# Kanban for Laboratories: A Step-by-Step Guide By the end of this guide you will be able to map your lab's workflow, build a kanban board on a wall or in free software, set work-in-progress limits that actually reduce bottlenecks, track cycle time, and run a board review that fits inside fifteen minutes. You need nothing installed to start. A whiteboard, a pack of sticky notes, and a marker will get a physical board running in an afternoon. If you prefer digital tools, a free Trello or Notion account is enough, and the steps below cover both. ## What Kanban Is and Why Laboratories Use It Kanban is a visual [workflow management](/knowledge/bioinformatics/snakemake-vs-nextflow-workflow-architectures) method that originated in the Toyota Production System, where engineers such as Taiichi Ohno used physical cards, called kanban, to signal when parts should move to the next stage of manufacturing. The method was later formalized for knowledge work, most prominently in David J. Anderson's book *Kanban: Successful Evolutionary Change for Your Technology Business*, and it now underpins much of modern research project management in software teams, hospitals, and laboratories. A kanban board has three basic parts: 1. **Columns** that represent the stages of your work, for example "Planned", "In preparation", "Running", "Analyzing", and "Done". 2. **Cards** that represent individual units of work, for example "Extract RNA from cohort B samples" or "Draft figure 3 for the manuscript". 3. **Work-in-progress limits**, usually abbreviated as WIP limits, which cap how many cards may sit in a given column at once. The core idea is the **pull system**. A person starts a new task only when there is free capacity, meaning an open slot below the WIP limit, rather than when someone pushes a new assignment onto them. In a laboratory this maps naturally onto real constraints. An incubator holds a fixed number of plates. A flow cytometer is booked by the hour. A single experienced person can only supervise two undergraduates at once. Kanban makes those constraints visible instead of letting them hide inside people's inboxes. The method suits laboratory work for a specific reason: lab projects are a mix of long-running, uninterruptible processes (a fourteen-day culture, a month-long animal study) and short, interruptible tasks (order reagents, reply to a collaborator, troubleshoot a primer). A board that separates these two kinds of work keeps the long processes visible while the short tasks stop from burying them. ## What You Need Before You Start For a physical board: - A whiteboard, a wall section, or a large corkboard - Sticky notes in two or three colors, one color per work type - Thin washi tape or marker lines for columns and swimlanes - Magnets or pins if you use laminated cards for recurring protocols For a digital board, any one of these free options works: - **Trello** (trello.com), which is the simplest and a good first choice for most labs - **Notion** (notion.so), which combines the board with a documentation database - **GitHub Projects**, useful if your lab already versions protocols or analysis code there - **Jira**, common in larger core facilities and bioinformatics groups No specialized software is required. The method matters far more than the tool. ## Kanban for Laboratories Step by Step This walkthrough builds a working board for a hypothetical [molecular biology](/blog/careers/molecular-biology) lab of four people: one principal investigator (PI), one postdoctoral researcher, one graduate student, and one [research technician](/blog/guides/research-technician). Adapt the details to your own setting. ### Step 1: Map Your Workflow Before You Draw Anything List the actual stages your work passes through, from idea to published or archived result. Interview everyone who will use the board. Write each stage on a sticky note and arrange them left to right. A typical research lab workflow looks like this: ``` Backlog | Planned | In preparation | Running | Blocked | Analysis | Writing | Done ``` Guidelines for this step: - Keep the number of columns between five and eight. More columns make the board hard to read at a glance. - "Blocked" is a special column or a label, not a stage. It holds anything waiting on an outside party, such as a purchase order, an animal protocol approval, or a sequencing facility turnaround. - "Done" must be defined precisely. "Done" for an experiment might mean "data deposited, notebook entry signed, reagents logged". Write that definition down before you build the board. ### Step 2: Decide What One Card Represents Card granularity is the single most consequential design decision in kanban. A card that says "Glioblastoma project" is useless, because it will sit in the same column for two years. A card that says "Load 12 samples on gel" is useful, because it moves within a day. Use these rules of thumb: - A card should take between half a day and two weeks of active effort. If it will take longer, split it. - Long-running, uninterruptible processes (cell culture, animal studies, long incubations) get their own card type and usually their own swimlane, which is a horizontal row across the board. - Recurring maintenance (equipment calibration, freezer inventory, media prep) can be one card per week rather than one card per item. For the example lab, define three card types: | Card type | Color | Examples | |--|--|--| | Experiment | Yellow | "Western blot for p-AKT, 6 conditions" | | Analysis or writing | Blue | "Fit dose-response curves for figure 2" | | Admin or maintenance | Green | "Submit IACUC amendment", "Order DLB reagents" | ### Step 3: Build the Board **Physical board.