 Three months into my first job out of college. I was working on a broken build one night in the office all by myself at 2 a.m. and not getting anywhere and seriously questioning the career choice I had made And I was tired. I was making mistakes and so I ended up. I just went home. I went home Got an insufficient amount of sleep came back in the next day to try and track down dependencies Is a very bad day at at our Job at Boeing with Boeing had missed its commitment with the United States Air Force on that project and You know fast-forward ten years later three different jobs later, but we're still You know some things are better, but some things are still the same right at least now Configuration management was cool and people willingly checked their code and to source control without me having to bug them to do So and I remember that connect software came out with a tool That allowed it was a beautiful thing. We were able to automate our builds and and deployments But yet there we were at 2 a.m. In the morning on release night dealing with sequel replication issues and wondering what version of that DLL we were supposed to use and You know ten years of sleep Deprivation takes its toll and so my heart goes out to a lot of people in that situation There's got to be a better way to do it in 2006 our team started using Kanban and it was the first time in my IT career That we finally found a way to deal with everything that was on my plate and still get a decent night's sleep and so You know for those of you, I know there's some out there who don't need any sleep Just just swap that out for something that you would like to get more of The point is it's really it's just frustrating and it's very hard for people who are Overburdened to have the time to take pride in their work and that that's a problem Right people should be able to have take pride in their work and not be so frustrated all the time I think the cycle of frustration is pervasive It's it's I see it everywhere right customers Want their you know, they don't know why it's taking so long for them to get their simple little feature They requested a few months ago Ops teams feel like there's just too much workload Upstream teams wonder where their requests went right there's not enough transparency And there's just this radical and constant shift in prioritizations or Sometimes there's a battle between all the parties and everything is a priority one Who who here's dealing with this? Right we got a couple brave souls raising their hands here about five to ten percent of the audience When everything is a priority one Context switching raises its ugly head right and context switching can be very expensive and irritating It's irritating to get interrupted when you're in deep thought trying to solve us a problem trying to find a solution Excuse me. We need access to this database on this server You know it takes time to get back into what you were doing It takes longer for things to get delivered when you have context switching So now we've got overburdened people who are irritated It's not it's not a good situation So what to do about it? I find that if we can if we can find a way to demonstrate to the people up the food chain Making the decisions on prioritization We at least have a shot at Negotiating how that prioritization process so work Sometimes just bringing visibility to the work is enough to help people understand and Making work visible. It's the first of six Kanban practices. I'm going to run through today Make work visible If you make work visible at least people can see what's going on right? This is knowledge work made visible and we do this for a number of reasons listed here You know, it's hard to manage invisible work when you make work Visual You know we can we can take in information much faster through our eyes than any other sense And you can quickly glance at this visual and you can see okay Well, there's three projects going on and we've got some red You know emergencies floating through the the top lane there the top expedite swim lane Those are live issues that maybe the knock is dealing with or there It may be some kind of security breach where we had all hands-on deck We're bringing visibility to the different types of work that we're dealing with I've got this Project lane where all the projects are flowing through and then we've got a miscellaneous swim lane Where we've got some gray tickets and purple tickets and the great tickets are they're like maintenance They're like CODB cost of doing business We just have a lot of that Purple tickets use your imagination. I like to think of those as R&D research and development or whatever it is You want to track? For those of you that are colorblind in the room And it's about 10% of the male population in the US So there's probably at least five or six of you out there that are colorblind We do have ways to address that by using different shapes or different patterns or just decide not to use red and green Since those are the colors that tend to be most confused The the point here is that Or somebody made the point earlier today Jeff Patton about data does not equal empathy But when you can visualize data in a certain way then it can help bring about an understanding You can see where things start piling up right this validate Q toward the end before close their things are piling up