable. The limits matter: board layout does not appear in Triage or Inbox, and it cannot be set as a universal default across the workspace, so each saved view needs its own choice.

A lane answers which bucket a card belongs to. It says nothing about what blocks what, which is why the third specification exists.

Model hierarchy and dependencies explicitly

Two structures solve two different problems. Parent-child hierarchy breaks a large item into smaller pieces: Linear’s sub-issues exist for work too big for one issue but too small for a project, and they are the practical way to split work shared across teammates. Sub-issues inherit some parent properties, and optional team-level automation can close parents or sub-issues when the other side completes.

Dependency links connect separate items that must move in order. Jira documents linked work items that surface and manage dependencies and related work, including links that cross spaces and link types configured by administrators. Those links can be shown on board cards through view settings, applied per user.

Lanes and links solve different problems. A lane tells you a card belongs to the platform team. A link tells you that card cannot start until the platform team’s migration finishes. Teams that only add lanes usually discover the missing half within a quarter.

Measure cycle time, not just completion

Cycle time runs from the moment work starts to the moment it finishes. Lead time runs from the moment the request was logged to completion. The gap between them is your intake queue, and it is frequently the larger half of the wait.

Jira’s control chart reports cycle time or lead time with an average, a rolling average and a standard deviation, and its documentation notes that past performance can be used to set targets for Kanban work. That report applies to company-managed spaces, which is a real constraint for teams on other configurations.

Linear gates analytics behind plan. Insights is available to workspaces on the Business and Enterprise plans and covers measures including time from issue start to completion and time from creation to completion, with percentile markers, burn-up charts, link sharing and CSV export. Trello’s pricing page lists a Dashboard view among paid-tier features; it is a reporting view rather than a documented cycle-time distribution, so treat it as a progress dashboard until you test it.

Averages hide the tail. When some cards finish quickly but a minority take much longer, the average describes neither group and your forecast will be wrong for both.

Four tests for flow control as work becomes more complex. Learning-curve notes are editorial judgment.
SpecificationWhat it controlsPass testWhat often gets mistaken for it
WIP cap with a pull ruleHow much work can sit in one stage at onceA named person can state the cap and what happens at breachA coloured column header nobody acts on
Second dimensionHow several workstreams share one boardLanes split by team, client or phase, and collapse cleanlyFive separate boards nobody reconciles
Dependency and hierarchyWhat blocks what, and what belongs inside whatEvery blocked card names the item it waits onA lane labelled with a team name
Cycle-time reportingHow long work actually takes, and how much it variesA distribution you can read, not just a single averageA completion chart with no elapsed time

Trello, Jira, and Linear compared

The three tools solve different problems, and the honest comparison is about where each one stops being enough.

Comparison of Trello, Jira and Linear on flow controls, structure, reporting, learning curve and published pricing evidence. Learning-curve ratings are editorial judgment. Confirm current prices with each vendor.
DimensionTrelloJiraLinear
WIP controlNo native column cap is established by Trello’s product page; use a team rule or automation and test itColumn constraints trigger red or yellow warnings; they do not change the displayed item countBoard layout is documented; test whether your plan enforces a hard cap
Swimlanes and groupingBoards and lists are the primary structure; lane grouping is not established by the product pageBoard columns map to statuses; verify lane configuration in the specific board before relying on itLanes group by team, cycle, assignee and more; not available in Triage or Inbox
Dependency and hierarchyCards can link to Jira work items; no native dependency model is established by the product pageLinked work items cross spaces; link types are admin-configurableParent and sub-issues split large work; optional automation can close them
Flow reportingDashboard view is listed on paid tiers; treat it as a progress dashboardControl chart reports cycle time or lead time with average and standard deviation; company-managed spaces onlyInsights covers start-to-completion and creation-to-completion time with percentiles; Business and Enterprise
Learning curveLowest of the three; a new user can be productive on day oneHighest; column-to-status mapping, workflow schemes and admin permissions all need configuringModerate; familiar to software teams already working in issue lists
Published pricing evidenceFree tier caps collaborators and boards; paid steps published; confirm billing termTiers published; exact per-user amounts require a quoteNo per-user amount confirmed; get a quote

Trello. The cheapest path to a working visual board and the easiest to teach. Trello’s pricing page lists a free plan for up to 10 collaborators and 10 boards per Workspace. It also shows a $6 monthly-billing per-user step and a $12.50 monthly-billing higher step, plus Enterprise pricing on an annual per-user basis. Confirm the annual monthly-equivalent wording and the plan each amount applies to before budgeting, because the page does not make every billing-term comparison fully clear.

