Organizing engineering teams has never been more consequential. As software systems grow in complexity, the way you structure the people building them directly shapes the quality, speed, and sustainability of what gets delivered. This is precisely why team topologies captured the attention of so much of the engineering leadership world when Matthew Skelton and Manuel Pais introduced the framework in their 2019 book.
The model offers something genuinely valuable: a shared vocabulary and a principled approach to designing teams around cognitive load, clear boundaries, and deliberate interaction modes. For many organizations, it has been a revelation.
But no framework is without its blind spots, and team topologies is no exception. In this analysis, we will examine what the model genuinely gets right, where it delivers lasting insight, and where practitioners tend to hit walls that the book does not fully address. Whether you have already implemented parts of this framework or are evaluating it for your organization, this breakdown will give you a clearer, more honest picture of its strengths and the open questions it leaves on the table.
What Team Topologies Actually Is
Team Topologies is an organisational design framework created by Matthew Skelton and Manuel Pais, published in 2019, built around a single governing principle: the way you structure your teams is not a neutral administrative decision. It directly encodes the architecture of everything those teams produce. The framework provides leaders with a precise, practical vocabulary for intentional organisational design, moving organisations away from reactive, historically-accumulated structure toward deliberate, evidence-based configuration. At its core, the framework defines four fundamental team types and three interaction modes that together determine how value flows through any organisation.
The four team types are stream-aligned, platform, enabling, and complicated-subsystem. Stream-aligned teams are the primary type, each focused on delivering a single business capability end-to-end; Skelton and Pais argue that ideally 90% of teams should be stream-aligned. Platform teams provide self-service internal infrastructure, reducing cognitive load on stream-aligned teams by treating their output as an internal product. Enabling teams act as specialist coaches, helping stream-aligned teams adopt new practices. Complicated-subsystem teams own deeply specialised components that would otherwise overwhelm stream-aligned teams. Alongside these types, the framework defines three interaction modes: collaboration (intensive joint working), X-as-a-Service (one team provides a decoupled service another consumes), and facilitating (one team develops the capability of another). A critical and counterintuitive insight is that not all communication is beneficial; focused, intentional interaction between specific teams outperforms indiscriminate connectivity.
The framework's intellectual foundation is Conway's Law, formulated by computer scientist Melvin Conway in 1967: any organisation that designs a system will produce a design whose structure mirrors its communication structure. As Martin Fowler's analysis of Team Topologies reinforces, when system design conflicts with organisational design, organisational design wins. Every time. This transforms team structure from an HR concern into an architectural one.
The Inverse Conway Maneuver follows directly from this logic. Rather than allowing team structure to emerge and then observing what architecture results, leaders deliberately shape topology first to produce the desired system architecture. The entire Team Topologies framework can be read as an extended, practical treatment of this maneuver in action.
The framework's diagnostic premise is equally important: most team structures evolve accidentally, shaped by historical incidents, individual personalities, political pressures, and temporary crises rather than intentional design. Organisations contain three simultaneous structures: the formal org chart, the informal network of personal relationships, and the actual value-creation flow. These rarely align, and the gap between them is where coordination overhead and cognitive overload accumulate.
Originally developed within software delivery and DevOps contexts, Team Topologies has expanded significantly. By 2025 to 2026, the framework explicitly addresses AI workforce readiness, product operating model acceleration, platform engineering, and enterprise transformation, signalling that team design thinking is now being pulled directly into the broadest organisational change conversations.
Why Cognitive Load Is an Engineering Problem, Not a Personal One
Perhaps the most consequential idea in Team Topologies is also its most quietly radical: cognitive load is not a personal resilience problem. It is an engineering variable. When a team's domain exceeds its cognitive capacity, delivery slows, quality degrades, and people burn out. The instinctive organisational response is to send the team lead to a time-management workshop or hire a coach. The Team Topologies diagnosis is blunter: the structure is wrong. The three supporting team types in the framework, platform, enabling, and complicated-subsystem, exist specifically to absorb complexity off stream-aligned teams, not to motivate those teams to cope with more of it. This reframe matters enormously, because it shifts the locus of accountability from the individual who is struggling to the system that generated the struggle. As Matthew Skelton has stated explicitly, the goal for leaders is not doing less but creating environments where teams can contribute safely, quickly, and sustainably.