** Draw vertical lines for columns, draw one horizontal line splitting the board into two swimlanes ("Wet lab" on top, "Analysis and admin" below), and place a sticky note with the WIP limit at the top of each column. You are done in under an hour. **Digital board in Trello.** Follow these commands: 1. Create a board named "Lab Kanban". 2. Create lists named exactly as in Step 1: `Backlog`, `Planned`, `In preparation`, `Running`, `Blocked`, `Analysis`, `Writing`, `Done`. 3. Create labels: yellow for "Experiment", blue for "Analysis or writing", green for "Admin or maintenance". 4. Open each list, click the three-dot menu, choose "Set limit", and enter the WIP limits from Step 4 below. 5. Add members to cards so ownership is explicit. Expected output: a board where dragging a card into a list that is at its WIP limit is either blocked or flagged with a highlighted count, depending on your power-up settings. When the `Running` column shows "4/4" in red, every lab member can see the constraint without a meeting. **Digital board in Notion.** Create a database with properties `Status` (select: Backlog, Planned, In preparation, Running, Blocked, Analysis, Writing, Done), `Type` (Experiment, Analysis, Admin), `Owner` (person), and `Started on` (date). Add a board view grouped by `Status`. The `Started on` date is what you will later use to compute cycle time, so do not skip it. ### Step 4: Set Work-in-Progress Limits ![Calculation of cycle time as WIP over throughput, giving 8 over 4 equals 2 months for the example lab.](/article-images/illustrations/kanban-for-laboratories-littles-law-cycle-time.webp) WIP limits are what separate kanban from a simple to-do list. Without them, every column fills up, nothing is visibly blocked, and the board becomes decorative. Start with limits based on real capacity. Count the people who do the work in each column and the equipment that constrains it, then set the limit near the true capacity, not above it: - `In preparation`: one card per person who does prep work. Four people doing prep gives a limit of 4. - `Running`: count actual concurrent capacity. If the lab has two incubators, a plate reader, and one person running gels at a time, a limit of 4 running experiments is honest. Five is not. - `Analysis`: one card per analyst, usually 2 in this example lab. - `Writing`: 2, because only the PI and postdoc draft manuscripts in this example. You can also use Little's Law, a standard result from queuing theory, to reason about the relationship between WIP, throughput, and cycle time: ``` Cycle time = WIP / Throughput ``` where cycle time is the average time a card takes from starting work to finishing, WIP is the average number of cards in active columns, and throughput is the average number of cards completed per unit of time. Worked example: suppose the lab completes an average of 4 experiment cards per month and keeps an average of 8 experiment cards in active columns. Then: ``` Cycle time = 8 / 4 = 2 months ``` Each experiment takes about two months from start to finish on average. If the lab wants a one-month cycle time without hiring, the equation says active WIP must fall to about 4 cards. That is the argument for WIP limits stated in numbers rather than opinions, and it is the calculation to bring to your next lab meeting when someone objects that limits "reduce productivity". ### Step 5: Write Column Policies A column policy states what must be true before a card enters a column and what must be true before it leaves. Policies prevent the most common board failure, which is cards drifting into "Running" that are not actually runnable. Example policy for entering `Running`: ``` A card may enter Running only when: 1. All reagents are in stock and logged in the inventory. 2. Equipment time is booked. 3. The protocol is written in the notebook or linked from the card. 4. Owner is assigned. ``` Example exit policy for `Running`: ``` A card may leave Running only when: 1. Raw data are saved to the shared drive with the naming convention YYYYMMDD_assay_initials. 2. The notebook entry is complete and signed. 3. Reagent lots are recorded. ``` Pin these policies next to the physical board or put them in the list description in Trello. When a card sits in the wrong place, you point at the policy rather than arguing with a colleague. ### Step 6: Populate the Board Move every active project onto the board as cards. Expect the first population to be humbling. Labs that run this exercise commonly find twenty to forty partially finished items, far more than the team realized. Do not try to run them all. Move everything to `Backlog`, then pull the top three to five items into active columns until the WIP limits are full. Everything else waits. This is the point of the method, and it is also where most of the initial argument happens. ### Step 7: Run a Daily Board Review Stand at the board (physically or on a video call) for ten minutes at a fixed time, for example 9:30. Three questions, in order: 1. Does anything in `Blocked` need action? Who owns it, and by when? 2. Is any column over its WIP limit? If so, what finishes first? 