Pawel Brodzinski on Leadership in Technology

Whatever it takes to lead a team, build a product, and run a business

Category: kanban

  • Agile Isn’t Immune for People Who Don’t Give a Damn

    I like presenting. I just love the fact that every time I deliver the presentation I find something new in there. It’s like having tiny epiphanies while you’re talking to a crowd. And I’m sure I wouldn’t have them if I didn’t overcome my fears and didn’t start speaking publicly.

    A thing which hit me last time I was delivering my Kanban Story presentation was how fragile agile is in terms of people who don’t give a damn. In the summary of presentation I make a point that Kanban may not be for your team if you work with undisciplined engineers.

    But what I realized while I was talking it is not only about lack of discipline. It’s about care. In Kanban people have to care about Kanban board. They have to move cards as they progress with features. They can’t abuse limits unless it is agreed among the team and team rules allow to. They should think how they could improve the board. They should keep the board up to date.

    Why it is so important? Because the board is single most important information radiator in Kanban teams. If contents of Kanban board are random you spread random information about project among the team and outside the team. You pretty much misinform everyone around. And now the important thing:

    In Kanban teams anyone can easily abuse Kanban board.

    It’s enough someone doesn’t care. He doesn’t care to move cards through the board or put his marker on task he’s working on at the moment or ignore transition criteria or whatever. Suddenly you have random Kanban board and no systemic tools to deal with the issue. People versus Kanban 1:0 at halftime. If you want to know how the match ends, well, you’ll have to read the blog in future since it is a subject for another post.

    It’s already half of A4 page and I haven’t really mentioned agile in general. Only Kanban. Kanban this and Kanban that. Don’t I have something else to say? Well, I do, thanks for bearing with me so far.

    One of reasons why Kanban is so easily abused by people who don’t give a damn is the fact that Kanban prescribes close to nothing. But enough of K-word. Scrum is more formal. Actually Scrum is pretty formal approach if you ask me. How it deals with the problem?

    A bit better. Stand-ups, time-boxing and planning meetings are all practices which help to deal with undisciplined team members. (I wanted to write “engineers” instead of “team members” but I’ve just recalled there are no roles in Scrum besides Scrum Master, Product Owner and Team Member and I prefer not to be killed by some Scrum extremist.) With these practices it is easier to make someone stick to convert folks who don’t care. If you work with Scrum you can’t work without time-boxing while the rest of the team has 2-week long sprints. You are asked the same questions as everyone else during stand-ups. There is more of a stick in Scrum than in Kanban. (Have I just said K-word?)

    XP goes even further. Rules are surrounding you everywhere, my poor engineer. It almost tells you how to pee. Not in pairs. fortunately If you work with XP and you don’t give a damn you’re probably already fired.

    But let’s face it – how many teams out there use eXtreme Programming religiously? Few. More of them follow just a couple of engineering techniques choosing Scrum as their process base. And even then we have whole variety of Scrumbuts, Scrumbans and Scrumwhatevers.

    Average agile approach in real world, if there happen to be something like that, would cover a very limited number of rules. I’d say that there would be less of them than Scrum proposes. And even then rules are often softened or teams choose those which are less painful to adapt.

    So we come back to the first point. The fewer constraints (aka enforced practices) we have in place the more easily our process can be abused by people who don’t give a damn. If your method base on a lot of common understanding instead of tons of process documentations and a number of strict rules buying all people in becomes crucial. Having few rules means that you give people power to create and adjust your software development/project management approach.

    It also means you give them power to destroy and harm it.

    If you like this post you should thank Michal who took the discussion up on Twitter.

  • When Kanban is the Best Choice

    Kanban, as any other methodology, isn’t a silver bullet. There are situations and teams when it shows its full potential but there are others where its impact will be limited. Where Kanban suits best then?

    Micro-sized teams

    It is said Scrum works best with teams of 7 or close to this size. Sometimes we deal with smaller groups. 3 people working on a project isn’t something very uncommon. For such micro-sized teams Scrum is often too formalized. You can limit a number of rules you follow and still keep good quality. And you have a bit more time to do the real work.