Static org charts make this problem worse by design. A hierarchy diagram represents reporting lines; it does not represent workstreams, communication frequency, or real accountability flows. Over time, actual work patterns diverge substantially from the official chart, and the gap between the two grows invisibly. Leaders who manage by the chart are, in effect, managing a fiction. The compensating behaviour is reorganisation: redrawing the boxes in the hope that the new configuration will better reflect reality. It rarely does, because the reorganisation is still drawn from org-chart logic rather than from an honest audit of where load actually sits and where dependencies genuinely run. The boxes change; the load distribution does not.
When structure is allowed to drift rather than being set with intention, three failure modes tend to compound over time. Sustained collaboration between teams, rather than signalling healthy cross-functional work, frequently indicates that team boundaries are simply unclear. Platform teams that mandate adoption rather than offering self-service capability generate dependency instead of autonomy. Enabling teams that never disengage create permanent reliance rather than transferred competence. Each of these patterns is a structural problem wearing the costume of a people problem. The cognitive load does not disappear; it relocates, typically accumulating on whoever absorbs ungoverned complexity by default.
This points to a broader shift in how mature organisations are beginning to diagnose dysfunction. The individual-deficit model, which attributes poor performance to character, skill gaps, or attitude, is giving way to a systems-level frame: this structure is generating the behaviour we are observing. W. Edwards Deming argued decades ago that the majority of workplace problems are produced by systems, not individuals. Team Topologies operationalises that argument for team design specifically, and the implications extend well beyond software delivery.
The implementation risk, however, is substantial and under-acknowledged. Changing team interaction modes is a costly intervention. It should only be undertaken when the existing model is demonstrably broken, not when a new framework sounds compelling. Practitioners and independent observers have documented a consistent failure pattern: organisations copy the labels of Spotify squads, SAFe trains, or Scrum ceremonies without internalising the cognitive-load principles that gave those structures their logic in their original contexts. The result is new dysfunction layered over old dysfunction. As Skelton has emphasised in discussions of Team Topologies applied to complex systems, the question practitioners must ask is not which label fits, but whose load does this structure actually reduce. Cargo cult adoption answers the first question while ignoring the second entirely.
Span of Control: Where Team Topologies Goes Quiet
Span of control has always been one of organisational design's most contested variables. How many direct reports can a leader meaningfully support? At what point does a management layer become a bottleneck rather than a bridge? Team Topologies engages with these questions at the team level, endorsing small units of roughly five to nine people and treating cognitive load as the ceiling that determines team scope. That guidance is genuinely useful. What the framework does not provide is a concrete, vertically calibrated model for how span should narrow as altitude increases, validated against anything older than a 2019 book or a mid-century law about communication structures.
This is not a minor omission. Five years of organisational adoption have demonstrated that Team Topologies transforms horizontal flows with real precision, yet practitioners continue to encounter the same unresolved question at the top of their structures: how should authority compress as organisations scale, and what does that compression do to the people leading at each level?
The Fractal Answer Team Topologies Does Not Offer
The Cone Model, the structural framework developed by William J. Oxfield in The Lonely Leader, proposes a historically grounded answer. Span narrows in a fractal ratio: approximately ten at the first tier, five at the second, two at the third, before resetting to a full circle of ten at each power of ten. This is not an arbitrary heuristic. Jethro's counsel to Moses in Exodus 18 produced precisely this architecture: leaders of thousands, hundreds, fifties, and tens. Roman cohort structure encoded similar ratios into military command for centuries. Robin Dunbar's research on stable social group sizes, which identifies natural cognitive layers of roughly five, fifteen, and fifty within the outer limit of approximately one hundred and fifty, maps onto the same oscillating pattern.
What this reveals is that relational load does not increase linearly with seniority. It oscillates. A leader who moves from managing a team of ten to managing a small executive group of two is not simply carrying more; they are operating in a structurally different relational environment. And crucially, at certain altitudes, a leader inherits a full peer circle again rather than a narrower one. The weight of isolation is not an inevitable consequence of climbing higher; it is a symptom of structural design that has not accounted for how relational geometry shifts at each layer.
