The AI Strategy No One Can Measure (Yet)
Why AI activity is easier to see than AI value
March, 2026
The interesting problem with AI measurement right now is that most organisations are not short of data.
They are short of evidence.
They can count licences, logins, prompts, active users, pilot numbers and time-saved anecdotes. They can show adoption dashboards. They can point to teams using copilots, assistants, workflow tools and AI-enabled features across the business.
They can usually produce a slide proving that activity is increasing.
Lovely.
But the harder question is still sitting there, looking mildly inconvenient:
did any of this change an important decision?
That is the confidence gap underneath a lot of AI strategy right now.
The practical question is more specific:
what is AI improving, who owns the outcome, and can the organisation stand behind the result?
The Shift
This is less about one tool than the spread of AI across everyday work.
Enterprise copilots, GenAI assistants, AI-enabled workflow tools and embedded features are helping people draft, summarise, analyse, compare, search, recommend and decide faster.
That matters because these tools create a lot of visible activity.
Someone used the assistant. A team generated summaries. A pilot reduced drafting time. A workflow produced recommendations. A dashboard showed usage going up and to the right, which is always comforting and occasionally meaningful.
But activity is not the same as value.
A tool can be popular and still have a vague relationship with the decision it was meant to improve.
A team can save time on drafting while the approval still sits in someone’s inbox for two weeks.
A model can produce a useful recommendation that nobody trusts enough to act on.
A pilot can look successful because it made part of the process faster while leaving the actual business decision untouched.
That is why the next useful shift in AI measurement is not another dashboard of tool usage.
It is a better view of what happens between AI-assisted work and an actual decision.
The Real Tension
Adoption data can be reassuring for a while.
A high licence activation rate can make an AI programme look healthy. So can prompt volumes, pilot counts, internal showcases and enthusiastic stories from early adopters.
Some of those signals are useful. None should be dismissed.
But they do not answer the question a board, CFO, executive team or business owner will eventually ask:
what changed?
That question has a way of clearing the room slightly.
Often, the value is probably there. A customer issue was resolved with better context. A manager had a clearer view before making a call. A piece of analysis that once took days took hours.
But if the organisation cannot trace that value to a decision, an outcome or an accountable owner, the story becomes hard to defend.
This is where AI strategy can start to wobble.
The technology may be working. The dashboard may be accurate. The pilot may have genuinely helped.
But if the dashboard can tell you who used the tool and not what decision improved, it may be measuring attendance, not value.
A neat little trap, really.
Shadow AI makes this harder. Useful work may be happening in tools the organisation cannot see, while the official dashboard confidently reports on the smaller system it can.
That is not measurement.
That is a torch pointed at only part of the room.
The Ripple Insight
The real gap is not measurement first.
It is ownership.
Someone buys the licence. Someone runs the pilot. Someone reports adoption. Someone else is expected to realise the benefit.
Then, when scrutiny increases, everyone can explain their part.
No one can quite explain the outcome.
That is not just a measurement problem.
It is an operating model problem.
Name the decision, not the tool.
If AI is supposed to improve customer retention, reduce risk, sharpen pricing, speed up triage, support forecasting or improve service recovery, then the relevant business leader needs to own the decision AI is meant to influence.
Otherwise, AI becomes another layer of activity sitting across an already complicated organisation.
More tools. More dashboards. More governance meetings.
Not necessarily more confidence.
The useful question is not:
how much AI did we use?
It is:
what decision changed, who owns it, and can we stand behind the result?
That question moves AI strategy away from technology consumption and toward decisions someone actually owns.
AI does not create value in the abstract. It creates value when it changes a decision, improves a process, catches a risk earlier, reduces rework, or helps a team act with better judgement.
If none of those things can be named, measured or owned, the strategy may still be active.
It may not yet be useful.
The Move
Before the next AI steering committee, board update or investment case, pick one important decision stream.
Not the whole AI portfolio. That way lies a long workshop and a very tired room.
One decision stream.
A customer escalation. A pricing approval. A field response. A claims decision. A campaign allocation. A workforce planning call. Something repeatable and close enough to the operating model to reveal the truth.
Then ask four questions.
How was the decision made before AI?
Where is AI now involved?
What changed in speed, quality, rework, escalation or confidence?
Who owns the result?
The point is not to build a perfect measurement system in one quarter. Good luck to anyone attempting that with a straight face.
The point is to create a sharper link between AI activity and organisational consequence.
Some AI use will prove valuable. Some will not. Some will simply show leaders where the organisation had learned to live with a bottleneck.
That is useful too.
The leaders who make progress from here will be the ones who connect AI to decisions the organisation actually cares about.
The practical test is simple:
what decision changed, who owns it, and can we stand behind the result?
Executive note for leaders
Before scaling the next AI pilot, dashboard or licence rollout, ask:
what decision is this meant to improve?
If the answer is unclear, the organisation may be measuring activity rather than value.
Name the decision, assign the owner, then decide what evidence would prove AI changed the result.
Name the decision, not the tool.
Subscribe
You can subscribe for free. Every edition is delivered directly to your inbox and published here on Substack.
If a piece raises a question, surfaces a pattern, or helps you think more clearly about a decision, I’d value the conversation.
Thanks for reading,
Stuart Gonsal MAICD
With occasional help from Springsteen, my Border Collie, who reminds me that clarity comes from movement 🐾.
Connect
LinkedIn – Follow for practical leadership in the AI era. ↗
Disclaimer
Everything shared in The Ripple Effect reflects my personal views and does not reflect those of my current or past employers, clients or partners. Any examples are illustrative, drawn from publicly known patterns or anonymised experience.