Jira. A strong fit for dependencies and flow reporting. Linked work items cross spaces, control charts separate cycle time from lead time, and administrators can configure link types. The cost is administrative: columns map to workflow statuses, some reporting is limited to company-managed spaces, and a red column header still relies on a human to act. Jira’s pricing page lists Free, Standard, Premium and Enterprise tiers; exact per-user amounts need a vendor quote.

Linear. A strong fit for a software team that wants the same data in board and list form. Lane grouping, parent and sub-issues, and a fast issue editor make it comfortable for engineering work. Analytics is the gate: Insights requires Business or Enterprise, so calculate the plan you actually need before assuming the reporting is included.

Verify prices and enforcement before you commit

Two checks stop most bad purchases. First, price: published per-user figures vary by billing term, seat count and region, and not every amount is published on the vendors’ pricing pages. Get a quote for your real headcount. Second, enforcement: build a scratch board, add a column cap, and try to exceed it. You will learn whether the tool warns, blocks, or ignores you, and that answer shapes how much process documentation you need.

Which tool fits which team

Pick by the constraint that hurts most this quarter, not by feature count.

Fit by team type and workflow constraint, not by universal ranking.
If your team looks like thisStart withWhy it fitsVerify first
Under ten people, one stream of work, a short cycleTrelloFast setup, a readable board and a free tier that covers a small WorkspaceWhether you need more than team convention to cap work
Software delivery with cross-team dependencies and reporting obligationsJiraDocumented linked work items across spaces and cycle-time reportingWhich spaces are company-managed, and who administers workflows
Product and engineering teams wanting board and list parity with sub-issuesLinearLane grouping, parent and sub-issues, and a consistent issue modelWhether the plan that includes Insights fits the budget
Operations or agency teams running several process typesScore every candidateMultiple workflows, external approvals and client handoffs stress all three tools differentlyDependency depth and whether a table or process view is needed alongside the board

Two rules do most of the work. If your pain is clutter from one simple stream, buy nothing and tighten the board you have. If your pain is a card that waits on another team, choose the tool that models that dependency and report it, because no amount of extra columns will surface a wait that lives in someone else’s queue. For a wider view of how boards differ on benefits and specifications, see this breakdown of Kanban tool benefits and specifications.

Move a growing team off a basic board in six steps

The sequence below moves one team, not the whole department, and it starts with what is already on the board. Do it once and you will have a defensible case for whichever tool you buy.

Prerequisites

  • A snapshot of the current board, including every card in an active column
  • One named process owner who can change the workflow rules
  • Completion dates for several cycles of finished work
  1. Audit the stale cards.Sort every card in an active column into three piles: work actually in progress, work waiting on someone, and work nobody is currently touching.

    A card in “In Progress” reads as active work whether or not anything is happening to it. The third pile is the history your board has been carrying, and it is usually the reason the board feels heavy.

    You will know the audit worked when every remaining card has a named owner who can describe the next action required.

  2. Map the stages work really passes through.Take one recently completed item and walk it from request to delivery with the people who did the work, writing down every handoff and every wait.

    Each stage where work can pile up deserves its own column. A single “In Progress” column covering design, build, review and testing hides the three places where the delay actually sits.

    The map is finished when every handoff in the walkthrough corresponds to a column or a mapped status, and no column exists that nobody can define.

  3. Set one capacity-based WIP cap.Count the average number of items sitting in your busiest status across several completed cycles, then set the cap at or just below that number. Set the in-progress cap below headcount so reviews and pairing have room.

    This is where ambition usually wins over evidence. A cap set from a target the team hopes to hit will trip on the first morning and be ignored by the end of the week.

    Run the cap across several completions. If it never trips, it is too loose to teach anything; if it trips constantly while work still finishes on time, revisit the number once, deliberately, and write down why.

