The technology sponsor can help you put AI to work, but someone in the business has to own what changes as a result.
So, the AI demonstration went well, the team liked it, the licences are bought and then someone at the leadership table asks the awkward question, ‘Tell me what, exactly, will be different in the business?’
If the answer is ‘people will save time’, you haven’t finished the investment conversation. You have only reached the interesting part.
There is a reason to press further. In McKinsey’s August 2026 survey, 80% of respondents said AI had improved their own productivity, while 37% reported a positive AI contribution to organisational earnings before interest and tax (EBIT), essentially unchanged from 2025. These are different, self-reported measures, not audited results or numbers we can subtract to calculate a benefits gap.
Higher-performing organisations also more commonly reported fundamental workflow redesign, senior leadership commitment and defined impact measurement processes. The survey shows associations, not proof of what caused the results.
My take is straightforward, if we want a business result, we need someone in the business who owns it, not just someone who sponsors the technology.
Give someone the authority to deliver
A benefits owner is accountable for the outcome and has the authority to change the process, people’s responsibilities and the way work is managed. They can decide what stops, what gets redesigned and where released capacity goes. A name on a project slide is not enough if that person cannot make those decisions.
This is not an argument for sidelining IT.
The technology sponsor remains essential for dependable systems, integration and support. Finance helps test the economics and validate the benefits, while risk colleagues help agree controls and acceptable exposure.
The benefits owner brings those contributions together around a business outcome, rather than expecting IT to deliver a result it doesn’t control.
That distinction matters because installing a tool and changing how a team works are quite different jobs. If every approval, handover and performance measure stays the same, faster drafting may simply feed the same old queue.
Follow the work, not the demonstration
Imagine a customer service team using AI to prepare replies. This is an imagined example, not a case study. Drafts now take less time, but every reply still waits for the same manager’s approval. Customers haven’t noticed much difference.
The service leader, acting as benefits owner, looks at the whole journey. Could routine replies go through a simpler, agreed check? Which cases still need expert judgement? What training will staff need, and who will deal with a wrong answer? Technology helps, but these are operating decisions.
Now there’s something worth testing. Can we reduce the time customers wait, without increasing errors or repeat contacts. We would need to establish the current position first, agree a target and a test period, then measure the full journey and include checking and rework, not just the minutes spent producing the first draft.
You also need to talk to the people doing the work. They will often know exactly where a promising shortcut creates another job downstream. A redesign done with them is more likely to work than a new tool dropped into an unchanged process.
Be precise about what ‘saving’ really means
Suppose the team releases some hours each week. That’s what we call capacity. It isn’t automatically cash back in the bank, and multiplying those hours by salary rates doesn’t make it so.
Cost reduction means an actual fall in spending, perhaps through less paid overtime. Avoided future spending means handling more work without a planned hire, it needs a credible account of what would otherwise have been spent. Neither should be counted again as another benefit.
Growth or capacity means using the released time to serve more customers or do more valuable work. The owner must explain where that time goes and how the extra output will be recognised.
Service quality means better experience for the customer, such as quicker resolution or fewer errors, whether or not it reduces costs.
All four can matter, they just are not interchangeable. A stronger service may be an excellent investment without being described as a cost-cutting exercise. Honest categories make for better decisions than an impressive total built from overlapping claims.
Put the business case on a review date
Before committing, agree with the baseline, the target, the timeframe and who will validate the result.
Record the assumptions too, expected demand, how much staff use the tool, and the level of checking required, otherwise, a change in workload can look like an AI benefit when it is not.
Count the full cost of getting there and keeping it working.
Licences are only the start, include integration, data preparation, training, staff time, process redesign, controls, monitoring, support and ongoing review. Finance should help test whether the expected outcome is worth that whole commitment, not just the subscription.
Then put a decision in the diary.
At the review, compare results with the baseline, check quality and risk, and examine whether the benefit holds after the novelty wears off. Decide whether to adjust, stop or scale. Scaling should follow demonstrated value and workable controls, not enthusiasm alone.
Stopping is not an embarrassment if the evidence says the investment does not stack up. Continuing without asking is an expensive choice.
Ask who owns Monday morning
At your next AI investment discussion, try five questions:
- Who owns the business outcome, and can they change the work needed to deliver it?
- What will we stop or redesign, rather than simply do faster?
- Are we claiming lower costs, avoided spending, more capacity or better service, and how will we tell?
- What’s the baseline, target and full cost, and who will validate the result by when?
- What evidence would make us stop, adjust or scale?
These are not questions to dampen ambition, they are how you turn ambition into something the organisation can deliver.
Your technology sponsor can help make AI available. Your benefits owner must make the changed way of working real on Monday morning. Before you approve the next investment, make sure someone owns that Monday, not just the launch.
Sources
- McKinsey, ‘The state of AI in 2026: On the road to ROI’, 25 August 2026. Dan Tinkoff, Lieven Van der Veken, Michael Chui and Tara Balakrishnan. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai?rd=1
Summary
AI investment needs more than a technology sponsor, it needs a business leader accountable for turning better tools into measurable results.
Real value comes from redesigning workflows, deciding how saved time will be used, and tracking business outcomes rather than counting licences or pilots.
About Bushey Technology Transformation
Bushey is a specialist boutique consultancy, with reach across Australia and APAC. Our Technology Transformation Practice provides independent, client-side delivery and advisory for Corporate and Government organisations. We take complex programmes through to benefits, not just to plan.
Our team has led senior technology functions and delivered high-profile programmes across finance, insurance, manufacturing, retail and the public sector. We have also recovered troubled programmes, including an APAC-wide Programme completed over 18 months alongside a global delivery partner.
At the core of every engagement is AssureChange®, our proprietary framework bringing Project Management, Change Management and Governance together around one goal, the business benefits you signed up for. We sit on your side of the table, so you always hear the real position while there is still time to act.
#AILeadership #BusinessTransformation #AIValue #BenefitsRealisation #ResponsibleAI

Comments are closed