ROAI. There’s an acronym to conjure with 😉
The Struggle for Return on AI Investment.
In the week where Steve Fox and I ran a series of discussions in our workshop “GBS – Accelerating Digital and AI Enabled Transformation” at the SSOW Fall Conference in San Diego, the topic of enterprise value creation in the age of AI is resonating everywhere.
Finance, GBS and Shared Services leaders discussed and debated what “enterprise value creation” actually means and how to align to achieve it.
Ultimately, its all about the P&L and Working Capital.
If we apply our “Think like a CFO” mindset, there are only 3 business outcomes that really matter for any initiative. They are the acid test for any transformation program;
- Strengthen financial resilience
- Improve cash flow, optimize capital, and manage risk
- Drive profitable growth
- Not just top-line revenue, but scalable, sustainable P&L and cash flow improvements
- Increase operational efficiency
- Cut waste, eliminate, streamline, automate to build a resilient, agile business
AI success will be measured in the same way.
AI is astonishing: fast, powerful, and learning every day.
But even as leaders roll out AI across their organizations, most still struggle to translate the technology into enterprise value, and now, Agentic is raising the stakes.
Isabelle Bousquette at WSJ wrote an interesting article on this, “Companies Struggle to Explain Their Own AI Investment Returns”.
Whilst many remain bullish on AI’s promise to transform business and create enterprise value, the value equation is a challenge, with some saying their initial planned budgets have as much as tripled this year.
Many executives observe that, in the majority of cases, their AI initiatives are costing more money than the value created.
That’s a hard lesson!
An AI built automatic out-of-office responder turned out to cost $10,000 a day. An outlier maybe, but a salutary example.
McKinsey’s latest research shows that 80% of workers feel they are getting a productivity boost from AI. However, only 6% of companies are getting real, measurable, financial ROI they can show to investors.
It is a confusing world, but this challenge is not new.
Individual tasks can be massively more efficient, but this may not flow through to financial benefit.
Why?
Because Tasks are NOT Jobs.
If your work is anything like mine, AI is providing massive productivity gains for certain tasks, even enabling tasks that were not realistic prior to AI.
But these tasks are also contributing to reshaping the nature of work. For every task that is mind-blowing with AI, there are also a series of new or changed tasks to consider and additional critical thinking to apply.
I am working even longer hours than I did before AI.
There is nothing new in this challenge. If everyone can save 20 minutes a day, it doesn’t necessarily reduce our costs by 4% or mean that we can reduce staffing by 4% (assuming an 8 hour day). I got hoisted on this petard early in my career.
There are also many reports of wastage coming from using Generative AI or agents for things that they’re simply not best used for.
In some cases, it has become became apparent that 95% of the problems AI agents were solving were better suited to deterministic workflows, a shift that saved one company between $20m and $30 million.
We have to choose the right tool for the task and consider the classes of automation we have at our disposal.
Our end-to-end value streams and processes are made up of hundreds, maybe thousands of tasks. Obviously, we should seek to eliminate any task that does not add value (and there are a lot of them when we look at our processes genuinely end-to-end!). But once we have defined the desired, streamlined end-to-end process focused on the enterprise value created, we need to ask what is the most appropriate mechanism with which to digitize?
For each task or group of tasks, is it best served by;
- Deterministic (Conditional) rules ?
- Probabilistic (Analytical) AI ?
- Generative (Conversational) AI ?
- Agentic Enterprise Workflow ?
- Or something else . . . ?
This is not an academic exercise. Selecting the right tool for the job is a key to enable genuine enterprise value creation.
The mindset shift required is to realise and emphasize that the goal of AI isn’t to replace workers, but to amplify them. Augmentation vs replacement.
There may be more replacement driven by “old school” deterministic, rules-based software.
The future likelihood is that the successful agentic workflows will be in the order of 30% probabilistic (AI) and 70% deterministic (classic rules based).
Tom Loftus observed, also in WSJ, that it would seem like the task of transforming a business to take advantage of AI—and make those investments pay off—is far, far more complex than cracking the latest big mathematical problem, Navier-Stokes.
It may just be that AI is allowing us to think differently and facilitate a structure of work that has a genuine “end-to-end” or value stream characteristic, that can integrate across the more rigid walls of the traditional, functional “silo” structure.
If AI can help “Silo-busting”, that may be the real enterprise value creation opportunity!
ROAI – The Struggle for Return on AI Investment.
You can read “AI Transformation’s Slow March” by Tom Loftus in WSJ here . . .
Thanks for reading . . .