Value Realisation · 21 September 2026
The value you don’t find is still value you lose
By The Cardaxia Team
The value you don’t find is still value you lose
Most businesses don’t have a shortage of ideas. They have a limit on how many ideas they can properly investigate. For decades, management teams have had to narrow the field early because there is only so much data a team can analyse, so many alternatives it can investigate, and so many assumptions it can challenge before a decision needs to be made.
A recent Harvard Business Review article, AI Is Revolutionizing Strategic Decision-Making, argues that AI changes this constraint. It allows businesses to search a much larger set of possibilities, build richer models from more information, and test potential decisions from multiple perspectives before management commits. For us, the interesting question is what happens when you apply the same thinking to value creation.
How much value never makes the shortlist?
Consider a typical cost, productivity or growth programme. Management already has hypotheses about where the opportunities are. Advisers add benchmarks and experience. Analysis identifies a set of initiatives and a business case follows. It is a sensible process, but it is necessarily selective.
HBR gives an example from M&A that illustrates how much that constraint can change. An AI-enabled scouting system combined semantic search with a database of more than 40 million public and private companies. It identified and scored more than 500 acquisition targets in less than a day before narrowing them to 15 serious prospects.
The important number isn’t 500. It is the 40 million that could be searched. Apply that principle to a business and the opportunity becomes interesting: combinations of expenditure, processes, customers, suppliers, technology, organisational choices and operating practices that may never have been seriously examined because doing so wasn’t practical.
AI reduces the cost of asking more questions. That should change how we look for value.
Discover more, then get sceptical
A longer list of AI-generated opportunities isn’t particularly useful. The quality of the shortlist matters, and this is where the second part of the HBR argument becomes important.
The article describes AI being used for structured challenge, with different agents generating an argument, attacking it and representing alternative perspectives. It also cites a field experiment involving 776 P&G commercial and R&D professionals. Individuals using AI matched the average quality of two-person teams without AI, while teams using AI were around 12% faster and more likely to produce top-decile solutions.
There is an important lesson here for value programmes. The same technology used to find an opportunity can also be used to challenge it: test the evidence, attack the assumptions, examine dependencies, model downside cases and bring different functional perspectives to the analysis.
Finding $10 million of potential value is relatively easy. Deciding how much of it deserves to become a management commitment is harder.
What this means for the Value Office
At Cardaxia, we think the Value Office needs to manage four distinct stages of value. They are connected, but each answers a different question.
Discover Value
Discover Value starts with a wider search. External evidence, internal data, specialist experience and AI can be combined to investigate substantially more of the opportunity space than a conventional review can practically cover. The objective isn’t to produce more initiatives; it is to improve the probability that the opportunities that matter make it onto the table.
Validate & Commit
An attractive hypothesis needs evidence before it becomes a commitment. The financial case needs to withstand challenge, assumptions and dependencies need to be visible, and uncertainty needs to be reflected in the value case rather than buried inside a single number.
Management judgement remains critical. AI can do more of the searching, testing and challenging, but executives still decide what the organisation is prepared to commit to.
Assure Value
The business case was only ever a prediction. Once execution starts, actual evidence becomes available: costs change, timing moves, assumptions prove right or wrong, and dependencies emerge. Each of these should change our view of the expected outcome.
A value case should therefore behave less like a spreadsheet approved at the beginning of a programme and more like a living model. It should update as the evidence changes and show where expected value is strengthening, weakening or moving somewhere unexpected.
Realise Value
Realise Value is the test that ultimately matters. A programme can finish on time, technology can be implemented and a process can be redesigned without the intended value ever appearing in the business.
This distinction is particularly relevant to SG&A. McKinsey studied 882 companies between 2013 and 2024 and found that 234 achieved a greater than 10% reduction in their SG&A cost ratio in one year and then sustained that performance for at least the following four years. Those companies improved their SG&A ratio at around seven times the rate of the overall group.
Finding value and sustaining value are different disciplines. A Value Office needs to do both.
The AI isn’t the advantage
HBR makes another point that is easy to overlook. The underlying AI models are becoming widely available, so access to them alone is unlikely to create much competitive advantage. The difference comes from what an organisation builds around them: its proprietary data, the way AI is incorporated into its processes, and its ability to act on what it learns.
The same applies to value creation. The opportunity isn’t to add AI to an existing transformation process. It is to reconsider the process now that some of its historical constraints have changed: search more broadly, challenge opportunities harder, commit with better evidence, keep testing the value case as reality unfolds, and measure success by what ultimately shows up in the business.
AI expands what we can investigate, but it doesn’t change the test. The value still has to be realised.
References
Felipe A. Csaszar, “AI Is Revolutionizing Strategic Decision-Making”, Harvard Business Review, September—October 2026.
Josh Peters, Jung Paik and Martin Rosendahl with Abhishek Shirali, “The SG&A challenge: Achieve excellence and outperform your peers”, McKinsey & Company, October 2025.