The Big Three Consultancies Are Selling You PowerPoint and Calling It Strategy
There is a slide we have seen, in some form, in roughly forty pitches from McKinsey, Deloitte, BCG, Accenture, and the rest of the cohort that calls itself “strategy.” The slide has a 2x2 matrix. The axes are “AI maturity” and “business value.” There are four quadrants, each labeled with an aspirational adjective: “Foundational,” “Transformative,” “Differentiating,” “Industry-Leading.” The client logo sits in the bottom-left quadrant. An arrow points diagonally to the upper-right. The arrow is labeled “Our Recommended Journey.”
The fee for the engagement that produced this slide is usually somewhere between two and twenty million dollars.
The slide is not strategy. The slide is a billing artifact. It is the deliverable shape that has emerged from forty years of charging hourly rates to produce documents that nobody can execute. We are not the first people to notice this. We are noticing it again because, in the AI era, the gap between what these firms sell and what their clients actually need has widened to the point of farce.
What the Big Three Are Optimized to Produce
A traditional management consultancy is optimized for one thing: producing artifacts that survive a board presentation. The artifacts are slide decks, frameworks, maturity models, and recommended roadmaps. They are designed to be read by senior executives who have ninety seconds per slide and to be defensible against challenge from inside the organization.
This optimization has structural consequences for what the firms can and cannot do.
They can produce documents quickly. A team of fifteen analysts can generate a hundred-slide deck in six weeks. The deck will be polished, internally consistent, and built on the same conceptual framework the firm has been refining for a decade.
They can supply a narrative. The narrative will sound new – because the labels rotate – and will be structurally familiar. There will be an industry context, a set of challenges, a recommended approach, and a phased implementation. The narrative is reusable across clients because the structure is reusable.
They cannot ship code. They cannot operate production systems. They cannot make architectural decisions accountable to the engineering reality of the client’s stack, because they do not have the engineers in the room who would have to live with those decisions for the next five years.
When the work being purchased is producing a strategy document, the Big Three are competent. When the work being purchased is changing how the organization operates, they are not. They never were. This was tolerable when the strategy document was the primary deliverable. It has become catastrophic now that the strategy document is, increasingly, the wrong artifact.
Why AI Has Made the Mismatch Worse
The Big Three sell their AI practices as the future. Most of those practices are sales channels for the same slide-based engagement, with AI as the topic. The maturity model is now an “AI maturity model.” The roadmap is now an “AI transformation roadmap.” The implementation phase is now scoped to start “after the strategy phase concludes,” which is the standard structure of every engagement these firms have ever run.
The structural problem is that AI delivery is not a strategy problem. It is an execution problem. The strategy is usually obvious: identify the workflows that have the highest leverage, build agentic systems against them, measure the outcomes, expand. The hard work is not deciding what to do. The hard work is doing it – specifying the workflows, building the agents, integrating with production systems, training the team, surviving the operational reality.
The Big Three are constitutionally incapable of doing the hard work. They will subcontract it. They will recommend a partner. They will write the integration plan. They will not, themselves, ship the integration. By the time the strategy phase is over, the client has paid eight figures for a document and is now starting the implementation from a standing position with a strategy that may or may not survive contact with their actual production environment.
We have seen this end three ways. The implementation gets handed to a different firm, which discards the document and starts over. The implementation gets handed to an internal team that interprets the document differently than the consultants intended. The implementation never starts, the executive who signed the engagement moves on, and the strategy enters the slow death of “we are working on revising it.”
What Is Actually Required
The work the client paid for – the AI capability that produces measurable business value – requires four things the Big Three are structurally unable to provide.
Engineers in the room from day one. Not as advisors, not as “subject matter experts” parachuted in from a vendor relationship, but as the same humans who will be on the keyboard when the system gets built. The strategy and the implementation cannot be sequential. They have to be the same conversation, run by people who can do both.
Production-grade specification. A strategy document that says “build a customer-service agent” is not actionable. A specification that says “the customer-service agent ingests Zendesk ticket text, classifies into thirty-seven action categories, drafts responses for the seventeen action types we have approved, escalates the rest to a named queue, and is measured on first-contact resolution rate and customer-sat delta against a control population” is actionable. The Big Three do not write the second kind of document, because the second kind requires the writers to know how the system will actually be built.
Operational ownership of the result. The person who writes the strategy has to be accountable to the operational outcome. If they are not – if the strategy phase ends and the strategists go home and a different team is on the hook for whether it worked – the strategy will be optimized for the document’s defensibility, not for the operation’s success.
A willingness to be wrong on the record. The kind of AI work that produces outcomes requires you to try things, watch them fail, and adjust quickly. That is incompatible with the consultancy contract structure, which is designed to ratify the strategy as correct and bill against the implementation of the correct strategy. The contract structure cannot accommodate “we were wrong about this part, here is what we are doing instead.”
The Vendors That Replace Them
The firms that have started to displace the Big Three in AI delivery share an opposite shape. They are smaller. They are operator-led. They have engineers on every engagement, on day one, who will personally ship the implementation. They charge for outcomes, not for documents. They are willing to be wrong on the record, because their contracts pay them for the result, not for the artifact.
Their pitch is not a 2x2 with an arrow. Their pitch is “here is the workflow we will instrument first, here is the outcome we will commit to, here is the timeline, and here is the team that will be in the room every day until it ships.” It is less impressive in a boardroom and more useful in a quarterly business review.
There are still plenty of clients buying the document. The document still gets purchased. The clients who are buying outcomes are buying them from a different population of firms, and the population is growing.
The Honest Statement
If your AI transformation is being led by a firm whose primary artifact is a slide deck, you are paying for the wrong thing. The slide deck is not the work. The slide deck is the bill.
The firms that will help you build something that survives next quarter’s planning cycle are the firms whose engineers are in the room writing specifications, building agents, and reviewing pull requests with your team. That is a smaller industry than the strategy consulting industry. It is also the part of the industry that ships.
Buy the work. Not the document.