and so that can signal a problem to us This is just a sample board right Kanban is you know, we're not prescriptively saying this is how your design should be This is just a board that worked quite well at a particular customer site And I always get the question about what physical board or electronic board What are some of the electronic tools we can use and I find it's great If you can use both if you can have both a physical board that you can You know feel and touch and move about and be creative with no constraints It's great But if your team has a high volume of ticket flow It might it doesn't make sense to put every single little ticket up on your board But maybe you could do a temporary board and put up those things that are irritating the team like all this work that's Piling up at the end you know where we have pile ups we have bottlenecks when Upstreams team when upstream teams Increase their throughput and the downstream teams aren't able to consume at You know what happens what good does it do if Development is delivering features faster and faster When the teams downstream can't consume it at that pace We can only move as fast as the slowest moving piece And that's why we have work in progress limits with Kanban, which is a second practice that we use We're putting Numbers here to say here's where our limit should be The standard way based on the three of constraints would be to limit the flow To the tempo of the slowest moving bit the constraint that's in the system because that's the fastest that we can Possibly go right and so where did it began? Where do we start? What is that magic number? I find the best way to start is just start where you are How much work do you have flowing through your system right now? Put that number up there make it visible and then experiment is that right? Is that too much are people overburdened? Probably then maybe we need to bring those numbers down So one way that you can have a bit of a bargaining chip With the teams that you're working with is to just demonstrate how much work you actually have in progress There You know because knowledge work is perishable The longer it takes from the time you have to set something aside because you got interrupted The longer it takes then to pick that back up and work on it. You might have have forgotten It'll just take time to to dig back into that Now Here's the deal about work in progress limits. They set they Bring in tension right work in progress limits introduce tension into the organization That you have to deal with it's not comfortable It's like the stress that Adrienne was talking about this morning You add stress into your system so that you can see where it breaks the whip limits introduce Tension and they help provoke the kind of conversations that need to be had and they may be uncomfortable But if it was easy anybody could do it, right? You know, we don't let our computers get to a hundred percent capacity utilization Why do we let our people? This is not sustainable When you're working day in and day out and at two in the morning It doesn't take very long before you get burnt out and people leave So we're trying to find a sustainable way a sustainable flow that we can keep up with the pace We've got two choices. We can either You know push back on the demand or we can increase the capability of the team Which is probably going to take some time, right? We need to build up that skill set knowledge Maybe we need to hire some more people, you know, what tools are they using? What's the capacity of the team to do the work? There's a lot of Attributes that make up the team's capability and as soon as one of them leaves or gets hired on then your capability probably goes down So what to do? How do we get a sustainable approach? What are some rules that we're using that are hindering us or helping us? And that's the third Kanban practice is to make process policies explicit Or this is saying what are the rules of the game? What are the rules at your organization that's either driving fast-flow or preventing it? And some rules if they're a tacit or a tribal You know, they're not well understood at least to new people. So let's look at some of these some of these rules So if you've got If you're trying to go in a certain direction and you have to stop in order to get there It's going to decrease your flow If the only way to get somewhere if the only way to turn left is to stop and wait for a green light Then you're gonna have this stop and go Things will take longer and people get irritated So we need to look at what are some other options besides this stop and go I think the roundabout is a nice example of that because they've they found a way to keep things flowing Right, they've minimized stop and go. You don't have to wait. There's no lights. It's cheaper throughputs increased by 20% because they measured it and measurement is The fourth practice, but first just an example of some rules here on this count-bam board first of all in the Done column here. We'll just walk through this real quick. There's a backlog Then they put work on deck the backlog by the way, this work isn't prioritized. It's not stack ranked Right. It's just a list of uncommitted Unprioritized items that somebody wants to get done. It's not committed to it's not prioritized Until it's pulled into the on-deck queue Combine by the way is a