Vibe-based AI spending is over

For the past three years, a significant portion of enterprise AI investment has been driven by something closer to competitive anxiety than strategic clarity. The tools were impressive. The demos were convincing. The fear of being left behind was real. What was often missing was a clear answer to the simplest possible question: what, specifically, are we trying to change - and how will we know if it’s changed?

That era is ending. Not because the anxiety has gone away, but because the patience for investment without evidence is running out. Boards are asking harder questions. CFOs are requiring more rigorous justification for continued spend. And the organisations that built their AI strategy on enthusiasm rather than outcomes are finding themselves in an uncomfortable position.

What ‘vibe-based’ actually means

The term sounds harsh, but it describes something real and recognisable. Vibe-based AI spending is characterised by investment decisions driven primarily by vendor demonstrations, competitive pressure, or executive enthusiasm - without a clear framework for what success looks like or how it will be measured.

It shows up in a few consistent patterns. Pilots that never scale, not because they failed, but because no one defined what ‘success’ meant precisely enough to know whether to proceed. Adoption metrics that go up while business outcomes stay flat. A portfolio of AI tools with overlapping capabilities and unclear ownership. An inability to answer, with specificity, what the organisation’s AI investment has changed.

“The shift happening now is from measuring activity to measuring impact. They are not the same thing - and the gap between them is where most AI value disappears.”

Research from the Larridin State of Enterprise AI report puts a number on the scale of the problem: 72% of enterprise AI spend is reportedly destroying value through waste. Not through bad intentions or poor technology, but through the absence of the measurement and governance infrastructure needed to know whether the investment is working.

The shift to outcome measurement

The organisations pulling ahead share a common characteristic: they define outcomes before they choose tools. Not ‘we want to implement AI in our finance function’ but ‘we want to reduce the time from invoice receipt to payment by 35%’. The AI investment follows from the outcome definition. The measurement framework is built before deployment, not retrofitted after.

This sounds straightforward. In practice it requires a level of strategic clarity that many organisations haven’t developed. It means being honest about which workflows are genuinely ripe for AI impact and which are being included because they’re visible or because someone is enthusiastic about them. It means defining baselines before you start, so you can measure change rather than just declare it.

It also means being willing to stop. One of the clearest signals of a mature AI organisation is the ability to kill initiatives that aren’t delivering - not because the technology is bad, but because the outcome isn’t materialising and the evidence doesn’t support continuing. Vibe-based spending is characterised by sunk-cost thinking. Outcome-based spending is characterised by honest evaluation.

What leaders who can prove ROI are doing differently

Across the organisations seeing consistent, measurable returns from AI, a few practices stand out.

They separate exploration from exploitation. Exploratory AI work - pilots, proofs of concept, capability building - is funded and governed differently from operational AI investment. The former is expected to produce learning; the latter is expected to produce returns. Mixing the two is where measurement gets muddled.

They build measurement before they build capability. Before an AI system goes live, the team can answer: what metric is this moving, what’s the baseline, how will we track change, and at what point would we decide it’s not working? These questions are harder to answer than they sound - which is exactly why most organisations skip them.

They account for total cost, not just tool cost. AI ROI calculations frequently undercount the cost of change management, integration, quality assurance, and the human time spent working around AI limitations. Organisations seeing genuine returns build these costs into their models from the start rather than discovering them after deployment.

They report on outcomes, not activities. The metric that goes to the board is not ‘number of AI tools deployed’ or ‘percentage of employees using AI’. It is the specific business outcome the investment was designed to move. This creates accountability at every level of the organisation and forces the honesty that activity metrics allow you to avoid.

The practical starting point

If you’re looking to shift from vibe-based to outcome-based AI investment, the most useful place to start is an honest audit of your current portfolio. For each significant AI initiative, ask three questions: What outcome was this meant to produce? Do we have a baseline measurement from before deployment? Can we demonstrate, with data, that the outcome has moved?

The answers will tell you a great deal about where you actually are. For the initiatives where the answers are clear, double down. For the ones where they’re not, either build the measurement infrastructure or seriously consider whether the initiative is worth continuing.

The shift from adoption metrics to outcome metrics isn’t just about satisfying the board. It’s about actually getting the returns that AI investment promises - and building the organisational capability to keep getting them.

AI adoption numbers are easy to generate. AI outcomes that change the business are hard - and require a fundamentally different approach to investment, measurement, and governance. The organisations making that shift now are building durable advantage. The ones that don’t will be having difficult budget conversations sooner than they expect.