  4. Write the pull rule.State, in two sentences, who may pull work into a stage and what happens when that stage is full. The usual answer is that the person pulling helps clear the stage before starting anything new.

    This is the step that separates a working limit from a decorative one. Where a tool only warns, as Jira’s documented column constraints do, the rule needs a human owner or an automation trigger.

    Test it by asking two team members to state the rule without looking at the board. If the answers differ, the rule is not written clearly enough to survive a busy week.

  5. Model dependencies and hierarchy.For every card that cannot move, create a link to the item it waits on. For any card that hides a multi-day piece of work, split it into sub-issues with a parent.

    Lanes group work; links connect it. A lane labelled with another team’s name still leaves you guessing whether their migration blocks your release.

    The structure is complete when every blocked card names its blocker, and that blocker exists on a board somebody owns.

  6. Baseline cycle time and book a review.Chart completion dates from the historical data you gathered and record both the median and the slowest group of items. Then schedule a short review of the board against that baseline.

    The chart describes the past. It only becomes useful when the spread is narrow enough that past variation predicts future variation, which is why the slowest group matters more than the average.

    The baseline is ready when the chart shows a readable distribution and the review produces one concrete change to the board per cycle.

Verify the result

  • No card in an active column lacks a named owner and a next action
  • Every delay the team named during the walkthrough has a column or mapped status
  • The in-progress cap has tripped at least once and prompted a conversation
  • Every blocked card links to a named blocker owned by someone
  • A cycle-time baseline exists and one review date is on the calendar

Troubleshooting the move

The WIP column is red every single day

Either the cap was set from ambition rather than observed capacity, or the real bottleneck sits upstream and work is arriving faster than it can leave. Check where cards wait before they reach the capped column. Adjust the cap once, write down the reason, and stop raising it to clear the colour.

Cards sit in review for weeks

Review is a queue with no assigned capacity. Give the review column its own cap, define what “ready for review” means so reviewers are not rewriting the submission, and name who is on review duty this cycle.

The reports look healthy but delivery still slips

Reports describe finished work. If the spread of completion times is wide, an average tells you almost nothing about the item in front of you. Compare your slowest group of items against your typical one and set expectations from the slower group.

Misconceptions that survive the comparison

More columns mean more control. Extra columns add places for work to sit. Without a cap and a pull rule they add width, scrolling and the illusion of precision.

A red header is a hard block. Jira’s documented constraints warn visually and do not change the number of items shown. Whether anything blocks a transition depends on workflow configuration and automation you verify yourself.

Swimlanes are dependencies. A lane answers which bucket a card belongs to. A link answers what must happen first. Teams that add only lanes keep missing dates.

A cycle-time chart is a delivery promise. It is a description of past work. It supports targets when variation is low and misleads when variation is high.

The tool creates pull discipline. No product can stop a person from dragging a card. Discipline lives in the rule, the review and the person who owns the process.

A missing feature on a public page means the feature does not exist. Features sit behind plans, admin settings and configuration. Score anything unverified by testing it on a scratch board.

Start with the audit, then choose the tool

Buy for the constraint you can name. Trello suits a small team with a simple, linear board and a tight budget. Jira suits a software organisation where cross-space dependency links and cycle-time reporting justify the administrative overhead. Linear suits a product and engineering team that wants the same issues in board and list form, plus parent and sub-issues, with analytics priced on a higher plan.

If your team runs several process types with client or external approvals, score each candidate against the four specifications before any demo. Operations and agency leaders usually find the deciding factor in dependency depth and reporting, not in card design.

The first three actions cost nothing. Audit the stale cards, map the stages work passes through, and set one capacity-based cap on the busiest column. Then review after several cycles, because a board that measures nothing cannot tell you whether the change helped.

Teamhood publishes its best Kanban boards page as a vendor resource. Treat it as a vendor shortlist, not an independent ranking, then score every name on it against the four specifications and your own price quote.

Ifeanyi Okondu

About the Author

Ifeanyi Okondu

Ifeanyi Joseph Okondu is a Product Manager, technical writer, and creative. Connect with him on LinkedIn

View all posts by Ifeanyi Okondu →
Comments

Be the First to Comment