One Layer Down on Kanban
The pair of essays on Scrum and Kanban worked with a particular version of Kanban: the version that appears most in the discourse, in the freely-available core guidance, in the conversations practitioners have when they pick the practice up and try to use it. That version of Kanban does substantial things, and the analysis of where its load-bearing assumptions sit and how they fail was honest about that version.
It was also incomplete. There is a body of guidance, called the Kanban Maturity Model (KMM), which addresses several of the failure modes those essays attributed to the practice itself. KMM exists, it is well documented, and it is not difficult to find if you know to look. The essays did not bring it in. The omission was real, and the failure-mode attributions need to be revisited with KMM in view, because some of them change shape when it is.
This essay does that revisiting. It does not retract the original analysis. The failure modes those essays described are real, they occur in the part of Kanban that most of the discourse actually transmits, and the description of where the load-bearing assumptions sit was accurate for what it engaged with. What KMM adds is a framework that re-reads several of those failures as failures of advisor judgement rather than failures of the practice, and that re-reading has force. KMM also introduces its own load-bearing assumptions, which the same analytical move surfaces and examines.
The move the foundational essay set out applies as readily to one’s own analysis as to anyone else’s. Every validator rests on assumptions. The original essays stopped at one depth. KMM is a layer further down. KMM’s own assumptions are a layer further down again. That is the work.
What KMM proposes
The Kanban Maturity Model is a body of guidance, owned by Kanban University, that organises Kanban practices along seven levels of organisational maturity, numbered 0 through 6. Level 0 (Oblivious) describes individuals managing their own tasks without any process awareness. Level 6 (Built for Survival) describes organisations capable of double-loop learning about their own identity and strategy. Between them sit Team focused, Customer-Driven, Fit-for-Purpose, Risk-Hedged, and Market Leader.
Each level has cultural values associated with it (Collaboration, Taking Initiative, and Transparency at level 1, for example, with different values at each subsequent level). Each level has a scope of attention (Task, Deliverable, Product/Service, Product Lines/Shared Services, and beyond). And each level has a set of specific practices grouped under the six general Kanban practices: Visualize, Limit WIP, Manage Flow, Make Policies Explicit, Feedback Loops, and Improve Collaboratively/Evolve Experimentally.
The practices at each level are calibrated to the cultural conditions of that level. Practices appropriate to a higher level presume cultural conditions that the lower levels have not yet developed; practices appropriate to a lower level make smaller demands on the surrounding system and are robust to its absence. The model holds that this calibration matters: a practice installed at a level whose cultural conditions are not yet in place is unlikely to function as the practice claims to function.
Between adjacent levels, the model places transition practices: practices designed to lay the groundwork for the next level’s main practices, so the cultural conditions catch up before the load-bearing practices arrive. The transition practices are typically less identity-disruptive than the main practices; they can be introduced in an organisation aspiring to the next level without provoking the resistance that the main practices would.
Two failure modes are named in the literature, and both belong to the advisor rather than to the practice. The first is overreaching: installing practices appropriate to a higher maturity level in an organisation whose cultural conditions cannot sustain them. The predictable consequence is that the practices fail in ways that look like the practice failing, when what has actually failed is the judgement that installed those particular practices in that particular context. The second is false summit plateau, or stagnation: stopping the journey at a level the organisation has reached without continuing to evolve. Both are advisor failures. Neither is read as a failure of Kanban.
A note about where KMM sits relative to the rest of the Kanban discourse. KMM is owned by Kanban University and is the centrepiece of their advanced training. It is also referenced from the Official Kanban Guide (the freely-available document that is most people’s first encounter with the method), but referenced is the operative word. The Guide names KMM once, as a pointer for readers who want more detail on implementation by maturity level, and otherwise presents Kanban as a method that can be applied at any scope with appropriate depth of practice. The framework that would tell you which practices belong at which level, and which would resist your installing higher-level practices in lower-level conditions, is not in the core guidance most practitioners encounter. It is in a body of work that sits at one remove. This matters for what follows, but it is not the topic of this essay, and the analysis below treats the model on its own terms.
Revisiting the failure modes
The two original essays surfaced several failure modes that affected Kanban. Some are addressed directly by the model. Some are addressed through different mechanisms. Some are not really touched. Honesty about each group is what the move asks for.
