AI Strategy & Insight | Brooklyn Solutions | Part 2 of a Series
Nick Francis · Co-Founder & CTO, Brooklyn Solutions · June 2026
Previous in this series: Before You Chase AI, Understand Your Processes →
Leading consultancies flagged it at the start of 2025. The data has only sharpened since. Procurement transformation programmes are failing at alarming rates — and the root causes have nothing to do with the technology. If organisations don’t fix what’s broken here, they will fail at AI for exactly the same reasons.
In the first piece in this series, I wrote about the importance of understanding your processes before chasing AI. That argument came from 25 years of watching organisations rush toward each successive technology wave — mobile, cloud, digital transformation — without doing the foundational work that makes any of those investments actually land.
Procurement is where that argument becomes uncomfortably concrete. Because procurement transformation has become one of the most thoroughly researched, extensively funded, and consistently disappointing areas in enterprise technology. And the reasons it fails tell you almost everything you need to know about why AI will fail in the same organisations, for the same reasons, if nothing changes.
What the Research Tells Us
Across multiple reports published by leading consultancies and analysts between 2025 and 2026 — covering hundreds of procurement organisations across financial services, advanced industries, life sciences, utilities, and the public sector — a consistent picture emerges. Procurement transformation failure rates sit between 70% and 84% of programmes that do not achieve their original objectives [1,2]. In the past year alone, the failure rate for AI-specific procurement initiatives rose from 17% to 42% — more than doubling in a single year [3].
Perhaps most striking is a finding from the Kienbaum 2025 study on organisational transformative capacity [4]: the higher the seniority of the leader, the more optimistic their assessment of how well the transformation is going. Boards and C-suites systematically overestimate their organisation’s ability to change. Project managers and programme teams — the people doing the work — see reality far more clearly.
“AI did not create the operating tensions that are now surfacing in procurement. It accelerated the consequences of leaving them unaddressed.”
— Procurement Insights / GEP AI Readiness analysis, 2025–2026 [6]
This perception gap is not a curiosity. It directly produces unrealistic timelines, under-resourced programmes, and the kind of inflated expectations that make failure almost inevitable before the first sprint is run.
The Three Root Causes I See in the Field
The research corroborates what I see consistently in our own transformation programmes — across defence, financial services, and utilities clients — where we work with enterprise procurement teams at various stages of their transformation journey.
The same three failure patterns appear, almost without exception.
Programme Design: Overreach Without Traction
Transformation programmes are routinely designed to do everything at once. They aim for the complete future state from day one, striving always for “more” rather than building toward it. The result is a programme that is perpetually at risk — always over budget, always behind schedule, always trying to absorb the next requirement before the last one has stabilised. What’s missing is a genuine MVP discipline: the willingness to define a lean, right-sized first release that creates real traction and delivers demonstrable value before expanding scope.
Frameworks like Agile / Scrum and Lean exist precisely for this — not as delivery methodologies bolted on after design, but as principles that should shape how a programme is conceived. The organisations that succeed are the ones that have accepted continuous, iterative change as the operating model — not a project they’re waiting to complete.
Governance: No Design Authority, No Ownership
This is the one that surprises me most — because I spent much of my early career in investment banking, where design authority is an understood and non-negotiable concept. Someone has to own the architecture of both the technology and the business process. Someone has to be accountable for the integrity of the system design as the programme evolves. Without that, every stakeholder pulls in a different direction, every team makes local decisions that contradict decisions made elsewhere, and the programme gradually loses coherence.
I also see a persistent failure around stewardship: individual process steps need named owners who are accountable for keeping those steps current, well-defined, and optimised. This is a core principle in Lean Six Sigma — every step has an owner, every owner has accountability, and the process map is a living document. Without governance structures that define decision-making authority clearly, and without a design authority that sits above programme politics, transformation drifts.
Capability and Clarity: The People Gap Nobody Budgets For
Research consistently shows that 80–90% of transformation budgets flow into technology. Under 2% goes into capability building — the reskilling, coaching, process redesign, and role clarity that determines whether the people in the organisation can actually operate what the technology enables. This is the most predictable failure in the field. Organisations hire for the “run” phase before they’ve properly designed it.
