Integrate AI-powered automation for class of service selection

Are you using a free to-do list app or “Kanban”-style tool? If so, have you considered the added value of more feature-rich applications providing more complete support for the Kanban method? Let’s take a look at the differences between those basic to-do list tools and those that provide well-developed support for managing the flow of your work. The payback on using the latter type of tool might be huge.Features common to all toolsLet’s start by looking at the functionality that all these tools provide, from the most rudimentary to the most developed. They help you in two ways:they help you visualize your workthey automate certain administrative tasksBasic visualizationVisualizing work helps you gather the low-hanging fruit in your efforts to improve how you manage your work. Time and again, people exclaim the benefits of their kanban boards: “For the first time, we can see all the different tasks we are called upon to perform!” Seeing what you have done in the past, what you are working on now and what you shall have to do in the future helps you to make better decisions about managing your work. It helps you answer questions like, “What shall we do next?” “What work is not progressing as fast as it should?” “Who is doing what?” and so forth.Basic automationAutomating administrative tasks is great for freeing you to spend your time doing work that delivers value. You can automate the sending of alerts. You can automate the creation of new work items on your kanban board. You can automate the interfaces among your various tools. You can automate the completion of data fields when the work items are of standard types. The degree to which a tool provides automation is generally proportional to the types of functionality it offers. For example, a tool offering extensive analytics typically allows you to automate the creation and distribution of those analyses. Obviously, a tool lacking such analytics will also lack their automation.A gradient of featuresAlthough visualization and automation may be beneficial, they do not complete the story of how tools may increase value. While visualization is beneficial, it puts the burden of analyzing situations and making decisions on individuals. Too often, those decisions are made in a seat-of-the-pants or emotional way. Many psychological biases mitigate against choices leading to greater value.Automation is beneficial, especially because it helps ensure the consistency of data quality. Automation helps us ensure that many of the little tasks that we might otherwise complete in a lackadaisical, inconsistent way, are done in a timely and accurate way. In reality, though, the types of automation provided by kanban tools can only help us gain a very limited amount of time.Many of the so-called “kanban” tools stop at visualization and a little automation. At the opposite end of a gradient of features is what I call integral kanban tools (see Figure 1). These tools are integral because they support the broader set of kanban practices and principles. What are the key differences between the basic to-do list tools at the narrow end of the gradient and the integral kanban tools at the opposite end of the gradient?In addition to visibility and automation, integral kanban tools provide key functionality for an organization seeking to truly optimize the flow of its work. This functionality includes:analyticsbuffer managementportfolio managementgranular access controlsupport for improvement cyclesThere are certainly other types of functionality that an integral kanban tool might provide. Rather than make an endless list, I believe the above are among the most valuable.AnalyticsAnalytics is at the heart of flow optimization. They provide an objective means for answering questions all service providers and consumers ask. “How long is it likely to take to complete this work?” “What are the most important factors slowing us down?” “What can we do to make our work more predictable?”Of course, analytics on their own do not answer these questions. They must be combined with each team’s periodic activities to understand how its work progresses and take actions to improve that work. But without analytics, such decisions are often made by the most senior person in the room; or by the person who argues the most; or in response to the customer that complains the most. In short, analytics provides us with a factual basis for making incremental improvements leading to long-term, sustainable benefits. Lacking good analytics, we often make decisions destined to reduce immediate pain, thereby creating bad habits and long-term operational debt.An integral kanban tool will typically provide built-in functions to calculate lead time distributions; estimate the probability of completion of a task by a given deadline; identify the outliers in a team’s work history; simulate future work patterns if resources are changed; among others. The analyses are focused on delivering value to customers rather than the old-style analytics of push-systems focused on keeping people busy.Typically, kanban tool analytics describe cycle time distributions, work in progress quantities, work item blockages, throughput, demand levels and flow efficiency. These statistics may be broken down by value stream, by service, by class of service, by initiative, by organization and by work item type. Analytics also provides a basis for predicting the future flow of work within a given level of probability.Buffer ManagementIntroducing buffers (queues) into the flow of work often helps to smooth out that flow. Buffering recognizes that the tasks of knowledge workers