3. Who needs help today? No one reports status to a leader. Everyone looks at the same board and talks to each other. Ten minutes, then back to the bench. ### Step 8: Track Two Metrics Only Start with exactly two measurements: - **Cycle time**: calendar days from the `Started on` date to the date a card reached `Done`. Track it per card type. - **Throughput**: number of cards finished per week, again per type. After four to six weeks, if you want a richer view, build a **cumulative flow diagram**, which plots the number of cards in each column over time. In a spreadsheet, plot each column as a stacked area chart with dates on the x-axis and card counts on the y-axis. Bands that widen over time show accumulating queues, which tells you exactly where your bottleneck lives. For the example lab, after five weeks the technician records these experiment cards: ``` Card Started Done Cycle time Western blot, p-AKT 2026-10-06 2026-10-10 4 days qPCR validation, gene X 2026-10-07 2026-10-17 10 days Flow panel, cohort A 2026-10-13 2026-10-24 11 days Immunostaining pilot 2026-10-15 2026-10-21 6 days ELISA, serum batch 2 2026-10-20 2026-10-28 8 days ``` Average cycle time for the five cards is (4 + 10 + 11 + 6 + 8) / 5 = 7.8 days. Throughput is 5 cards in 5 weeks, or 1 experiment per week. These two numbers, updated monthly, are the baseline against which any process change (a new reagent supplier, a shared equipment booking system, an additional technician) can be judged. ### Step 9: Add a Weekly Replenishment Meeting Once a week, fifteen minutes, the team reviews the `Backlog` and decides what enters `Planned` as slots open. This is where the PI states priorities. Anything not chosen waits in the visible backlog, which converts "why is my project not moving" from an emotional conversation into a board conversation. ### Step 10: Improve One Thing at a Time Each month, pick the single widest queue visible in your metrics and change exactly one thing about the process, then re-measure. For example, if the `Blocked` column consistently holds cards waiting on reagent orders, the change might be moving order approval to a standing weekly decision instead of ad hoc email approvals. Small, measured, reversible changes are the mechanism of improvement in kanban. Large redesigns cannot tell you what worked. ## A Research Project Plan Template Use this template as the starting card content for any new project. It is written in markdown so it pastes directly into Trello card descriptions, Notion page bodies, or a plain text file. This doubles as a lightweight research project plan template for labs that do not need formal project management software. ``` # Project: [Short descriptive name] Owner: [Name] Collaborators: [Names] Card type: Experiment / Analysis / Writing / Admin Target completion: [Date] ## Question [One sentence. What are you testing or building?] ## Success criteria [What result, file, or figure means this card is Done? Be specific: "Western blot image with 6 lanes, quantified, saved to /projects/GHB/fig2/"] ## Steps - [ ] Protocol written or reviewed - [ ] Reagents in stock, lots recorded - [ ] Equipment booked - [ ] Execute - [ ] Data saved and backed up - [ ] Analysis complete - [ ] Notebook entry signed ## Dependencies [Blocked by: card link, person, or approval] ## Log 2026-11-11: Card created ``` Fill every bracket before the card leaves `Backlog`. A card that cannot be fully filled out is not ready to enter `Planned`, and that rule alone eliminates a large share of stalled laboratory work. ## Key Parameters at a Glance | Parameter | What it means | Typical starting point | Adjustment rule | |--|--|--|--| | Number of columns | Stages of your workflow | 5 to 8 | If a card sits in one column for weeks, split that column | | Card granularity | Size of one card | 0.5 to 2 weeks of effort | Split any card estimated over 2 weeks | | WIP limit per column | Max concurrent cards | 1 per person or 1 per instrument | Lower the limit if cycle time is too long, never raise it to clear a queue faster | | Cycle time | Start-to-done calendar days | Track, do not set a target | Judge changes against your own baseline | | Throughput | Cards finished per week | Track per card type | Rising throughput with falling WIP means the system is improving | | Swimlanes | Horizontal rows for card types | 2 to 3 | Add one only when a work type has different stages or policies | | Blocked aging | Days a card sits blocked | Review daily | Escalate any card blocked more than 5 working days | ## Practical Tips for Laboratory Kanban These kanban for laboratories tips come from the failure patterns that recur across labs adopting the method: 1. **Color-code by work type, not by person.** Color per person hides the mix of experiment, analysis, and admin work. Color per type reveals imbalances immediately. 2. **Put long incubations on the card.** Write "4C overnight, day 3 of 5" on the card itself so a glance at the board tells you when it will move. 3. **Give shared equipment its own mini-board or swimlane.** The microscope, the sequencer, and the cell sorter are constrained resources with queues of their own. 4. **Keep a "recurring" row.** Media prep, inventory, and calibration repeat every week. One card per recurring task per week, always in the same column position, keeps them from being forgotten without cluttering the flow. 5. **Photograph the physical board every Friday.** A timestamped photo preserves the board state for lab notebooks and for anyone on vacation. 6. **Let the person doing the work move the card.** Cards moved by the PI on behalf of others stop reflecting reality within two weeks. 7. **Start with a two-week trial on one project only.