mostly a pole system. We're pulling work into our queue based on our capacity or the team's Capacity to handle the work if there's no capacity then the work doesn't get pulled in It's like only take you know only take on what you can chew, right? Finish. Let's let's focus on the end here focus on finishing work before pulling in new work And then they've got some rules here Back up. They've got some rules down here. What's the definition of done on some of these their definition of done? Right down here They need you know Some work the criteria is they got a cab approval first or maybe it needs an architectural review or Info sec needs to buy off on it Then the first rule is stop hiding the rules Get them out there where people can see them make them aware of what the rules are It's much easier to have a conversation about changing a rule or improving it if it's explicit If people are unsure it's not explicit, then it's harder to improve it But what about prioritization? That seems to last year was I was speaking about the problems that ops teams face as far as dependencies and distributed teams and Interrupt driven work, but I'm finding now that probably one of the biggest issues is prioritization Conflicting priorities causing a lot of heartburn. So what's the priority? What is the prioritization process? Is that explicit? How do things get escalated? What is that process? Is it just by whoever yells the loudest or is it based off of political weight? Right where your bot somebody's boss tells your director and then it comes over it down and then things get escalated so Come up with a policy for prioritization and escalation I want to point out this box in the corner here These were you know, we have these red Expedites that are going through the top lane and projects going through the middle lane and these items in the box here It's a spot to put items that need further analysis. They need further retrospection They got delivered and they're done, but there were problems with them and we want to address them later So we put them on a spot on the board there for us to peek at later What if one of the rules was that work got prioritized by cost of delay? right in in finding a way to bring some non Emotional logical quantitative reasoning to how we should prioritize and a guy who's been doing a lot of work on this is Troy McGonis from Focused objective. He uses Monte Carlo simulation to Determine different timings and and forecasting and I think this is a great start for ops Right, what didn't go out or get done that could have or should have right? We can calculate the cost of delay per project per day, but then in the ops world we also need to take account of keeping production stable and secure Right that we have more variability there. There's a higher risk. So that needs to be factored into this equation What is the probability of getting a DDoS attack or some security breach? And how much would that cost the the company? If you're tracking those kind of issues in your system if you're tracking expedites Then you can measure them fourth Kanban practice We're measured. We're very interested in measuring the flow of work that goes from beginning to end and the individual amounts of cycle time so always a Always a great conversation. What is lead time? What is cycle time? I'm just going to simplify it by saying lead time is when the customer requested something to when they got it and Cycle time is how long it takes once you've Committed to do that work, but you can calculate the cycle time for each cue on your Kanban board and Then that's going to provide you some historical data To use for forecasting and you won't have to rely so much on guessing or estimating how long things are taking Aging reports are fabulous You if you're measuring cycle time You'll have you know, you can get an average and a median and a mean But the aging report is is just it's a huge eye-opener, right? if you've got things and this is real data, right things sitting in Implementation cues for over a hundred and eighty days If you can pull up how long or it's like rotten fruit, right? the longer it sits the you know the costlier it is the The more frustrated people get right so think about aging reports bringing attention to some of these dependencies and Then the rate of the flow This is a good one and you've already got it if you're using JIRA You can see the rate of incoming work versus the rate of outgoing work And if you're incoming work, which would be in the red or brown if that slope is Exploding out Instead of closing up with the green green is the work. That's closing You'll be able to see really quickly in a one-shot glance if your team has a lot much more demand than they're able to deal with bottom line is Looking at being objective at the data, right love couldn't resist this slide You know, how much time do we really spend waiting on third-party vendors? Let's be realistic here Next up is feedback loops and offering up opportunities to be heard frequently and often kind of like merge early and often Now one thing we've had a lot of success with lately is lean coffee. Anybody go to lean coffees Got a couple lean coffees here. Lean coffee real quickly is an opportunity to have a bit of a more intimate session with five or six people you go grab coffee you get out of the Office and you sit down and everybody writes a