vary considerably in effort and duration, even when those tasks are of the same type. As a result of this variability, the WIP limits for each phase of the value stream can only be approximate. Buffering between value stream phases helps reduce the starting and stopping of work.Furthermore, a buffer column provides a visible way of distinguishing between items in a given phase on which work is in progress and those items that have been completed in that phase. When a worker completes a work item, she slides the corresponding card into the following buffer column. When another worker is ready to start a new item in the following phase of the value stream, he picks the item from that buffer column (see Figure 2).There is always an initial buffer in the flow of work—the “Ready” or “Requested” column on the card board. This is the column in which you find work items to which their respective customers have committed but on which the team has not yet started work. Experts generally recommend that buffers be set up in response to particular issues that a team might have. Buffers for the sake of buffers add complexity without necessarily smoothing the flow.Figure 2: The third column, "Analysis Completed", is a buffer. The time spent in this column is non-value-added.Up to this point, almost any to-do list application or kanban tool can define buffer columns. Many of them can even define WIP limits to buffers. There are at least two areas, however, where an integral kanban tool can add value.First, the time a work item spends in a buffer is non-value-added. As we saw above, integral kanban tools should be able to calculate and report the flow efficiency of a given value stream. Being able to identify the time work spends in a buffer allows the tool to calculate flow efficiency automatically. (Alas, the time spent in non-buffer columns is not necessarily value-added time, thereby complicating the calculation.)Integral kanban tools might also calculate and set the size of buffers to optimize overall flow through the value stream. The size of a buffer is none other than the WIP limit applied to that buffer. Methods for calculating and managing such scheduling risk have grown out of classical project management. Whereas these methods have the goal of minimizing risk during the execution of a particular project, they might be adapted to minimize risk for the repetitive execution of a value stream, rather than executing a single instance (as in a project). The Enterprise Service Planning method works along these lines. Calculating optimal buffer size based on historical data and algorithms contrasts with the trial-and-error, experimental approach that teams must otherwise use. Further, various software scheduling automation tools implicitly manage job queue size as part of the scheduling of tasks to distributed resources.I am not aware of any kanban products on the market today that have such features. The enterprising kanban tool developer might find inspiration in methods such as the Activity Schedule Risk method, the Event Chain method or Critical Chain Scheduling.Portfolio ManagementBasic kanban tools might be suitable for individuals or for individual teams. However, as soon as an organization reaches a certain size, it will have multiple teams. Those teams interact and frequently hand off the results of their work to each other. Thus, it is important to have visibility not only of the work of individual teams but also of how that work fits together in the complete value chain of the organization. This is the purpose of kanban portfolio management.There are two principal aspects to kanban portfolio management. The first is to provide a high-level, aggregated view of the initiatives or projects of the organization. The second aspect is to manage the interactions among teams.Aggregated views of workWhen I refer to initiatives or projects, I am talking about some arbitrary container for, or aggregation of, a set of work items. This container might be what we traditionally call a “project”, but it could also be, for example, a category of tasks for operating and maintaining an organization’s infrastructure. For marketing organizations, it might be a single marketing campaign. For financial organizations, it might be the set of tasks performed to track, manage and report on a financial cycle. In short, it is any grouping of work items at a level higher than the tasks of a single team.The fundamental concept of portfolio kanban is that each work item may be related to one or more higher-level cards visualized on a separate, portfolio card board. Each of those higher-level cards is used to aggregate and track the overall progress of the initiative or container. As the kanban cards of the various teams move through their respective value streams, their related portfolio cards are also updated automatically (see Figure 3).Figure 3: An aggregated, portfolio view (top) and a detailed team view (bottom)Portfolio management provides high visibility of the status of work to those responsible for initiatives, individually or collectively, without their getting bogged down in the details of how each team is managing its work items. It provides them with a factual basis, for example, for reallocating resources among the various participating teams. Thus, the portfolio kanban board supports a higher-level management cycle, providing aggregated performance information on an initiative by initiative basis. In this sense, portfolio kanban boards are sometimes described as “team of team” boards.In addition to supporting the work of initiative management, portfolio boards also support individual teams. For example, a key repeating