Three horizons. Most organisations only manage one.
When you ask your leadership team what return they expect from AI adoption, you are likely to get one of two answers: either a confident efficiency number (hours saved, costs reduced, throughput increased) or an uncomfortable silence dressed up as "It's difficult to calculate, so we are still exploring it."
Calculating the real ROI of AI adoption, not simply as a new disruptive technology but as a new value-creation paradigm, requires your leadership team to hold three very different definitions of success simultaneously:
What efficiency can we capture now?
What value can we create for our stakeholders in the medium term?
And the one that most teams quietly avoid: why does our organisation exist in a world where AI will be doing more of what we used to do?
No sector is being forced to answer all three of those questions more publicly than universities. That is why they are worth paying attention to, not as a model to copy, but as an early warning system for what is coming.
THE THREE HORIZONS OF AI ADOPTION ROI
Most AI adoption ROI conversations are built around the horizon one: efficiency and productivity. That is where the business case is easiest to make, the wins are fastest to show, and the board is easiest to satisfy. Efficiency gains are real, they matter, but they are not a strategy for business success.
Look at universities right now. They are under serious pressure as margins are shrinking, costs keep rising, and public funding is under greater scrutiny. AI adoption is legitimately the way to help them manage that pressure by automating administrative workflows, streamlining student services, and trimming operational overhead: measurable, defensible, necessary.
This is where most businesses would focus, but universities are largely focused on horizon two, and I explain why below. They are trying to save their existing business model: How do you assess students when AI can write the essay? What is the value of a lecture when the content is freely available and infinitely personalisable? What does research mean when AI can synthesise literature, generate hypotheses, and accelerate analysis? These are not future concerns. They are live operational problems sitting on the medium-term horizon.
And the reason universities are prioritising horizon two over horizon one goes back to their historical strength and largely unchallenged position. Universities didn't have to be efficient; most are not private institutions under pressure to turn a profit, so they haven't had to build muscle for productivity gains. The medium-term question of how they maintain stakeholder trust among students, employers, governments, and academics takes priority. They need to ensure assessments work, lectures remain valuable, and research is still seen as a distinctly human endeavour.
Most businesses can extract enough value from horizon-one efficiency gains for now without confronting horizon two.
And then, underneath both horizons, is the long-term question universities can no longer defer- horizon three: why should someone go to university at all? When the future is genuinely unclear, and AI will transform most of the careers graduates are preparing for, the notion that "university exists to prepare you for your future" becomes heavily contested. No amount of short-term efficiency or revised assessment models can resolve this.
Read that third horizon question through a business lens:
In a world heavily augmented by AI, why should someone use my organisation and what it offers, at all?
McKinsey's State of AI 2025 found that while 88% of organisations now use AI in at least one business function, only 6% qualify as genuine high performers generating meaningful bottom-line impact. That is not a deployment problem. It is a definition problem. Organisations are measuring AI ROI against the horizon they were already managing, and calling that the whole picture.
Universities expose this gap most visibly. An institution that successfully redesigns its assessment models while students are questioning the value of tuition is not winning. It is losing more slowly. If the long-term purpose question is never answered, the efficiency gains are just buying time.
Businesses face exactly the same dynamic, just with different labels on the horizon.
Horizon one: where can AI reduce cost and accelerate output?
Horizon two: how do we redesign our value proposition when AI is changing what customers actually require from us?
Horizon three: what is this organisation for in a world where AI is doing more of what we built it to do?
Three horizons. Three different definitions of success. All live at the same time.
The organisations that are dodging the third question are just deferring the hardest part of the calculation.
We built our institution as AI-native from inception, using AI as a lens for curriculum and assessment (horizon two) and for operational optimisation (horizon one). And our daily conversation is not about tools or efficiency. It is about why we should exist at all, and what future of the university we are trying to create. That question puts enormous pressure on every choice we make. We are learning to move on ground that is not stable, and never will be. It is the most uncomfortable position I have occupied as a leader.
AI ADOPTION ROI REQUIRES LEADERSHIP TEAMS THAT CAN HOLD ALL THREE HORIZONS AT ONCE
The practical implication is uncomfortable but clear: measuring AI adoption ROI properly means your leadership team must work across all three horizons simultaneously, not sequentially.
That means:
Horizon one: capturing short-term efficiency wins while actively asking which of those efficiencies are building toward something, and which are simply buying time.
Horizon two: addressing the medium-term disruption to your value proposition before it becomes a crisis, rethinking what you deliver to customers, employees, and partners in a world where AI is changing what they are willing to pay for.
Horizon three: naming the long-term purpose question out loud, even when the answer is not yet clear, because the cost of not asking it is higher than the discomfort of sitting with uncertainty.
Watch university leadership teams right now. See who is holding space for all three horizons simultaneously: cutting costs today, redesigning teaching and assessment for the medium term, and genuinely interrogating what a university is for when the world their graduates are entering is being remade by AI.
Universities cannot continue doing what they do today. That is now a given. The ones that survive and matter will be the ones that figure out what they are for in the world that is actually arriving.
The same is true for your organisation.
The question is not whether AI delivers ROI. It does. The question is whether you are measuring the right return, across all three horizons, before it is too late.