    Frequent priority changes

    “Walking on water and developing software from specifications are easy if both are frozen.” Unfortunately we deal with a lot of changes as we build software. There are new features; importance of tasks changes, new top priority bugs requires instant attention. Our response is moving to more flexible approaches. We try to avoid BDUF projects. We switch to agile methods employing short iterations. We even make iterations shorter and shorter. It allows us to change priorities frequently. Once every couple of weeks if we take typical Scrum implementation.

    The problem is when priorities happen to change even more frequently. Once every few days. Or even every single day. And yes, there are such projects. Kanban is a great answer for them. Feel free to change priorities every day. As long as it is well-grounded it shouldn’t ruin your project.

    Maintenance projects

    A typical maintenance project routine looks like this:

    1. Whenever high-priority bug is submitted fix it as soon as possible
    2. Low-priority bugs becomes high-priority ones when resolution deadline approaches; then see above
    3. Whenever a client orders change request (CR) and there’s no high-priority bugs – try to do it as soon as possible
    4. If there are more than single concurrent CR ask project manager about priorities
    5. If there are no bugs or CRs do some refactoring or other improvement job

    Sounds like ideal Kanban playground, doesn’t it? That’s typical case of event-driven development (not event-driven programming) where you don’t actually have a roadmap or something but you do whatever new day brings. After all you don’t expect to have a bug submitted tomorrow, or do you?

    Multiple small projects

    Working on several rather small projects or sub-projects with the same team at the same time is pretty difficult. Resources (what a nice name for your people) are usually insufficient since it is harder to synchronize stream of small orders to keep it at the same level all the time and bringing more people just-in-case isn’t the best business strategy around. This ends up with (surprise, surprise) a lot of priority changes and trade-off games. “We can complete this additional project on time but we’d fail to meet a deadline here or there.”

    With standard structured project management approaches coordinating different threads with ever-changing priorities becomes pretty much a hell. What Kanban does is it organizes workflow so the main, well almost the only, thing you should care about is setting priorities at the beginning of workflow.

    Common part

    The common part for all of environments above is they don’t require many constraints to work. Few simple rules which come with Kanban should be enough to get things done. Another common thing is mid- and long-term planning is hard or even close to impossible, which is another problem hardly resolvable with more structured approaches. These two things are the most specific for environments where Kanban shows its full potential.

    This isn’t really a post which is a part of the Kanban Story but if you found it interesting you should like the story as well.

  • The Kanban Story: Swarming

    If you’re into Kanban you probably have heard the term swarming. Actually chances are good you’ve heard the term despite your lack of interest in Kanban.

    What is swarming?

    In short swarming is all about getting more people to do the task than you’d get otherwise, in normal situation. An example: you have a bug which normally would be fixed by a single developer, but for some reason you have the whole team working on it. And you already have another question…

    Why, oh why?

    Why would you want people to swarm around a small task? Typically teams do this when they face either a blocker, which hampers whole team, or a very difficult problem, which is hardly solvable by a single person. Either way you get more people to move things further, which would be impossible or at least very difficult otherwise.

    Swarming in Kanban, part 1

    Up to this point swarming is something teams do intuitively at times. The reason why the term is mentioned so often in the context of Kanban is one of Kanban rules: limiting work in progress. Because in Kanban we want to have as few ongoing tasks as possible, yet as much as it still does make sense, we have limits for different stages of our process. And it when we hit the limit we treat it as a problem which we want to solve and we want to solve it fast. That’s why we swarm.

    Swarming in Kanban, part 2

    There is another thing which you may read about, which I call everyday swarming. Everyday swarming is done by teams which swarm by default, when no emergency situation happens. The reason standing behind this behavior is trying to make cycle time and/or lead time as short as possible thus flow of tasks very rapid and smooth. Oh, and you have fewer ongoing tasks then too, which means you have less work in progress, which means you have less waste, which means you are oh so lean.

    Our story

    I must admit I don’t really like the idea of everyday swarming. I mean cycle time is definitely shorter but I’m so sure about throughput. Flow is smoother but you kick in a lot of micro-switches when a few people need to integrate their work into one piece of code since they all have worked on a single feature. You also may invite some issues with version control this way since you’ll be working on the same chunks of code. For me the net weight of swarming by default is negative.