couple things on a post-it note that they have energy to talk about and Then people you get three votes People can vote all three on one or vote three across different items And then you run them through a Kanban just real simple to do doing and done and you time box the items and to do in doing so after five minutes the Beeper goes off and you thumbs up or thumbs down and you the team decides if you want to continue the conversation on But we can take these and we can start to see patterns week after week after week and you can It gives the team an opportunity to feel like they've been heard you can rotate people in and out to keep it a small group Fantastic way to get feedback that you wouldn't get otherwise unless you know you go out for beers after work It's impossible Sometimes to discover these kinds of issues at larger meetings like ops review But we still need to do ops review Ops review is a monthly where where link coffee would be weekly ops review is monthly And this is where we're gathering everybody in the organization And we're looking at the data and the team leads are presenting That basically they're presenting the balance that their team is dealing with the demand That's coming on their team and then their ability their capability to meet that demand That's what we're interested And everybody here is the same message at the same time This is actually a picture of the first Solvay conference in 1911 in Brussels And I kind of liked it because the subject was radiation and the quanta and Einstein is the youngest Physicist he's second from the right When we're looking at what kind of data to measure and track and present at ops review Here you go these these five things across trend and across the variability Will be quite useful and the the top three work in progress cycle time or lead time and throughput You can get all of those out of a cumulative flow diagram, which is one of the most used con bond Charts and many of the electronic tools will automatically generate those for you But if not you can do it using Excel So there we have six combat practices all on one sheet Click review. So most importantly make work visible because it's hard to manage invisible work and oftentimes You know if it is out there on a wall in front of people At least they'll see it especially especially if it's on the way to the coffee or on the way to the VP's desk The last place I was at the VP doesn't have a lot of time to go look at stats in Jira Right, but if you put it on the way to his desk, it's like covert or Covert con bond right or stealth con bond If you can get that if you can design your con bond system in such a way that it's revealing what is irritating to the people doing the work and Irritating to the customers that the work gets delivered to Now you've got a way to start having that conversation That needs that needs to be had and to find some of that empathy that Jeff was talking about this morning number six is Collaborate an experiment with the with models and theory and so here's a list of reading material that the people in the con bond community often Talk about the first is the goal This is a story about the theory of constraints, which is the study of bottlenecks famous book that Gene Kim's Phoenix project I guess you all got a copy of that Anybody who's wouldn't working and ops for some time can relate to that Lean startup Just you know Eric talks about having this This meeting where you pivot right you decide if you're gonna pivot or not And I liken that to the ops review meeting where you're looking at the data to see if you need to go in that direction or change And then Deming Deming's the guy who went to Japan in the 50s and taught all these Japanese businessmen about why Why organizing teams in functional silos does not work because we don't deliver work that way we deliver work across different functions That fabulous fabulous read there and then down on the bottom right David Anderson's book on con bond It's for the leading how to do book on con bond And then the commitment books a new book out It's about real options theory and looking at options and more of a business language use if we can learn to speak in the language of the business and Have them understand that options expire quite useful. It's a graphic Business novel and then David Reiner since flow book So if you want to get into the deep math on queuing theory to understand Why using cues is helpful, you know con bonds based on cues, right? Not so much on timeline. We manage work through cues It's a leading cues are a leading indicator because we know the more work That's in the queue the longer things are going to take great read there and then Troy Troy McGinnis's book on using Monte Carlo simulation to do Forecasting and cost of delay so just Ending up with grumpy cat here I mean sometimes I said earlier that con bond gave me a way to deal with everything on my plate and still get a decent night's sleep Dealing with everything on my plate doesn't mean having to say yes to everything We can only do so much and sometimes we have to push back and say no the team does not have capacity to deal with that right now and And and and that's okay people don't expect you to do everything, right? I mean you can't you can't do a million things at the same time It's just final slide here and picture of the Con bond for ops game that