Expert on every file. Why our anti-pyramid model changes deliverable quality.
The pyramid model is the consulting industry's default. The partner sells the work, the manager supervises it, and the analysts and associates produce most of what the client actually sees. Our model is different by design. Expert on every file means the person who advises is the person who does the work, and the deliverable the client receives is shaped by the judgement of an operator with twelve to twenty years of practice rather than the translation of that judgement through three intermediate layers.
What the pyramid model produces.
The pyramid model is the engine of the global consulting industry, and it works for the type of work the industry was built to serve. Programme rollouts where the methodology is well understood, where the team needs to scale to thirty or fifty people, and where the partner's role is to set the frame and the bench's role is to populate the plan. For that kind of work, the pyramid is efficient. The expert judgement is encoded in the methodology, the manager carries it, and the bench produces the artefacts.
The model breaks on transformation work where the judgement is the deliverable. AI strategy, operating model redesign, portfolio prioritisation, governance work, sovereign cloud arbitration. The conversations that produce value in these engagements are the ones where the partner is in the room with the right context, the right scars from previous engagements, and the authority to make calls. When the partner has set the frame and stepped out, the work continues but the judgement does not.
The translation loss is the part the client experiences most directly. The expert insight that won the engagement is filtered through a manager who heard it secondhand, an associate who heard it thirdhand, and a slide that distils it past the point of resonance. The client knows. The follow-up meeting is the one where the partner is asked back to repeat the conversation, because the original judgement has not survived the chain.
The B One model.
Our model is founder-led by design and structured to keep judgement at the centre of every file. The partner who sells the work is the partner who runs it. The expert who advises in the first conversation is the expert in the room at the steering committee three months later. The deliverable the client receives is shaped by an operator who has spent twelve to twenty years building the practice they are now using, not by the translation of that practice through three layers.
The model is supported by a small bench of expert partners and operating partners, each of whom carries files independently and runs them end to end. Where additional capacity is needed for production work, the Tech Factory engine provides it from our Bali and Paris benches, but the partner remains the operating owner of the file and the judgement remains in their hands. The structure looks unconventional from the outside. From the inside, it is the simplest way we know to produce the kind of deliverable our clients hired us to make.
The hiring that sustains the model is indexed to the work. We grow when the right experts are available and when the pipeline of work justifies it. We do not grow because the prior year produced revenue. The discipline costs us volume. It earns us the bench depth that the model requires.
The expert who advises in the first conversation is the expert in the room at the steering committee three months later. The deliverable is shaped by that operator's practice, not by the translation of it through three layers.
What changes for the client.
Four things change for the client compared to the pyramid model, and the four together explain why clients who have worked with us once tend to come back rather than test the market. The first is continuity of judgement. The partner who hears the brief is the partner who shapes the response. There is no information loss between the partner who sold the work and the team who runs it, because they are the same people.
The second is the pace of decisions. In the pyramid model, the manager has to escalate a meaningful question to the partner, the partner has to recover the context, and the answer comes back two days later filtered through one more layer. In our model, the partner is in the meeting where the question arises. The answer is given on the spot, with the full context that the partner has been carrying from the start.
The third is fewer translation layers. The deliverables our clients receive read like documents an experienced operator would write, because that is who wrote them. The reasoning is direct, the trade-offs are explicit, the recommendations are owned. The polish is functional, not theatrical. Clients tell us regularly that the deliverable looks lighter than what they expected and reads heavier than what they are used to, and the second part is the part they remember.
The fourth is accountability. When the partner who sold the work is the partner who runs it, the line of accountability does not break. If the engagement does not land, the partner cannot defer to the bench. If it lands well, the partner cannot claim it without having been in the work. The model removes the ambiguity that the pyramid model relies on to manage risk.
What this requires from us.
The model is operationally hard to run. It requires a hiring profile that is heavily weighted to expert practitioners, an economic model that does not depend on leverage, and a sales discipline that does not over-sell relative to the bench available to deliver. We have built each of these deliberately, and each carries a cost that volume-driven firms do not have to pay.
Bench discipline is the most important. We accept fewer engagements than we could sign, because we will not run an engagement without a partner of the right profile on the file. The opportunity cost of that discipline is visible. The benefit is that the engagements we do run land with the quality our clients hire us for, and the reputation effect compounds.
Indexed hiring is the second. We hire when the right experts are available and the work justifies it. We do not hire to fill an attrition gap or to match a competitor's headcount. The hiring conversation is slower, the profiles are more specific, and the offers are made to candidates who have the practice depth our model requires.
Careful capacity is the third. We watch the load each partner carries and decline new engagements when the bench is close to full. The discipline is unglamorous. It is also the only way the model holds.
Why we keep it hard to scale.
We are sometimes asked why we do not move to a hybrid model that would let us grow faster. The honest answer is that the hybrid model breaks the promise the firm is built on. The moment we accept that some files will be carried by a manager rather than a partner, the pyramid logic creeps back in, and the deliverable quality we defend with our clients moves with it. The constraint is the point.
The size of the firm will grow over time, because we hire the experts who want to do this kind of work and we win the kind of engagement the model serves. The shape of the firm will not change. The four offices in Paris, Dubai, Singapore and Bali will deepen rather than multiply, and the bench will grow in expert capacity rather than in junior volume.
If you are weighing your next consulting engagement and the question of who will actually do the work is sitting in the back of the conversation, the move we recommend is to ask explicitly. Who will sit in the steering committee in three months? Who will write the recommendation slide? Who will be in the room when the call gets made? The honest answer to those three questions tells you which model you are buying. The brief form below routes to the partners who will run the work, not to a relationship manager.
Frequently asked questions.
Does the model cost more than a pyramid firm?
On a daily-rate basis, the partner days cost more than a bench-heavy mix. On a total-engagement basis, the answer depends on the work. For transformation engagements where judgement is the deliverable, our model usually costs less, because the engagement runs shorter and produces fewer translation cycles. For volume rollouts, the pyramid model is cheaper, and we do not compete for that work.
How do you handle capacity when the bench is small?
By accepting fewer engagements than the pipeline allows, and by being explicit with clients about when we have capacity. The pricing reflects the capacity constraint. The transparency is what allows the conversation to be honest and the relationship to last.
What if more capacity is genuinely needed mid-engagement?
The Tech Factory engine provides additional capacity for production work without breaking the model, because the partner remains the operating owner of the file. When the additional capacity needed is also expert judgement, we bring in a second partner rather than escalate a manager. The cost is real and the conversation with the client is open.
How does this work on multi-year rollouts?
By staging the engagement so that the partner is on the file at the moments where judgement matters and the Tech Factory carries the production work in between. The model is not designed for engagements that need a permanent thirty-person team on the ground. For those engagements, the pyramid firms are better placed and we say so.
How do partners move between offices?
By design. The four offices are not local franchises. Partners run files that take them between Paris, Dubai, Singapore and Bali as the engagement requires. The bench mobility is what keeps the firm coherent across geographies and what allows a client in one office to draw on the practice of another.
What happens when a founding partner steps back?
Succession is treated as a long-horizon question. The firm is built around a small group of partners who carry the practice, and new partners are added on a slow rhythm tied to demonstrated capacity to run files in the same model. The size of the firm will never be a function of growth-target pressure.
Where this lands
How we'd take this further with you.
Writing is one thing. Shipping is the other. Selected work from the partners writing here.
See the work