The Vertical Dimension Team Topologies Leaves Unmapped
The framework's core critique of org charts is that they fail to model real communication flows. That critique is directed horizontally: at the gaps between teams, the invisible dependencies, the unofficial collaborations that org charts cannot show. There is no corresponding critique of how org charts fail vertically, specifically, how they obscure the relational load on leaders who sit above multiple teams of teams with no structural peer relationships inside the organisation they govern.
Team Topologies optimises for flow between teams. It does not model how authority and trust interact as altitude increases, how decision latency compounds through management layers, or why a senior leader can implement the framework's horizontal principles with genuine skill and still feel structurally isolated at the apex. The runn.io summary of Team Topologies captures the framework's strength well: it gives leaders vocabulary for horizontal dysfunction. That vocabulary simply stops at the vertical boundary. A senior leader who successfully applies Team Topologies to their organisation's stream-aligned teams may have solved the flow problem entirely, while remaining in a condition the framework has no language to name, diagnose, or address.
The 7/14 Split: A Structural Distinction Team Topologies Does Not Make
The Cone Model, developed by William J. Oxfield in The Lonely Leader, introduces a framework that draws a precise structural boundary Team Topologies leaves unmarked. Of the twenty-one things people need to flourish at work, only seven require a leader's positional authority to generate: Direction, Vision, Mission, Mentoring, Reward, Advocacy, and Autonomy. The remaining fourteen, including Belonging, Safety, Correction, Teamwork, Trust, Training, Recognition, Friendship, Encouragement, Accountability, Celebration, Challenge, Forgiveness, and Care, cannot flow downward from a reporting line. They exist only in the lateral space between peers. This is not a soft observation about workplace culture. It is a structural claim with measurable consequences.
The structural implication deserves to be stated plainly. A leader who attempts to generate all twenty-one needs will burn out and still fail, not because of insufficient commitment or poor character, but because fourteen of those needs are architecturally inaccessible from the top. Belonging cannot be mandated. Forgiveness cannot be delegated. Friendship cannot be scheduled into a one-to-one meeting. These conditions emerge from how teammates treat one another across hundreds of small, lateral interactions each week. When a leader absorbs responsibility for generating them anyway, the relational load becomes unsustainable and the needs remain unmet regardless. This is a topology problem, not a motivation problem.
Team Topologies addresses a different but related topology problem with genuine precision. Its X-as-a-service and facilitating interaction modes are specifically designed to reduce unnecessary collaboration overhead between teams, cutting the cognitive noise that prevents stream-aligned teams from maintaining flow. Book summaries of Team Topologies consistently highlight the framework's core argument that not all communication is productive, and that deliberate interaction boundaries protect team focus. This is sound structural reasoning. What those modes cannot see, however, is the relational health inside a team once the inter-team friction is reduced. A perfectly configured set of team interactions can still house a group of people who lack psychological safety, peer accountability, or genuine mutual care. The framework's vocabulary offers no instrument for naming or measuring that deficit.
This is precisely where the 7/14 split functions as a complementary diagnostic layer rather than a competing one. Team Topologies asks whether teams are interacting in the right mode. The 7/14 split asks whether the humans inside those teams are receiving what they need to sustain performance over time. Both questions are necessary. Neither answer substitutes for the other. An organisation can achieve excellent inter-team flow architecture and still watch its highest-performing teams erode from within, because the fourteen peer-generated needs were never structurally enabled.
Google's Project Aristotle provides corroborating evidence that this concern is not theoretical. Studying 180 teams over several years, Google found that psychological safety was the single strongest predictor of team effectiveness, outranking individual talent, tenure, and technical skill. The other leading factors, including dependability, meaning, and relational clarity, map directly onto the Cone Model's peer-generated fourteen: Safety, Trust, Belonging, Encouragement, and Forgiveness are all present in Aristotle's findings. Team Topologies adopts a "team-first" mindset and acknowledges that teams function as living organisms rather than static groupings, but it provides no structural mechanism for how those relational conditions are generated or sustained at the peer level. The Cone Model's 7/14 split names that mechanism explicitly, making it diagnosable and, crucially, designable.
The Gap Team Topologies Leaves Open: Leadership Isolation at the Apex
There is a gap in the Team Topologies framework that its own internal logic makes visible, even if the framework itself never names it. Team Topologies optimises for the flow of work between teams with genuine rigour: it maps cognitive load, defines interaction modes, and engineers out the structural conditions that cause bottlenecks. What it does not model, and does not attempt to model, is what happens to the person at the top of the structure once that flow is functioning well. The framework's unit of analysis is the team. The apex leader is the designer of the system, not a node within it, and that positional exclusion is never examined.
