.png)
The AI Illusion: What Really Counts as Business Value?
Why faster processes and hours saved don't necessarily deliver a financial return, and what leaders should consider before investing in AI.
AI is making it easier for businesses to automate work, produce analysis and accelerate processes. But when does greater efficiency translate into measurable business value?
That question becomes particularly important in a business turnaround, when cash, time and management capacity are already under pressure.
On 18 September 2026, I chaired The AI Illusion at the Institute for Turnaround (IfT) National Conference in London, joined by Sarah Hollyhead of Marlin Equity Partners and Ilona E Simpson of Agent & Capital.
We discussed what artificial intelligence can realistically contribute to business recovery, and how leaders should distinguish genuinely valuable improvements from activity that simply looks productive.
It's a subject closely connected to my work as a Partner at Owendale Advisory, where we help organisations understand how AI is changing their markets, identify where it can create value and develop the leadership and organisational capability needed to act.
The conversation kept returning to one distinction.
AI activity is not the same as business value.
A task can get faster without costs disappearing.
The question I came away with was: which business problem are we trying to solve, and how will we know whether AI has helped?
What will customers still value?
Automating routine reporting or removing a bottleneck can be worthwhile. But neither necessarily answers a more fundamental question: will customers continue to value what the business sells?
A more efficient back office will not solve the problem if the business's products or services are becoming easier to copy.
For investors, that question extends beyond current profitability to what will remain distinctive and valuable to a future buyer.
Decide which outcome you are pursuing, then agree how to measure it.
Choosing an AI project the business can deliver
For a business with limited cash or management capacity, an AI initiative needs to be material enough to deserve attention and manageable enough to deliver.
That means considering business readiness, technical readiness, data and affordability together.
Start with the licences and information the business already has, and what you can learn from comparable businesses. Not every opportunity requires a large data programme.
But a narrow dataset is different from an unreliable one. AI may help reconcile inconsistent records. That does not mean it can supply information the business does not have.
Use the pilot to check whether people can use the result, whether the process changes are workable and whether the economics justify taking it further.
Establish the baseline before the pilot begins.
Hours saved are not automatically money saved
Time saved might let the business produce more, improve service, avoid additional hiring or reduce costs. It might also be absorbed into other work without a measurable financial benefit.
Multiplying hours saved by an hourly rate does not, on its own, establish a return.
What happens to those hours next? Does expenditure change? Is another initiative claiming the same saving?
The same applies to the extra reports, dashboards and analysis AI can produce. Are they useful enough to justify the effort and cost?
Costs also extend beyond licences: integration, training, human oversight, rework and running old and new processes in parallel all matter.
Check the day-to-day cost, rather than relying on the demo cost.
Payroll savings are not the only benefit worth measuring. Extending asset life, deferring expenditure, improving service or strengthening resilience can all matter.
But anticipated benefits, expenditure deferrals and realised savings should be described accurately and not treated as interchangeable.
Who owns the business outcome?
A technically sound solution can stall because teams have different priorities, incentives or views about who should change.
Delivering a tool, changing a process and realising a benefit are connected responsibilities. They are not necessarily the same job.
The outcome owner needs authority to change working practices, resolve cross-departmental issues and stop activities that aren't delivering.
Finance has a separate role: challenge assumptions, identify double-counting and confirm whether claimed benefits translate into profit and loss or cash.
The person delivering the benefit should not be the only one assessing it. But finance should not end up running the whole transformation.
Changing how people work
A model can work technically and still be of little use if it lacks context, staff do not trust it or the process around it has not changed.
Someone needs to understand how the work fits together, decide who does what and see it through.
The team running the business may not have the time or skills to take that on.
Access alone is not enough, either. Training, internal champions and incentives need attention. This is especially true when people feel threatened by the change or see little reason to share efficiencies they have developed.
Using AI yourself can help you understand its strengths and limitations.
It does not tell you where the business should invest.
AI governance should help decisions happen
Too little governance leaves activity disconnected from business priorities.
Too much can turn a manageable improvement into an unnecessarily complicated programme.
An internal information feed should not face the same scrutiny as a system whose failure could have serious operational consequences.
Who can approve the next step, challenge a supplier's assumptions or decide that the evidence does not justify further spending?
Those responsibilities are better settled before a problem emerges than by a committee assembled afterwards.
Decide what must remain human and what happens when AI fails
We also discussed what businesses should not touch.
Where customers value relationships, authenticity or expert judgement, removing the people involved may damage what the business offers.
Once an AI service sits inside a critical process, what happens if it becomes unavailable or its output cannot be trusted?
The business needs to retain enough experience to challenge results, recognise mistakes and continue operating.
A fallback process is of limited use if the knowledge required to run it has disappeared.
What changes when AI is introduced during a business turnaround?
Start with the cash runway, the immediate operational constraints, the tools already available and the commitments you can change.
AI might help a practitioner understand the business faster or address a specific bottleneck.
A more extensive redesign might offer greater theoretical value but demand cash, skills and management attention that aren't available.
The first 30 days offer an opportunity to assess whether AI could materially improve the outcome.
Knowing what to leave alone is part of that judgement.
AI should support the recovery priorities, not become a competing agenda.
Six questions to take into your next leadership meeting
What are we already doing with AI? Where is it being used, what are we paying for, and what evidence shows it is helping?
What are competitors and comparable businesses doing? What has worked or failed elsewhere? Is AI changing the basis of competition, or mainly offering a different way to operate?
Which business problem deserves attention now? Is it material, manageable and deliverable within the cash and time available? What should we deliberately leave alone?
How will we measure the benefit? What is the baseline, what will it cost in full and what will happen to the time saved? Who will check that the benefit has been delivered?
Who can make the necessary changes happen? Is there an accountable owner with the authority, skills and capacity to act across departments? Are the people affected equipped and motivated to change?
What would make us continue, change course or stop? What evidence will we review, who will decide and how will the business operate if the AI service or its outputs cannot be relied upon?
My main takeaway from the conversation with Sarah and Ilona was how much of this is familiar turnaround work: understanding the business, setting priorities, challenging the numbers and making change happen.
AI changes what may be possible. It does not remove the need for those disciplines.
About the author
Rebecca Hastings is the founder of Lucent Search, a retained executive search firm specialising in senior AI, data and technology leadership. She is also a Partner at Owendale Advisory, working with organisations on AI strategy, business transformation and the organisational implications of technological change.
Her work connects the practical challenges of AI implementation with the leadership capability, authority and decisions needed to turn strategy into business results.
This article draws on a panel chaired by Rebecca at the Institute for Turnaround National Conference on 18 September 2026. Originally published in What Strategy Becomes on LinkedIn on 19 October 2026.
Comments