    Basically that’s why we avoid swarming whenever possible. When there is some serious problem to solve and standard approach is not enough we naturally help each other. One person after another joins the rescue team and this basically ends as swarming. But other than that we prefer to have only one person working on the task at the same time. Well, of course there are exceptions like discussing design or testing and bug-seeking/bug-fixing but these all are pretty natural things for any software development team. After all it would be pretty schizophrenic to discuss the design with oneself.

    It looks like we swarm as David Anderson initially proposed, despite ignoring David’s teachings when it comes to lead time. I think this is the most natural way of swarming. But of course you remember the rule to experiment like crazy, so you may want to try something different in your team.

    Read the rest of Kanban Story as well.

    http://blog.brodzinski.com/2009/10/kanban-story.html

  • The Kanban Story: Coarse-Grained Estimation

    Recently I told you what a screwed way we chose to measure lead time in our team. Unfortunately I’ve also promised to share some insight how we use lead times to get some (hopefully reliable) estimates. So here it goes.

    Simplifying things a bit (but only a bit) we measure development and deployment time and call it lead time (and don’t feel bad at all about that). So how do I answer to our customers when they ask me when something will be ready?

    That’s pretty easy. If we talk about a single feature and there is no other absolutely top priority task, I take a look at the board to track down a developer who will be first to complete his current task and discuss with him when he expects to finish it. Then I know when we can start working on the new, super-important feature ordered by the client. Then it’s enough to add two weeks (which is our average lead time) and make some effort to look oh, so tired as I’ve just completed such a difficult task of estimation. Oh, I need to tell the customer when they’re going to get the darn feature too, of course.

    This however happens pretty rarely. We try to keep our MMFs (Minimal Marketable Features) stick to their name which means they are usually small. This also means that described situations, when client wants just one feature from us, are pretty much non-existent. That’s why you might have noticed I didn’t take into consideration a size of a feature in described scenario. In the real life we usually talk about bigger pieces of functionality. What we do then is we split the scope into small MMFs and then use two magic parameters to get our result.

    One of parameters you already know – it is an average lead time. However there’s one more needed. It takes 13 days for the feature to get from the left to the right side of the board, but how many features are on the board at the same time? I took a crystal orb and it told me that on average we have 4,5 MMFs on the board. OK, I actually did a simple analysis of a longer period of time checking our completed MMFs and I got this result, but doesn’t a crystal orb sound so much better?

    Now, the trick I call math. On average we can do 4,5 MMFs in 13 days. Since I have, let’s say, 10 features on my plate I have to queue them. If I worked with iterations it would take 2,2 iterations (10/4,5) which basically means 3 iterations. But since we don’t use time-boxing I can use 2,2 and multiply it by 13 days of my lead time and get something about 30 days. Now I don’t start with empty board so I have to add some time to allow my team to finish current tasks. On average it would be half of lead time so we should be ready in, say, 37 days.

    And yes, this is rough estimate so I’d probably go with 40 days to avoid being blamed for delivering estimate which looked precise (37 looks very precise) but was just coarse-grained estimate.

    That’s basically what I went through before telling my CEO we’re going to complete management site for the product we work on in 3 months. Well, actually I didn’t told him yet, but 3 months there are. No less, no more.

    Although this post may look as it was loosely coupled with the Kanban Story it is definitely belongs there. So go, read the whole story.

  • The Kanban Story: Measuring Lead Time

    During AgileCE conference I had a discussion with Robert Dempsey about measuring lead time. I never really thought much about the way we count lead time in our team and talk with Robert triggered some doubts.

    What We Measure

    As you already know on our Kanban board we have backlog, todo queue, several steps describing our development and deployment process and finally done station.

    OK, so when do we stamp the starting date for the feature? We do it when the card goes from todo queue into design station, which is the very first column in our development process.

    When do we stamp ending date then? You may take your best guess and um… you’ll be wrong. No, not when the sticky note is moved to done column. Actually we mark ending date when a feature makes its way to live column which is third station from the right.

    And this is different from what you may have heard from Kanban gurus out there. Don’t blame me, I’ve already told you thought-leaders don’t know it all.