The structural logic of this exclusion is straightforward. When an organisation is designed around team topologies, every team gains defined peers, interaction modes, and collaboration patterns. The leader who holds accountability for the whole structure gains none of these, by design. They are the only person in the organisation operating without a counterpart at the same level of authority, context, and consequence. This is not a reflection of temperament, emotional intelligence, or leadership maturity. It is the necessary geometric consequence of how authority structures work: the apex position cannot, by definition, be placed inside a team. The isolation is designed in.
What makes this gap particularly instructive is that Team Topologies earns its credibility precisely by making a systems-level diagnosis. The framework's core argument is that poor outcomes are not caused by individual failure; they are caused by structural misalignment between team design and organisational intent. Cognitive load is not a personal resilience problem, it is an engineering variable. Bottlenecks are not caused by difficult individuals, they are caused by poorly designed interaction modes. This is the same diagnostic move that The Cone Model applies to the leader at the apex: leadership loneliness is not a personal failing or a character deficit to be coached away. It is a structural design condition, and it therefore requires a structural cure. The Cone Model's central claim, that the apex should be occupied not by one person but by a leader and a genuine deputy with real delegated authority, applies the same systems-level reasoning Team Topologies uses throughout its framework to the one position Team Topologies leaves unexamined. The framework stops short of extending its own logic upward.
No other framework in the current organisational design landscape fills this gap. SAFe, Scrum, and the Spotify model are all designed to optimise work within and between teams. They do not introduce any mechanism for the relational condition of the senior leader. The apex dyad as a structural design element remains entirely unoccupied ground across the field.
The practical consequence of this gap is specific and observable. A senior leader who has successfully restructured their organisation using team topologies principles has, in doing so, created better-flowing teams and a sharper experience of positional isolation simultaneously. The deliberate reduction of unnecessary coupling across the structure, one of Team Topologies' explicit goals, concentrates relational exposure at the apex rather than distributing it. The structural problem has not been solved. It has moved upward. The teams now have peers and defined collaboration patterns. The leader, having engineered those conditions for everyone else, has none of them for themselves.
The Apex Dyad: A Structural Cure, Not a Succession Plan
The structural answer to leadership isolation at the apex is not a wellness program, a coaching engagement, or a peer forum for senior leaders. It is a second person at the top, holding real authority, with a genuinely shared accountability horizon. The Cone Model calls this the apex dyad: not a chief of staff who manages upward logistics, not an executive assistant who controls the calendar, and emphatically not a succession candidate being groomed for a future role. Those relationships, however valuable in their own right, are defined by their asymmetry. The apex dyad is defined by its structural equivalence inside the authority structure.
Delegated authority is the operative term here, and it matters precisely. The genuine deputy in The Cone Model holds decision-making power in defined domains. This is not courtesy authority or provisional authority granted when the leader is travelling. It is real, bounded, and understood by the organisation to be real. The consequence of this distinction is significant: because the deputy carries genuine accountability, they can challenge the leader without political cost. They are not seeking approval, not managing their exposure, and not calculating how a disagreement might affect their performance review. They share the risk. That shared risk is what converts a supportive relationship into a peer relationship, and a peer relationship is the only thing that resolves structural isolation at the apex. Administrative support, however skilled, cannot do this because it does not share the weight.
The structural location of the solution is what separates the apex dyad from every popular alternative. Mentoring operates outside the authority structure. Coaching operates outside it. Executive peer networks, leadership circles, and advisory boards all operate outside it. They are genuinely useful for many things, but they cannot address the specific form of isolation that comes from being the final accountable point in a hierarchy. That isolation lives inside the structure, at the precise place where authority terminates and nobody above you shares your context. Placing a solution outside that structure, however thoughtfully designed, does not reach the problem. The apex dyad places the solution inside the authority structure, at the exact point where the isolation originates. That is not a subtle distinction; it is the entire argument.