Failure modes KMM addresses through level sequencing
The essays described WIP limits creeping upward under stakeholder pressure. They described boards surfacing bottlenecks to teams without the authority to address them. They described operational reviews that surfaced nothing of substance. They described probabilistic forecasting treated by the surrounding organisation as evasion. Each of these was presented as a failure of Kanban: a place where the practice’s mechanisms fail to deliver because the conditions for their delivery are absent.
KMM reads these differently. WIP limits as a system-level discipline (the kind that gets breached under pressure) sit at level 3 in KMM’s architecture, where they require cultural conditions including agreement, leadership at all levels, and unity across the team-business interface. A board that surfaces bottlenecks and expects the team to address them is operating in the territory of level 4 or 5 practices, where the cultural conditions include data-driven decision making and a kaizen culture across the organisation. Operational reviews are a level 4 practice, requiring substantial preceding cultural development. Probabilistic forecasting with reference classes and Monte Carlo simulation is also a level 4 practice, requiring the data-driven decision-making culture that level 4 develops.
If the cultural conditions are not present and the practice is installed anyway, KMM says the practice will fail. The failure is predicted by the framework. It is named in the literature as overreaching. And it is attributed to the advisor’s judgement, not to the practice.
The corresponding level 1 and level 2 patterns are robust to the absence of those higher-level conditions. Per-person WIP limits within a team. Aggregated personal kanban boards. Basic team kanban with emergent workflow. Qualitative metrics. These practices do not require the cultural conditions that the higher-level practices need, because they are designed for contexts where those conditions have not yet developed. They do less; what they do, they do reliably; and they create the conditions under which the next level’s practices can later be introduced.
This is a substantial re-reading. Failure modes that the essays treated as load-bearing assumptions of Kanban-the-practice are, on KMM’s reading, consequences of installing the wrong practices for the cultural conditions. The assumption that the team has the standing to refuse work, that the surrounding organisation can receive probabilistic signal, that someone with authority will respond to what the board surfaces: these are not assumptions Kanban silently borrows from somewhere outside itself. They are cultural prerequisites that KMM says you should know about, check for, and not violate by installing practices that depend on them when they are not in place.
The original essays did not engage with this. They worked with a version of Kanban in which the level distinction was not visible, and read the consequences of installing the wrong level’s practices as Kanban failures. With KMM in view, the failures of category one are advisor failures of overreaching, and the original attribution is not sustainable in the form the essays presented it.
Failure modes KMM addresses through cultural attention
The pair of essays raised what we called the informed-discretion bet: the claim that Kanban serves a practitioner who brings their own discipline, who engages with the practice’s proposals as substantive guidance, who runs the mechanisms as a system rather than as a collection of unconnected artefacts. We noted that this bet fails where the practitioner does not engage in those ways, and that the failure looks like the practice hollowing out: policies that exist on paper but are never revised, WIP limits that are breached quietly, operational reviews that surface nothing.
KMM does something specific with this. It treats informed discretion itself as a cultural property that develops at particular maturity levels, and it makes the cultural conditions an explicit object of attention rather than something the practice silently borrows from. The cultural values associated with each level are precisely the substrate that informed discretion depends on. Without the dispositions associated with level 2 cultural development (which include acts of leadership and understanding), the practitioner does not have the relationship to the practice that the bet on informed discretion needs. KMM’s response is not to assume the disposition is present; it is to use transition practices to develop the conditions under which the disposition can form.
Get Bruno Baketarić’s stories in your inbox
Join Medium for free to get updates from this writer.
This is not a complete answer. A team can sit at the cultural level KMM says is required and still fail to engage substantively with the practice’s proposals. The development of informed discretion is a tendency KMM works to encourage, not a guarantee it provides. But the framework addresses the question by making the cultural conditions a deliberate object of guidance, which is more than the version of Kanban the original essays engaged with does.
The same logic applies to the visibility-without-agency observation. The essays noted that a board surfacing a bottleneck to a team without authority produces signal that goes nowhere. KMM would say: a board surfacing system-level bottlenecks is a higher-maturity practice; the cultural conditions for it to translate into action include leadership at all levels and the values that accompany Risk-Hedged or higher organisational maturity. If those conditions are absent, the practice that surfaces the bottleneck was the wrong practice to install in the first place. The remedy is not to expect agency the system has not developed; it is to use level-appropriate practices that do not require it, and to use transition practices to develop the conditions under which higher-level practices become viable.
Failure modes KMM addresses partially or differently
Two failure modes from the original essays do not map as cleanly onto level sequencing or cultural development.