    What we measure is time which passes before we start actual work on feature to the moment it goes live.

    What We Don’t Measure

    What is left behind then? First and the most important thing is the time feature spends in todo queue waiting for some developer to become free and start working on it. If you were trained by the father of Kanban – David Anderson – you’ve probably heard something different but stay with me, I have a good explanation. Or so I guess.

    Another thing which is left outside is the last part of our process. There is documentation station where we (surprise, surprise) update documentation. This is done after pushing new version to production.

    It looks like we cut something on both ends to make our lead times look better, doesn’t it? Anyway what we gather as lead time doesn’t really describe time which passes from the moment of decision to build the feature to the moment it is done-done. Thus another question arises.

    Why, Oh Why?

    The way we measure lead time came in natural way but chat with Robert forced me to make some explanation up to justify our practice.

    Time Spent in Todo Queue

    We left time spent in todo queue out basically because content of this station is changing pretty often. Sometimes feature lives there just for a day or so just to go back to backlog when priorities change. And believe me they do change every now and then. Sometimes a feature stays in todo queue for a longer time as it is always pushed to the second or third place because, well, priorities change.

    There is another reason too. The basic reasoning for adding time spent in todo queue to lead time is that you should be able to tell your customer how long it would take from the day 0 (when they tell you they want the feature) to the moment they get it in production. It is pretty rare case when developers are able to start working on a new feature immediately so it is natural that some delay will appear when feature is waiting for a free developer.

    I’m not convinced though. Actually very much depends on additional circumstances. The feature the client is asking for may be the only high priority thing to do at the moment but it is pretty unlikely. If the client asks just for one feature lead time will be different than if they asked about a list of ten features. If you have limit of 3 in todo queue you would need to put 7 features in backlog anyway and time they spend in backlog won’t be measured.

    If you have a few clients and you need to balance your workload among them it becomes even more complicated since product owner (or whoever is your priority setter) has to decide which feature can be put on the top of the todo queue, which can occupy second or third place and which has to go into backlog.

    Basically in any situation but the simplest measuring time spent in todo queue wouldn’t help us much and that’s why we decided to exclude it from lead time.

    Time Spent on Documentation

    With documentation the situation is a bit different. Documentation isn’t a real part of our product. We create it for our internal reasons – to make life of our sys admin easier and to avoid problems if he was hit by the bus. This means that, from client perspective, our product is complete as soon as it goes live. Even though we have to do some housekeeping later it doesn’t affect our ability to deliver.

    Theoretically it might be a problem if documentation phase would be time-consuming and would steal time we’d prefer to spend for something else, namely deployment or testing. However looking at the past in the team it isn’t a problem so we may safely throw documentation out of lead time.

    The next thing is using lead time to create some estimates but that’s the subject for another post.

    If you liked this post you should like the whole Kanban Story too.

  • Why We Don’t Write Test Cases Anymore

    Almost a year ago I shared an advice to use test cases. Not because they are crucial during testing or dramatically improve the quality of the product (they are not), but because of value you get when you create test cases.

    A confession (and yes, you’d guess it anyway if you read the title): we don’t write test cases anymore.

    We stopped using them and the reason isn’t on the list of shortcomings of test cases. Actually I was aware of these shortcomings a year ago and I were all “test cases are great” anyway. What have changed then?

    We dropped test cases as a side effect of implementing Kanban, but you can perfectly use both if you like. In our case one of effects of switching to Kanban was making our pieces of functionality pushed (pulled actually) to development smaller. Before the switch we had pretty big features which were split into several (8-15) detailed user stories. After the switch we have much smaller features which would make 2 or 3 detailed user stories if we didn’t drop writing user stories at all.

    And the reason for making features smaller was simple – smaller features, smoother and more flexible workflow.

    Initially we were connecting test cases to features, not user stories. It was so because pretty often one testing flow was going through a few different user stories. I told you they were detailed. When standard feature-size went down we realized there’s much less value in preparing test cases.

    Once again: main value of creating test cases is thinking about specific usage scenarios, looking for places forgotten during design. The more complex feature the bigger are chances there is something screwed up. With small features test cases lost much of their value for us since most problems we were able to locate instantly and fix problems without additional technique. And yet the effort needed to create and maintain test cases was still significant. So we dropped the practice.