** Boards adopted lab-wide on day one usually collapse. One project, one month, then expand. ## Common Mistakes and Limitations These are the kanban for laboratories common mistakes that account for most failed adoptions: 1. **No WIP limits.** A board without limits is a to-do list with extra steps. Every column fills, nothing is visibly constrained, and cycle time never improves. Set limits on day one, even imperfect ones. 2. **Cards that are too large.** "Write the paper" cannot be a card. Split it into "Draft methods", "Make figure 1", "Make figure 2", and so on. 3. **Columns that mirror the org chart instead of the workflow.** Columns named after people or job titles cannot show where work is stuck. Columns must represent stages of the work itself. 4. **Defining Done too loosely.** If "Done" means "experiment happened" but the data were never deposited or the notebook never signed, cards vanish from the board while work remains. Write the exit policy and enforce it. 5. **Ignoring the Blocked column.** Cards blocked on a purchase order for three weeks represent three weeks of lost cycle time. Review the blocked column every single day. 6. **Measuring people instead of the system.** Cycle time varies by experiment type. Compare a card against its own card-type baseline, never against a colleague's numbers. 7. **Board decay.** A board that is two weeks stale loses all credibility, and no meeting can recover it. The daily ten-minute review exists precisely to prevent this. 8. **Copying a software team's board unchanged.** Labs have uninterruptible multi-week processes that software teams do not. If a column makes no sense for bench work, change it. The board serves the lab, not the other way around. One limitation applies throughout: every laboratory has different constraints, equipment, staffing, and regulatory obligations, so the limits, policies, and columns here are starting points that each lab must calibrate against its own measured cycle times rather than adopt as fixed rules. ## Frequently Asked Questions ### What is kanban for laboratories? Kanban for laboratories is a visual workflow management method that uses a board with columns, cards, and work-in-progress limits to make lab work, constraints, and bottlenecks visible to the whole team. It originated in manufacturing and was adapted for knowledge work, and it fits labs because it handles both long-running experiments and short administrative tasks on one board. ### How is kanban different from Scrum or a Gantt chart? Kanban is continuous and does not require fixed-length sprints, upfront estimates, or role changes, while Scrum organizes work into time-boxed sprints with defined ceremonies. A Gantt chart shows a planned timeline with dependencies, which suits construction-style projects, whereas kanban shows the current state of work and pulls new tasks as capacity opens. Research with unpredictable experiment outcomes usually fits kanban better than either alternative. ### How many columns should a lab kanban board have? Five to eight columns is the workable range for most labs. A common starting set is Backlog, Planned, In preparation, Running, Analysis, and Done, with Blocked handled as a labeled state rather than a separate stage. If you cannot read the whole board in five seconds, you have too many columns. ### What WIP limit should a lab start with? Start with one card per person per active column, and set the Running column limit to the true concurrent capacity of your equipment and staff. A four-person lab with two incubators and one shared plate reader might start with limits of 4 in preparation, 4 running, and 2 in analysis, then adjust based on measured cycle time. ### Can kanban work for a single researcher or a two-person lab? Yes. A single researcher can run a personal board with three columns (To do, Doing, Done) and a hard WIP limit of two or three in Doing, which prevents the common pattern of five half-finished experiments. Two-person labs benefit the most from a shared board because it replaces status conversations entirely. ### Should a lab use a physical board or digital software? ![Decision path asking whether the team is in one building, whether metrics are needed, and whether cards must link to files.](/article-images/illustrations/kanban-for-laboratories-physical-or-digital-board.webp) Use a physical board if everyone works in the same building, because it is faster to update, impossible to ignore, and visible during conversations. Use digital software when team members are remote, when you need automated cycle time metrics, or when you must link cards to files and protocols. Many labs run both, with the digital board as the system of record. ### How do I represent a month-long experiment on the board? Give long experiments their own card type and swimlane, and move them through the same columns but with sub-state labels on the card, for example "Day 3 of 14" or "In wash step". Long-running cards should have a defined daily or weekly checkpoint written on the card so that "Running" does not become a place where cards sit untouched for a month. ### How do I know whether the board is actually working? Track two numbers, average cycle time and weekly throughput per card type, and compare them month over month against your first-month baseline. Secondary signals that the board is working include a shrinking Blocked column, cards that accurately reflect reality at the daily review, and lab members who update the board without being asked.