The parallel to Team Topologies' Inverse Conway Maneuver is direct and instructive. The Inverse Conway Maneuver proposes that if you want a particular software architecture, you should deliberately design the team topology to produce it rather than waiting for the architecture to emerge from existing structures. The logic is proactive: shape the structure first and let the desired outcome follow. The apex dyad applies the same reasoning to leadership health. Rather than waiting for isolation to express itself as a performance decline, a retention crisis, or a strategic blind spot, you shape the apex structure in advance to preclude it. The structure becomes the intervention, not the response to a crisis already in progress.
History validates this principle with notable consistency. When Jethro counselled Moses in Exodus 18, the diagnosis was structural overload and the prescription was a formal hierarchy of delegated authority, with Moses retaining only what required his specific position. Roman military command, from the Republic's paired consuls to the cohort's internal command chain, repeatedly embedded a formal second at the apex as a structural necessity, not an administrative convenience. Modern special forces units carry a designated deputy with operational authority in defined conditions for exactly the same reason: resilience requires that authority never collapse to a single point. These are not echoes of the same management fashion. They are independent rediscoveries of the same structural truth, across millennia and contexts, which is precisely the kind of evidence that warrants serious analytical attention.
How to Use Team Topologies and The Cone Model Together
These two frameworks operate on different but adjacent layers of organisational structure, and that distinction matters. Team Topologies engineers the horizontal plane: how work flows between teams, how interaction modes are defined, and how cognitive load is distributed across team boundaries. The Cone Model engineers the vertical plane: how relational authority is structured within teams, how peer-generated human needs are met across the fourteen dimensions that leaders cannot provide alone, and how the person at the apex avoids structural isolation. One framework does not displace the other. They stack.
A Diagnostic Sequence, Not a Choice
The practical integration begins with sequence. Start with Team Topologies' interaction-mode and cognitive-load assessment to map where structural friction exists in team-to-team flow. This surfaces the horizontal problems: teams with poorly defined boundaries, coupling that has drifted from collaboration into dependency, platform teams that have become bottlenecks rather than enablers. Once that horizontal map is clear, layer the Cone Model's Full Cone 360° assessment across the same teams to identify relational health gaps operating inside and above them. A team that passes a cognitive-load threshold on paper may still be failing on belonging, psychological safety, accountability, or peer trust. Those fourteen peer-generated needs do not show up in any interaction-mode audit. They require a different instrument entirely.
For the leader coordinating the whole structure, the Find Your Deputy assessment from The Cone Model addresses the specific gap that no Team Topologies tool touches. It is designed to determine whether the person at the apex has a genuine second with real delegated authority, or only the appearance of one. The distinction is structural, not relational. A title does not create a deputy. Delegated authority does. Without that assessment, organisations routinely assume the apex problem is solved when it has merely been papered over.
The Cargo Cult Risk Is Identical in Both Frameworks
Both frameworks are principle-first, and that is where most adoption failures begin. Copying team topologies without understanding Conway's Law produces the same architecture the organisation was trying to escape, only with new labels. Copying a cone structure without understanding the 7/14 split produces a hierarchy that looks relational but still places all fourteen peer-generated needs on the leader's shoulders. The assessment layer in both frameworks exists precisely to prevent pattern-borrowing without diagnosis. Structure without diagnosis is reorganisation theatre.
Why AI Transformation Makes Both Frameworks Necessary Simultaneously
As organisations redesign team structures for AI workforce integration through 2025 and 2026, the relational stakes do not diminish; they sharpen. Matthew Skelton and Manuel Pais addressed this directly in a September 2025 Thoughtworks podcast, noting that AI adoption intensifies the tension between governance and autonomy inside teams. That tension is a relational problem as much as an architectural one. The leaders coordinating these transitions face compounded isolation: rapid structural change, unclear authority boundaries, and deputies who may lack the genuine delegated power needed to share the weight. Both frameworks are needed at the same time, applied to the same organisation, reading from the same diagnostic starting point.