The cycle-time-and-learning-loop conflation we described is not obviously a maturity-level problem. A team at level 4, with sophisticated metrics and well-developed cultural conditions, can still collapse cycle time and learning loop into a single variable and lose the distinction between what the practice controls and what the surrounding feedback dynamics control. KMM does not directly address this confusion; it addresses adjacent questions, including how to use metrics appropriately at each level, but the specific conflation we named is not a level-progression issue. Where KMM helps is in its insistence that metrics serve specific purposes and that those purposes should be made explicit; a team that has internalised that discipline is less likely to conflate two different signals into one. But the conflation is not a failure of cultural maturity; it is a failure of analytical care, and the framework treats it as adjacent rather than central.
The fled-from-Scrum adoption pattern is similarly oblique to KMM’s architecture. The team that has fled Scrum and adopted Kanban is not progressing through maturity levels; they are carrying a posture from a different practice, shaped by what they were leaving rather than by what they are entering. KMM has things to say about adoption posture (it is explicit, in its description of resistance, that practitioners bring identities and emotional commitments that interfere with practice adoption), but the specific pathology of using Kanban as relief from prescription rather than as substantive engagement with proposals is not centrally addressed. Where KMM helps is at the level of advisor judgement: an advisor who reads the team’s posture as fled-from-Scrum has reason to start with level 1 and level 2 patterns rather than installing higher-level practices that the team will use as licence for absence. But the issue is in the relationship to the practice, not the practice itself, and the framework treats it indirectly rather than as a primary object of attention.
Failure modes KMM does not address
Some of what the original essays surfaced sits outside KMM’s frame, and honesty requires naming it.
The chain-of-pressure dynamic at the organisational level, the way pressure transmits through hierarchies and compresses cognitive space at each link, is not addressed by KMM. The framework treats the organisation as the unit of maturation, and the cultural values it describes are properties of organisational maturity. The mechanism by which executive-level pressure produces the conditions for advisor overreaching, by which buying-side pressure converts a careful diagnostic question into a procurement decision, by which the cultural conditions for level 3 cannot develop because the surrounding system keeps reproducing the formation that prevents them: none of this is in KMM’s frame. The model says: do not overreach. It does not say: here is why the pressure that produces overreaching is structural to organisations, or what to do when the advisor lacks the standing to refuse what the buyer is asking for.
This is not an indictment of KMM. A model is a model; it cannot do everything. But the failure modes that arise from the chain-of-pressure dynamic are still failure modes, they still produce the consequences the original essays described, and KMM addresses them only at the level of advisor exhortation rather than at the level of the dynamic that produces the failures. Where the surrounding system is structurally unable to provide the conditions KMM says a level requires, the framework’s response is to install lower-level practices that do not need those conditions. That is sound advice. It does not change the fact that the conditions for higher-level maturity may never develop, because the mechanism that suppresses them runs at a level KMM is not operating at.
One final observation belongs here. KMM names two failure modes, not one: overreaching and false summit plateau, or stagnation. The original essays surfaced things that, on reflection, mostly map to overreaching. They did not surface stagnation as a Kanban failure mode: the team or organisation that adopts Kanban, reaches a stable state, and stops evolving from there. KMM addresses both failures. The original analysis only saw one of them. This is a small piece of evidence about KMM’s coverage being broader than the analysis above has probed, and it belongs in the honest accounting of what is being engaged with.
What KMM itself assumes
KMM substantially mitigates several of the failure modes the original essays surfaced. That mitigation is real. It also rests on assumptions, and the same move that surfaced the original failure modes will, applied honestly, surface KMM’s own load-bearing structure.
The first is an existence assumption about advisor capability, and it is more demanding than the previous essays in this series acknowledged. The architecture depends on someone (coach, internal change agent, leader) who can read the organisation’s current cultural maturity accurately, who has the standing to install level-appropriate practices rather than what the buying side is asking for, who can resist the pressure to promise higher-level results on a faster schedule than the work can honestly support, and who can work on the environment itself to develop the preconditions under which the next maturity level can emerge. The last of these is the most demanding. The advisor is not only installing practices that fit the present cultural conditions; they are doing the longer-horizon work of shaping the environment so that future cultural conditions become reachable. That work runs into the same mechanisms the earlier essays described: management formation that selects against the dispositions the next level requires, chain-of-pressure dynamics that compress the cognitive space within which environment-shaping happens, organisational programmes that buy delivery against milestones rather than the work KMM is asking for, whose pace cannot be promised in advance because it is a consequence of how the cultural conditions actually develop. The literature names the advisor failure; the architecture also depends on the advisor’s presence in the form the literature assumes. The Kanban Management Professional credential is the foundation accreditation, awarded after substantial training in the practices and the mechanics. Many practitioners stop there, which is reasonable: the foundation knowledge is real and useful. But KMM sits at one remove from that foundation, in the advanced part of the curriculum, and the population that has gone on to internalise it is a smaller subset of even the foundation-credentialed group. The broader population calling itself Kanban coach is larger again, has not necessarily encountered KMM at all, and is not selected for the judgement the framework requires, let alone for the standing and patience the environment-shaping work additionally needs. The advisor who has internalised KMM, has the standing to act on it, and has the room to work on the environment alongside the team exists; whether they will be the advisor the organisation actually engages is a separate question, and one KMM holds at arm’s length.