question for each team is “Now that we have the capacity to start a new work item, which item shall we choose?” Knowing how each ready work item contributes to broader initiatives may help the team make this decision.Top management of the organization should find it useful to aggregate initiatives into higher-level containers, such as the implementation of corporate strategies. Portfolio kanban tools enable multi-level aggregation. For example, work items might be aggregated into a portfolio of projects. Projects might be aggregated into a portfolio of strategies.Inter-team dependenciesThe second aspect of portfolio management concerns the dependencies of work items handled by one team on the work items handled by another team. Knowledge and visibility of these dependencies may help prioritize work items. Suppose there are two work items ready for work. All other things being equal, it might be better to start an item for which another team is awaiting the results.Figure 4: A tool might provide a visual indication of a dependency outside of the current teamAnother example concerns the cases where two different teams regularly hand off work to each other in the middle of their respective value streams. In an ideal world, such cases would not exist. But in our constantly changing and unpredictable world it often happens that a work item needs to be parked on one team’s kanban board until it is successfully handled on another team’s kanban board. An integral kanban tool makes this easy to do and transparent to understand (see Figure 4).Granular Access ControlIndividuals and simple organizations using the Kanban method rarely have significant issues with access control. Each board has an owner who decides who else may access that board. Access often means the right to change the board itself, as well as to create, modify and delete cards. This owner-dependent approach is unsatisfactory in other organizations, especially when the confidentiality of work requires strict and auditable controls.An integral kanban tool enables ad hoc and highly granular access rights, distinguishing among administrators, board owners, task executors, those with viewing rights, modification rights and yet other roles. Rights might also reflect a user’s position within an organizational structure in addition to the role being played. For example, an organization might have multiple business lines. Each board might be associated with a business line. Access rights might be different for each business line and each board (see Figure 5). Figure 5: An example of granular work item access rights per roleIn addition to granular access control, some kanban tools also provide single sign-on functionality. This feature helps to increase service availability while reducing the support effort.Support for Improvement CyclesVisualized work and analytics provide data that are useless unless the organization exploits them as part of its continual improvement activities. How can an integral kanban tool support these improvement activities?Figure 6: Position of the integral kanban tool in improvement cyclesMany products (not specifically kanban tools), free and paid, support the meetings during which decisions are made about improvements. Some tools support structured meetings, such as scrum retrospectives. Others are more free-form (and agile!). These tools capture ideas (often in the form of virtual sticky notes). They might provide a means to capture emotional responses, sometimes anonymously. Some tools support collaborative diagramming, mind-mapping and such, or have generic whiteboards.Analysis of captured input is typically supported via structured templates for the classification of ideas and feelings. Tools might formalize voting procedures or other types of decision-making.Few, if any, tools complement these subjective inputs with objective data drawn directly from the tools tracking team performance. In other words, the tool might provide some analytical data but it provides no functionality to make use of those data during the improvement activities. Each organization uses (or ignores!) those data as it sees fit, without kanban tool support.For tools that do provide some support, the ability to create work items directly in a kanban system is the most commonly found feature. Such tools thereby close the loop of the improvement cycle (see Figure 6). Let’s examine in more detail how integral kanban tools might support improvement cycles.Support for ExperimentationKanban encourages improvement via experimentation. Such experimentation consists of incremental changes to the parameters controlling the flow of work. If the experiment results in improved flow, the team should probably continue to adjust the parameters along the same lines. If, on the other hand, the experiment results in lesser flow, the team should probably roll back to the previous state. Or, at least, it should try a different change.The typical examples of incremental changes include:reduction of work in progress limits (normally, WIP limits are increased only as part of a roll-back to a change) (see Figure 7)changes to the quantity of resources available to perform workchanges to the definitions of classes of servicechanges to service profilesadding or removing buffers to work queueschanges to work commitment policiesinter aliaFigure 7: Analysis of experiments on the effect of a WIP limit on lead timeThe analytics should be able to present and compare statistics from before and after any incremental changes. Such comparisons might be visualized via changes in time series data. (Some organizations might also wish to store aggregated values as part of its system of record for