we've been playing Okay, I think we're out of time Okay, okay So we have time for a couple of questions if you want to come forward just line up behind the mic and we'll just take them serially so one question that I had is you talk about queuing for kind of the system of work and a Lot of the work that I have The systems do is also very cute So is there some kind of like grand unified theory of con bond where I can get those to work together and kind of a feedback loop Are you talking about having? Ration my prioritization But everything just having you know kind of a conceptual framework that that touches you know both on the the flow that we do Well, you know the flow that flows through the systems. That's all cute and you know limited to work in progress It's I see them as the same things I mean it sounds like it is it a question of tools. Yeah, just yeah Tools and you know concepts and you know trying to apply the same thing Yeah, so one cool tool is it's called They didn't pay me to say this or anything, but it's Jim flow and So it's electronic tool, but yet you have a board to it's what prints out the tickets It's got a QR code on it. Are you familiar with this now? You can put it up on your board you put a time-lapse camera in front of it And as people are pushing tickets across the board it takes pictures and automatically updates the the tool So it's the best of both worlds so you can watch the video of it at you know, Jim flow I am J-I-M-F-L-O-W I think them. I think the message that lesson is What's the what's the most irritating thing you guys are dealing with? How do we bring some visibility to that so we can improve it? Yeah, okay? Yeah We got one more okay Two two quick ones In your in your side about there we go In your side about quantitative measures you had quality listed as one of those items And I was wondering if you had time to expand on that. How can can then help us measure quality? Yeah, that's a really hard one So a couple things is rework How many times a ticket comes back well you or if they're getting closed and then they're getting reopened Why is that did we not understand something did you know? Did we miss communicate? So rework is is hard and then also I think looking at the number of expedites is a telltale sign Right if if 63% of the work we're doing is expedites does that sound right? I mean it could be that's that's how your team works, you know If you're at help desk or your call center That's the nature of the demand of your work and that's to be expected But if that's not your situation and that seems really high Then measuring the number of expedites because like that could be an indication that Previous work that was done wasn't done at the level of quality Yeah, I mean it doesn't tell you why may not tell you why right you have to dig into sure why I can see why that Is does our architecture not allow us to to work in this fashion? Okay, thanks. Okay. Thank you Hi, so it's curious You know I work in an ops team and Like a lot of ops teams we you know, we've been looking at a variety of different ways to try to implement, you know Limits of work in progress or like pull models or whatnot but our organization insists on us doing like long-term forecasting and and Committing to piles of work like you know one to two quarters ahead of time Which and we know that doesn't really work, right? So do you have any advice on how to like how to how to mitigate that like inside an organization? Yeah, are you tracking that now? Are you tracking the work? That's not happening two quarters later. Oh, yeah Yeah, we're tracking it a lot better now, so it sounds like you probably have the data there to make a case on Something that's not working that you can demonstrate maybe out of ops review To we just demonstrate here's what happened. We did it, you know, we missed our commitment Here's why It could be though that I mean cost of delay isn't always financial sometimes it's right political and if that's the the problem then you've got a I found that just trying to I Found the best way to get empathy empathy is to show the data in a non-emotional way because Ranting I couldn't get ranting to work, right? But I mean we're still kind of Left with this situation where we're being forced to commit to a pile of work that then we can't complete and then we're forced to Play kick the can and it just seems that we're constantly trying to catch up to that instead of getting yeah And so I mean Deming's book there's something that's driving that behavior What what is that and that's I think what you need to bring visibility to tell me how you're gonna measure me And I'll tell you how I'm gonna behave and if there's You know if people are being measured or their merit reviews or they've just come on board and they only have 90 days Or I don't know what your situation was. I'd love to talk to you afterwards to hear more but I hear I hear that a lot and I think that if you can show and Demonstrate well actual data that has not worked Yeah, for us it's becoming a barrier to getting the people to adopt these tools because they can't wrap their heads around The idea that we would not commit to getting anything. Yeah Yeah Rather than just focus to bring him to bring him to the conference. Thank you. Thank you Thank you everyone for attending. We have a 15 minute break and we have our next talk at noon