Only 4% of organisations without direct CEO ownership of AI strategy report established ROI from it. Among those where the CEO actively leads it, that figure rises to 14%, three times higher[1]. The same KPMG research found that CEO accountability correlates with a 2.7x increase in meaningful business value from AI investments. Those are not incremental differences. They suggest that the question of who is accountable for AI inside your firm is at least as important as which tools you deploy.
For a wealth manager or IFA principal, this deserves more than a quick read. Most advice firms are currently somewhere between “we have a few people using ChatGPT” and “we have a head of operations running an automation project.” Very few have made AI strategy a personal leadership priority. That gap is where most of the return is being left behind.
Why leadership ownership changes what gets built
When AI is delegated entirely to a technology lead or an enthusiastic paraplanner, the projects that get funded tend to be the ones with the clearest technical path, not the highest strategic value. A compliance checklist automation, a document templating tool, a client portal integration, all useful, all tractable, all below the level of decisions that change how a firm operates.
The KPMG and University of Texas at Austin research, which analysed 1.4 million real workplace AI interactions, found that the highest-impact users were not the ones who optimised prompts[1]. They were the ones who framed problems well and guided thinking iteratively. The same dynamic applies at firm level. When a senior decision-maker owns the strategy, the questions change: not “can we automate this task?” but “which of our firm’s constraints would we most want removed, and is there an AI-shaped solution to any of them?”
That reframe is what separates firms that find real ROI from those that accumulate a collection of tools that never quite justify their cost.
What CEO ownership actually requires
It does not require the principal to become technically fluent. It requires three things.
First, set the constraints yourself. Decide which client-facing processes AI can touch and which it cannot, before the tools are selected. In a regulated advice firm, every AI-assisted output in a client-facing or compliance context requires human review before it is used[1]. That is not a technical decision. It is a governance decision, and it needs to come from the top. Under the Senior Managers and Certification Regime, accountability for systems and controls that affect client outcomes rests with named individuals, and a delegated AI project does not move that accountability.
Second, measure from the start. The era of AI experimentation budgets is closing. Finance functions are now requiring measurable outcomes before approving further AI deployment. If you are starting a new automation project this quarter and you have not agreed what success looks like, in hours recovered, in error rate, in cost per client serviced, you will struggle to build the internal case for the next phase. More than 40% of current AI projects are forecast to be cancelled by 2027[2], and the ones most at risk are those that cannot demonstrate a return in terms the business cares about.
Third, keep the accountability visible. This does not mean attending every workflow configuration session. It means making clear internally that AI strategy is a firm priority, that resource requests related to it will be taken seriously, and that there is a named person (you) who will be answering for the outcomes. That signal changes how the rest of the firm engages with the work.
The question of who is accountable for AI inside your firm is at least as important as which tools you deploy.
The governance gap in most advice firms
The research finding about ROI is striking in part because it suggests the governance gap is still very wide. Only 14% of CEO-led organisations report established ROI, which means even among firms with the strongest structure, most are still finding their way. Among those without it, the figure is 4%.
In financial advice specifically, there is an additional pressure. AI governance has moved from internal best practice toward a regulatory expectation. The FCA has been explicit that it expects firms to understand and be accountable for any AI systems they use in client-facing or decision-support contexts[3]. If that accountability is diffuse, spread across a compliance officer, an IT contractor, and a paraplanner who set up the workflow, it is not really accountability at all.
This is not a reason to slow down AI adoption. It is a reason to structure it so that what you build holds up to scrutiny, and so that the person at the top of the firm can answer clearly for what is running and why.
What this looks like in practice at a small advice firm
A principal running a ten-adviser firm does not need a formal AI steering committee. The practical minimum is closer to this:
Map what is actually running. Before you can own the strategy, you need to know what tools are in use across the firm, not just the ones you approved. Shadow AI use, advisers using personal ChatGPT accounts for draft suitability letter language, for instance, is common and creates liability exposure that you may not be aware of.
Separate the tiers. Not every AI use carries the same risk or the same upside. A tool that drafts internal meeting notes is categorically different from one that touches client-facing output. Decide where the firm’s tolerance sits on each, and make that decision explicit rather than leaving it to individual judgement.
Name the review step. For any AI-assisted output that goes to a client or into a compliance record, there should be a named human reviewing it before it is sent or filed. That step needs to be documented, not assumed. The tool drafts; the adviser signs off. That structure also happens to be what protects the firm if a client ever challenges the quality of advice.
Review the list quarterly. The AI tool landscape is shifting faster than annual review cycles can track. ChatGPT’s share of the AI assistant market fell from 65.3% to 46.4% in May 2026 alone[4], the first time it has dropped below 50%, which means firms that standardised on a single provider are already operating in a different competitive environment than the one they planned for. A quarterly check-in on what the firm is using, what it costs, and whether it still makes sense is a reasonable governance minimum.
The honest version of the 3x finding
Three times the ROI from CEO-led AI is a compelling headline. The honest version is more specific: it is three times more likely that a firm reports established ROI, which is not the same as three times the financial return. It means that firms with leadership accountability are more likely to have built the measurement infrastructure to know whether the investment is paying off, and to have shaped projects around outcomes that the business cares about.
That is still the important finding. Most firms that are not seeing a return from AI are not seeing it partly because no one senior enough has decided what a return would look like.
If you are the principal of an advice firm and you have left AI strategy to someone else, the research suggests that is the single change most likely to move the needle, not a new tool, not a larger budget, but your personal involvement in setting the direction and holding the firm to account for outcomes.
If that is the conversation you want to think through for your own firm, a discovery call with Cordrey Consulting is a good place to start.
This article is for informational purposes only and does not constitute regulated financial advice or a compliance opinion. Consult a qualified compliance professional for advice specific to your firm.
Sources
[1] KPMG and University of Texas at Austin (2026) CEO AI Leadership and ROI Outcomes, KPMG. Available at: https://kpmg.com/us/sophisticated. [Industry report. Vendor-sourced. Supports the 3x ROI and 2.7x business value figures, and the finding on collaborative, iterative AI usage.]
[2] Zen van Riel (2026) ‘Why 78% of AI Agent Pilots Never Reach Production’, Zen van Riel, AI Engineer Blog, 31 May 2026. Available at: https://zenvanriel.com/ai-engineer-blog/ai-agent-scaling-gap-pilot-production-2026. [Supports the 40%+ project cancellation forecast for 2027.]
[3] FCA, ‘Rethinking Regulation in the Age of AI’, Financial Conduct Authority, 26 June 2026. Available at: https://www.fca.org.uk/news/speeches/rethinking-regulation-age-ai. [Supports the statement on FCA expectations for AI accountability in regulated firms.]
[4] Digital Applied (2026) ‘ChatGPT Drops Below 50%: AI Assistant Market Share 2026’, Digital Applied, 19 June 2026. Available at: https://www.digitalapplied.com/blog/ai-assistant-market-share-2026-chatgpt-below-50-percent-analysis. [Supports the market share figures for ChatGPT.]