the performance of work.)Suppose the tool can produce a line graph of rolling mean lead time for a certain category of work item (for example, Figure 8). The rolling mean might change (for better or for worse) following any incremental change to the system. Ideally, the kanban tool should automatically annotate the graph with the moment of change, especially if the graph does not make the probable cause and effect visually explicit. This is an example of the visualization of changes using time series data.Figure 8: Cycle time before and after an incremental changeOther types of analyses and corresponding visualizations might include:changes to the length and thickness of the tail in the probability distribution functionality. This analysis might be used, for example, to adjust the rules for identifying the class of service of work itemschanges to the standard deviation of a subset of lead time values. This analysis helps identify changes to the reliability of a team’s performancechanges to the randomness of variation in lead time values. This analysis might help to identify the introduction or removal of systematic errors in the work management system.Being able to roll back implies that the tool retains in memory the state of a flow before the change. As such, the tool could have a “roll-back” button to simplify the procedure. Even more interesting would be the ability to maintain notes about the incremental, including the reasons for making the change, the expected result, the actual result, if and when the change was backed out and why it was backed out.Given that teams and the work they perform are complex systems, it might be dangerous to blithely extrapolate from historical data. Saying, “We already tried that change, but it didn’t work” is hardly ever a reason not to try again, given that many other components of the system are likely to have changed, too, in the interim. We might expect, someday, that integral kanban tools will have some machine learning capability, enabling them to assess the interactions of a large number of system parameters. Alas, there are many steps to master before reaching that stage.Support for Financial AnalysisAt the end of the day, an organization will (should!) consider the financial aspects of its improvement initiatives. While the tuning of such flow parameters as WIP limits incurs few direct costs, if any, changes to the available resources, such as the number of team members, is one of the greatest costs to the organization.Suppose a team presents to the relevant service owner (and budget holder) a request for more resources. If the team is mature, it will have gone through a series of optimization steps, such as the five focusing steps in Goldratt’s Theory of Constraints, before making such a request. Thus, the decision to expand a team needs to take into account not only the costs of the new resources but also the probable impact of those resources on the flow of work. Ultimately, the decision should consider the value of the projected increased throughput and/or reduced cycle times to the organization and other stakeholders.An integral kanban tool can provide many of these data, assuming that the costs of resources used and the value of work items completed are made known to the tool. Using techniques similar to projections of the probable completion date of work, the tool could project the probable impact of added resources.However, the deciders need to complement the data from the Kanban with data from other sources. A Kanban tool is unlikely to be able to project underlying changes in the drivers of demand for services, changes in competition or regulatory changes. The tool will not know how much cashflow is available to pay for the new resources. It can only extrapolate from historical data on the assumption that such conditions will not change, which assumptions might be unwarranted.Choosing a kanban toolI have outlined above the potential benefits of an integral kanban tool. Those benefits can be realized only if the organization using the tool has sufficient maturity to make good use of the functionality. There is no single “best” kanban tool. There is only the tool appropriate to the needs and capabilities of the organization, at any moment in that organization’s development.An integral kanban tool that can be as simple to use as a basic to-do list tool has significant advantages over the latter. The basic to-do list tool is limited in its ability to evolve as the user organization matures. The integral kanban tool, on the other hand, allows the organization to progressively make use of its rich functionality, but only when that organization is ready to benefit from that functionality.In this sense, integral kanban tools are very different from high-end ERP tools. The typical ERP tool cannot be used at all until a large investment is made in analyzing needs, configuring the tool, training the users and testing the tool. Kanban tools are usable with only a few minutes of configuration effort. They are agile tools designed for agile work methods. Indeed, for organizations just starting to use the kanban method, I would strongly urge them to start with the simplest of value streams (Ready–Doing–Done) and a minimum of rules and automations.In this article, I have provided a framework for considering the value of tools to support the Kanban method. Organizations may use this framework to inform their reflection on how to improve the flow of their work. They might find that the cost of an integral kanban tool is paid back many times over.Figures 2-5 in this article are based on the kanban software tool Kanbanize. The other figures are the work of the author.