Defining what the right roles look like — who runs the process, who optimises it, who supports it, who has final sign-off — needs to happen before deployment, not after. The RACI model (Responsible, Accountable, Consulted, Informed) is a basic but essential tool here: it forces the conversation about ownership before ambiguity becomes paralysis.
The Primer Process Principle
Cutting across all three failure patterns is a fourth, often underappreciated mistake: attempting to transform everything at once. We consistently see organisations wanting to apply AI and agentic capabilities to all of their priority-one processes simultaneously — a big-bang approach that bypasses any meaningful proof-of-concept discipline and sets programmes up to fail before they’ve found their footing.
The argument for starting small isn’t just about managing risk, though it does that too. It’s about something more fundamental: the first process you transform should serve as a primer for the factory that will eventually transform all the others.
The goal of your first transformation isn’t just to prove you can transform a process. It’s to prime and flush the entire machinery — the programme, the governance, the change management — so that you know how to do it before you do it at scale.
When an organisation selects a well-bounded, manageable first process, several things happen in parallel. The team learns how to map a process properly. They discover where their data is clean and where it isn’t. They stress-test their governance model in a lower-risk environment. They validate their change management approach with real users. And critically — they generate a success story and a blueprint that can be carried, with confidence, into the next process and the one after that.
Jumping straight to the most complex, most politically sensitive, most cross-functional processes with a team that has never delivered an AI-enabled transformation before is, at best, optimistic. What looks like ambition is often just an absence of sequencing discipline. The organisations that scale AI successfully are almost always the ones that started with a single, well-chosen process, proved the value, documented what worked, and then moved — with a blueprint in hand — to the next.
Choose your first transformation process not because it has the highest potential return, but because it is manageable enough to succeed, meaningful enough to matter, and simple enough to teach your organisation how to transform. That first success is the blueprint for everything that follows. Without it, you are asking every subsequent process to be both the pilot and the production run simultaneously.
Why This Directly Blocks AI Adoption
In the first blog in this series, I made the argument that you cannot effectively adopt AI without first understanding your processes end-to-end: the individual tasks, the inputs and outputs at each step, what “good” looks like, and what data you have to evidence it. That argument stands. But procurement transformation failure adds a second, more urgent dimension.
It’s not just that organisations don’t understand their processes. It’s that the three failure modes above — overreaching programmes, absent governance, and under-invested capability — mean that organisations often don’t have the internal infrastructure to transform at all. And AI doesn’t fix that. It exposes it.
Layering AI onto a transformation programme that lacks design authority, process ownership, and capability investment doesn’t accelerate the programme. It accelerates the failure. S&P Global Market Intelligence data shows the failure rate for AI-specific procurement initiatives rose from 17% to 42% in a single year [3]. That’s not an AI problem. That’s an organisational readiness problem wearing an AI label.
Agentic AI — the kind of AI that doesn’t just answer questions but takes actions, monitors obligations, raises alerts, and coordinates across systems — requires even more rigour than conventional automation. An agent needs to know exactly what step it is executing, exactly what the success condition is, exactly when to escalate to a human, and exactly who that human is. If your programme doesn’t have those things defined for a human-operated process, it certainly can’t define them for an agent.
This is something we’ve built into Ask Brooklyn’s architecture from the ground up. Our BISO-28 AI governance policy, the centralised logging of every prompt, decision context, and model response, and our human-in-the-loop design aren’t compliance overhead. They’re the proof that the underlying process is well enough understood to be governed. If you can’t govern it, you don’t understand it. If you don’t understand it, you’re not ready.
What Good Actually Looks Like
The organisations that navigate procurement transformation successfully share a consistent set of characteristics. They are not, notably, the ones with the largest budgets or the most sophisticated technology stacks.
- ✓Adopt an MVP-first approach. They define a lean first release, create genuine traction, demonstrate value, and use that momentum to expand scope. They treat change as continuous, not as a project with a completion date.
- ✓Establish design authority early. Someone owns the architecture of the process and the technology. That person has the authority to make decisions and the accountability to defend them. Programme politics don’t override design integrity.