    It looks like my advice is: make your features smaller, and then you’ll be able to drop user stories and test cases. And I must say it does make sense, at least for me. What do you think?

  • The Kanban Story: Kanban Boosters

    During my talk at AgileCE I mentioned three things as biggest Kanban boosters in our case:

    One of comments I heard about this part was: “Pawel, these things aren’t Kanban-related – they would work in any environment.”

    Well, I’ve never said they’re exclusive for Kanban. My point is: Kanban is pretty simple approach – it really tells you to do just a few simple things leaving most of other aspects open. This means you may (and should) use some help from other techniques/practices which set additional constraints or organize other areas of your process. And boosters I mentioned work exactly this way.

    Co-location allows you to keep your Kanban process as simple as possible. You don’t need to use any software to visualize current state of the project – old-school, hardware whiteboard is enough. It also helps to exchange information between team members which is always important but with minimal formal information exchange in place it plays even more crucial role.

    No-meeting culture brings us “do it now” attitude and saves a lot of time usually wasted on meetings. And we discuss different things more often than we would otherwise because launching a discussion is totally easy: you just start talking.

    Best engineering practices help us to keep the code quality under control. That’s a thing which is basically omitted by Kanban itself so you need other tools to deal with it.

    Now, you can perfectly take your Scrum (or whatever kind of) team and use the same techniques and it would most likely work. Oh, it would as long as your team isn’t too big (co-location and no-meeting culture don’t scale up very well) and you don’t have a list of engineering practices enforced by your approach anyway (like in XP).

    So no, these concepts aren’t exclusive to Kanban. They just work in specific environments. Yours either fit or not. It doesn’t really matter if your process name starts with S or K or X or P or whichever letter sponsors your day.

    After all when I think about Kanban I don’t isolate the method from the rest of our environment – that would be stupid. If Kanban works for us it is because the whole setup is good enough, not just its Kanban part. And these boosters are a part of the story.

    Read the whole Kanban Story.

  • The Kanban Story: Kanban Board Revisited

    Kanban doesn’t prescribe much – if you follow The Kanban Story you already know that. Kanban Board which is a key tool in Kanban isn’t pre-designed or something. You need to tailor your own. That’s how we did with our Kanban Board.

    We decided to start with something simple and then adjust things as we go. We redesigned Kanban Board because it is meant to be changed. But even then it wasn’t the final shape.

    When we left the board last time it looked like that.


    What was the problem?

    On a general level we switched to more frequent deployments. At the beginning we didn’t have any system in production but it has changed. Sort of. Anyway our endpoint was no longer “ready to ship” – we wanted to see new features live.

    Before the change we were marking MMFs as “in production” by throwing sticky notes out. The problem was there was no limit for “ready to ship” station, so features could wait there on a huge stack waiting for deployment. Something rather opposite to what we’d like to see.

    The other thing which pinched me was documentation. We had a stage for this purpose but it appeared we were documenting after we’d pushed feature into production. This was perfectly fine since we don’t sell boxes with up-to-date documentation inside and it was no problem at all to have documentation updated a couple of days after deployment. The problem was our Kanban board didn’t correspond with this.

    After playing with some ideas we came with version 3.0 of our Kanban Board.


    What has changed?

    The left part hasn’t changed at all. We were happy with it.

    The first change was merging “test” and “ready to ship” columns into one under the name of the former. What we got was testing stage which was separated into ongoing and completed phases. When a feature was tested it was ready to ship. We no longer expected documentation to be updated before shipping since we weren’t working like this anyway.

    Another thing was adding “live” column which means feature was successfully deployed. I didn’t like the idea of adding another “deployment” station since deployment is very fast in our case.

    Documentation was left as the last but one column and at the end “done” station emerged. After all it is such a nice name for a column.

    Limits

    Redesigned part of the Kanban Board reflected our workflow much better but required also limits adjustments. Instead of rethinking whether we should add 1 here and subtract one there I wiped out all limits and rethought them from the beginning.

    Limits of 3 for “todo” and 4 for “development” had worked well so they got their way back to the board.