Three Mistakes Leaders Make When Implementing Team Topologies
Mistake 1: Treating the Org Chart as the Organisation
Team Topologies is explicit on this point: the static org chart is a structural liability, not a design tool. It shows reporting lines, not workstreams. It shows titles, not interaction modes. Leaders who initiate a Team Topologies implementation by redrawing the org chart are not redesigning their organisation; they are relabelling it. The result is what practitioners now call reorganisation theatre: a visible restructuring exercise that changes the boxes, renames the teams, and produces a new slide deck, while leaving the underlying architecture, boundaries, and interaction patterns entirely unchanged. Viktor Cessan, whose pitfalls analysis has circulated widely among practitioners, states the failure plainly: the topology derives from a clear picture of the architecture and flow your product needs, not from the org chart you already have. That sequencing distinction matters enormously. Leaders default to the org chart first because it is politically legible and immediately actionable, while defining product architecture first requires harder conversations. The cost of that shortcut is an implementation that optimises for today's delivery constraints rather than tomorrow's structural needs.
Mistake 2: Optimising Flow Without Addressing Relational Load
The cognitive load framework in Team Topologies is one of its strongest contributions. By distinguishing intrinsic, extraneous, and germane cognitive load, it gives teams and their designers a vocabulary for engineering workload, not just managing it. What it does not address is the parallel problem one level up. A leader can implement a textbook-compliant topology, achieve measurably better flow across stream-aligned teams, and still be structurally isolated, overloaded with decisions that have no clear owner, and making progressively worse calls as relational and decisional burden accumulates at the apex. The same logic that justifies reducing team-level cognitive load applies directly to the leader's own structural position; the framework simply does not name it as a variable. Applying cognitive load thinking only to the teams below, while leaving the leadership layer structurally unreformed, is the organisational design equivalent of fixing the factory floor while the operations director works without adequate support. The flow optimisation is real. The isolation at the top is equally real.
Mistake 3: Skipping the Deputy Question
Most senior leaders assume that assembling a leadership team resolves the question of structural support. It does not. A leadership team is a set of direct reports. A deputy, in the Cone Model's precise use of the term, is something structurally different: a person who holds genuinely delegated authority and can act, decide, and hold accountability in the leader's absence. That distinction, between structural support and the appearance of it, is consequential. Direct reports without delegated authority create the sensation of a second layer while concentrating decision rights at the top. The result is a bottleneck disguised as a team. The Find Your Deputy assessment from The Cone Model is a practical diagnostic entry point for testing this assumption; it surfaces whether the person a leader considers their second actually holds the authority, trust, and relational proximity the role requires. Critically, this does not require a full organisational redesign as a precondition. It requires honest structural diagnosis, which is precisely what most senior leaders skip when they are busy validating a Team Topologies rollout and assume the leadership layer is already sorted.
Structural Design Is the Conversation — Team Topologies Started It
Team Topologies made a genuine and lasting contribution to organisational design. It moved the field from intuition, politics, and inherited accident toward intentional, principle-based structure. The shift it represents, from symptom-level reorganisation toward systems-level diagnosis, is one the broader field genuinely needed. That contribution stands on its own terms.
The precise limitation is this: the framework optimises the flow of work between teams. It does not address the relational structure of the people leading them. Specifically, it leaves untouched the structural isolation at the apex, the condition where the person at the top has no genuine peer, no real deputy with delegated authority, and no relational architecture to match the positional one. That is not a criticism of Team Topologies. It is a description of where the conversation must continue.
Two assessments extend that conversation into actionable diagnosis. The Full Cone 360° evaluates relational health across the entire cone of leadership, not just team interaction modes. The Find Your Deputy assessment tests whether the apex of your structure holds a genuine second or only the appearance of one. Both are available at theconemodel.com.
The connecting principle is the same one Team Topologies used to reshape team design: structural problems require structural solutions. Loneliness at the top is not a character flaw to be coached away. It is a design condition, as diagnosable and as solvable as a team crushed under unsupported cognitive load. Team Topologies proved the method works at the team level. The Cone Model applies the same logic one layer higher.
Conclusion
Team Topologies offers a genuinely powerful lens for solving one of engineering leadership's hardest problems: how to organize people so that systems and teams can both thrive. The framework earns its reputation by naming cognitive load as a first-class concern, providing clear interaction modes, and giving teams a shared language to reason about structure. At the same time, practitioners must go in with open eyes. The model is a starting point, not a complete playbook; real organizations will encounter gaps around culture, politics, and continuous evolution.
The takeaway is not to adopt or reject the framework wholesale. Instead, take what it gets right, stay honest about what it leaves open, and build from there.
Start by mapping your current team interactions. Identify where cognitive load is silently killing momentum. Then use the vocabulary to spark better conversations. Progress begins with clarity.