- ✓Invest in capability proportionally. They resist the temptation to spend 90% of the budget on the platform and expect the people to figure it out. Role design, capability development, and process stewardship are treated as deliverables — not assumptions.
- ✓Map process ownership explicitly. Using frameworks like Lean Six Sigma and RACI, they ensure every step has a named owner, every owner has clear accountability, and the process map is maintained as a living document.
- ✓Apply a formal AI suitability assessment before deploying AI. Using a structured prioritisation framework — assessing task frequency, data availability, complexity, and regulatory risk — they identify which specific tasks are genuinely ready for AI augmentation versus which ones need process work first.
The Uncomfortable Truth About “Transformation”
The word “transformation” has become a problem. It implies a journey from one fixed state to another — a beginning, a middle, and an end. Organisations design programmes with that shape, budget for that shape, and then wonder why the outcome doesn’t hold once the programme closes.
The organisations that extract sustained value from technology — whether from ERP, from digital procurement platforms, or increasingly from AI — are the ones that have accepted that change is not a programme. It’s an operating model. The elaboration of requirements, the refinement of processes, the optimisation of AI agents: these don’t stop at go-live. They’re the ongoing work of a mature, well-governed function.
Change once it’s been done is a ridiculous concept. The organisations winning with technology are the ones that have made continuous improvement the way they work — not a project they’re waiting to finish.
This is as true for AI as it was for every technology wave before it. The difference is that AI, particularly agentic AI, makes the gap between well-governed and poorly-governed organisations more visible, more quickly, and with more significant consequences.
Looking Ahead
The next piece in this series will go deeper into what AI readiness actually looks like in a procurement and supplier management context: how to assess which processes are genuinely ready, how to sequence AI deployment to build rather than undermine confidence, and how human-in-the-loop governance translates from principle to practice.
For now, the question every CPO, CIO, and CTO needs to be honest about is this: if your procurement transformation programme is struggling — over budget, behind schedule, not delivering the traction it promised — is that a technology problem? Or is it one of the three things above?
Because whichever it is, AI won’t solve it. And without solving it, AI will make it worse.
Is your procurement transformation AI-ready?
Brooklyn Solutions works with enterprise procurement teams to assess process maturity, establish governance, and sequence AI deployment safely — from first use case to agentic AI at scale.
About the Author
Nick Francis, Chief Technology and Marketing Officer at Brooklyn Solutions
Nick Francis is a well-established and experienced CxO delivering Digital & Security-focused Transformation through the design, build, and deployment of cost-effective, highly automated industry-leading solutions. Nick has experience working across the private and public sectors in industries such as Financial Services, Insurance, Legal, Utilities, Retail, Public Sector and Government. Specialised in transformation activity to optimise processes, operational expenditure, and increase productivity. Significant experience in compliance, risk & control activities in highly regulated industries, standardisation of technologies, streamlining of internal processes and continuous improvement driving consistency and efficiency across an organisation whilst holding Customer, Colleague and Partner experience at a premium.
Research references
- McKinsey & Company — multiple reports 2021–2025, including “Transforming Procurement Functions for an AI-Driven World” (February 2025), survey of 300+ procurement leaders. Consistently report ~70% of digital transformation programmes fail to deliver meaningful results.
- Bain & Company — 2024 business transformation study: 88% of business transformations fail to achieve their original ambitions. BCG similarly reports 70–75% failure rates across digital transformation programmes. Walden University doctoral research (2026) corroborates a range of 66–84%.
- S&P Global Market Intelligence / World of Procurement (Substack, August 2025): AI procurement initiative failure rate rose from 17% to 42% in a single year.
- Kienbaum — 2025 study on transformative capacity of organisations (reported via SI Labs, March 2026): C-suite and board-level leaders systematically overestimate transformation capability. Separately: 80–90% of transformation budgets flow into technology; under 2% into capability building.
- Prosci — Best Practices in Change Management, 12th Edition: Initiatives with excellent change management are 6–7 times more likely to meet their objectives.
- Procurement Insights / GEP AI Readiness analysis (2025–2026): “AI did not create the operating tensions… AI accelerated the consequences of leaving them unaddressed.”