    The next one, and the hard part, was limit for “live” station. Remember our pseudo-iterations? Because of this approach we deploy a few features at once. So far there were two or three in each package. However it happened a couple of times that an emergency feature was added during a pseudo-iteration. Since this kind of feature was meant be included in the same deployment package this made a maximum of 4 MMFs possibly being shipped at once. That’s how “live” stage got its limit.

    The same reasoning applied to “test” station. Since it was possible to move 4 sticky notes at once to “live” column, they all had to be taken from the previous one, thus we couldn’t set there anything less than 4. We didn’t want to set anything more either, so the lucky number was exactly this: 4.

    “Documentation” got a limit of 2 since it’s often convenient to document 2 corresponding features at the same time.

    Looks like we get more and more columns over time. I just wonder whether that’s a good thing.

    If you liked the post you may want to read the whole Kanban Story.

  • The Kanban Story: Pseudo Iterations

    One of reasons why I like Kanban so much is because it doesn’t force you to formalize your process. You don’t need to set strict time-boxing for example. If you want to – fine do it, but if it doesn’t suit you fine no one forces you.

    If you happen to work in environment where priorities change very often, and we do work like that, you’re likely to reject time-boxing since this approach doesn’t reflect process you follow. We didn’t want time-boxing and this was one of main reasons we rejected Scrum initially.

    Iterations were bad. At least that’s how we thought.

    After some time I realized we basically had iterations. Sort of. We didn’t set time constraints like “we’re going to do features which fit 2-week period.” But we started grouping features into releases like “stories A, C, E and F which are top priority at the moment are going into the next release.” It was taking as much time as needed to complete a group of stories so it wasn’t time-boxing really but still a small and (mostly) closed set of work items to do.

    Why we started doing that?

    • To make releases easier to plan. As I saw sticky notes moving from the left to the right I could estimate when we were going to deploy new set of features in production environment.
    • To have more frequent releases. With no real release planning we ended up with not very frequent deployments. At least not frequent enough for me. When we added some planning we set some boundaries for our work and made it easier to deploy more frequently.
    • To make version control look less like hell. We don’t use distributed version control so every time we want to release we need a branch in our code repository. If we had a few branches actively used bugs would have to be fixed in each of them, so we’d like to avoid having plenty of active branches. If we had only one branch and we wanted to release only completed features, leaving those under development aside, it would be quite a challenge. Having small feature sets going into development and adding a branch every time we start new set brings an order to version control.

    So we ended up having pseudo iterations. Not as strict as in Scrum but still. I can’t say how long the next iteration would take but it most likely won’t take more than 3 sticky notes. I can’t promise I won’t add anything to the scope during the iteration but if it happened I had good reasons to do it (e.g. adjusting the app which would stop working after deployment to take latest case).

    Seems like iterations aren’t such a bad thing after all.

    Check out the whole Kanban Story.

  • The Kanban Story: Throwing Sticky Notes Out

    When you read about Kanban the thing which usually isn’t described in details is what happens with sticky notes which make their way to the right side of the board and are in the last column of your Kanban Board (however you call it).

    To be honest I didn’t put much thought in that when we were starting with Kanban. Our sticky notes were just stacking up in ready-to-ship column until we no longer could read them easily. And it was just a matter of time when they were going to come unstuck and fall being trashed by cleaner and forgotten forever.

    What we decided to do is to file a sticky note when a feature is actually shipped.

    We don’t ship every MMF separately although there are some who do that. We rather look for a complete set of features and ship them in a package. That’s why out “done” column was named “ready to ship” since it is one thing to have some feature ready and completely separated to implement it in production environment. At least it is so in our case.

    The key thing for our team is to have feature ready to deploy as soon as possible, not to have it live as soon as possible. It is so because due to a bunch of external dependencies (interoperability) deployment is pretty painful and rather long process which we definitely don’t want to do twice a week.

    However when we decide to ship something – there comes everything which is ready at the moment. This is the operation which empties our ready-to-ship column. I take all sticky notes which were in the column and add them to the stack stored in my drawer. After all we need some documentation, don’t we?

    Read the whole Kanban Story.