The second is a mechanism assumption about cultural progression. KMM holds that level 1 cultural values, given the right transition practices, evolve into level 2 values, which evolve into level 3 values, and so on. This is a strong claim about how organisational culture actually moves. The earlier essays in this series, particularly the ones on management formation and on the chain of pressure, argued that some cultural conditions do not evolve through local practice changes because the formation mechanism that produces them runs at a level the local practice cannot reach. The advisor installs level 2 patterns appropriately; the team’s culture develops; the surrounding organisation does not, and the conditions for level 3 progression never arrive. KMM is a model of evolution within a system. Whether the surrounding system permits that evolution, and whether the formation mechanisms that shape the surrounding system can be reached by team-level or service-level practice changes, are questions the framework does not answer.
The third is an exclusivity assumption about coverage. KMM’s framework (seven levels, six general practices, a catalogue of specific and transition practices) covers a substantial territory. Whether it covers all of it is a different claim. The cycle-time-and-learning-loop conflation is not a maturity-level problem. The fled-from-Scrum adoption pattern sits at an angle to the level progression. The chain-of-pressure dynamic operates outside the unit of maturation KMM takes as its object. A reader who treats KMM as a complete map of how Kanban can fail is making a claim KMM itself does not make, and which the framing-problem section of the foundational essay made it clear no model can support. KMM is one map. Other maps would surface other failure modes. KMM’s coverage is part of its value; its non-coverage is part of what makes it a model rather than the territory.
The fourth is a reach assumption, and this is where the discourse observation from the opening section comes back. KMM’s mitigation is real where KMM lands. Whether KMM lands is itself a question. The body of guidance that addresses overreaching is not where the practice is most commonly first encountered. The Official Kanban Guide names KMM once as a pointer for those who want more depth. The practitioner who reads the Guide and starts an implementation will not, by default, know that overreaching is a named failure mode or that a framework exists for avoiding it. The advisor who has not encountered KMM cannot install practices KMM would recommend or refuse practices KMM would forbid. The mitigation is structurally available; whether it is structurally accessible is a different question, and the answer to that question is largely shaped by editorial choices KMM itself does not control.
The recursion
This analysis rests on its own assumptions.
The reading of KMM I have offered is a position, and someone closer to the literature would draw the boundaries differently. The decision to treat the advisor-failure framing as a substantial re-reading rather than a deflection is itself a choice. The grouping of failure modes into four buckets (addressed through level sequencing, addressed through cultural attention, addressed partially, not addressed) is a way of organising the analysis that another framing could redraw. The claim that KMM’s coverage is partial relies on a sense of what the territory contains, and that sense is itself frame-dependent.
This is the condition. The analysis went one layer below where the original essays stopped. KMM is a layer further down. KMM’s assumptions are a layer further down again. The next layer is whatever a reader brings to questioning what this essay has said. The recursion does not resolve. It makes the weight visible at one more depth than it was at before.
Close
The original essays accurately described what fails when these conditions are absent. The Kanban Maturity Model is the most serious attempt to make those conditions a deliberate object of guidance, and it substantially re-reads several of what the essays called Kanban failures as failures of advisor overreaching. That re-reading has force. The original failure-mode attributions, in the form the essays presented them, are not sustainable with KMM in view.
KMM’s mitigation rests on its own assumptions: about advisor capability, about cultural progression, about coverage, and about the reach of the guidance itself. Those assumptions are the kind the foundational essay’s move surfaces and examines, and they connect directly to mechanisms the earlier essays in this series have already described. Where the conditions for KMM’s mitigation are not present, the failures the original essays described will continue to occur, regardless of what the framework says about whose failure they are.
None of this resolves the question of what to do. It makes the weight visible at one more level down than it was before. That is what the move is for.
