AI TransformationMeasured before it is committed
We start from your profit and loss account, identify the use cases likely to pay their way and treat each pilot as an investment, with a cost, a stopping point and an expected return agreed before work begins. S&P Global reports that companies abandon, on average, 46% of their AI proofs of concept before they reach production.
For chief executives who have launched AI pilots or are considering them.
The use cases likely to pay their way, quantified on your data and ranked by return on investment.
Each pilot has a cost, a stopping point and an expected return, agreed before work begins.
If you ask for it.
Why so many pilots come to nothing
| Study | Finding | Scope |
|---|---|---|
| MIT NANDA, The GenAI Divide, July 2025 | 95% of organisations see no economic return from generative AI | 300 public initiatives, 52 interviews, 153 survey responses; limited sample Source |
| S&P Global Market Intelligence, March 2025 | 42% of companies had abandoned most of their AI initiatives, against 17% a year earlier; 46% of proofs of concept abandoned before production, on average | More than 1,000 respondents, North America and Europe Source |
| IDC, March 2025 | Of 33 proofs of concept launched, 4 reach production | Data, processes and infrastructure not ready Source |
| Gartner, July 2024 | Gartner forecast that at least 30% of generative AI projects would be abandoned after proof of concept by the end of 2025. | Data, risk, costs, unclear business value Source |
The causes recur: an unclear indicator, data that is not ready, a process that has not been reviewed, and no one to decide to stop.
Our approach
- Starting from the profit and loss account. We identify where the value lies, then where AI is the best lever. Sometimes it is not, and we say so.
- Quantifying each use case on your data. Gain, full cost, timescale, risks. We rank them by return on investment.
- Designing the pilot for production. Before starting, we agree the indicator the business already tracks, the target value, the stopping point and the person entitled to change the process.
- Measuring two objectives separately. The economic result on one side, the learning of your teams on the other. A pilot may train your teams; it should not serve as an excuse for an absent result.
- Deciding. Scale up, correct or stop. Stopping at the agreed point is a sound decision, not a failure. If you wish, we then oversee the roll-out.
The expected benefits
Gains measured on your own indicators, not on demonstrations.
Less money tied up in pilots that lead nowhere.
Teams trained while working on their own cases.
A roadmap your finance team can budget for.
The risks we address from the outset
Incomplete or scattered data.
An unchanged process around a new tool.
Usage costs that drift.
Confidentiality and unregulated use.
Compliance with the EU AI Act.
Dependence on a single supplier.
What a pilot without results costs
An illustrative example for one business, with its assumptions. The amounts are orders of magnitude.
| Item | Assumption | Amount |
|---|---|---|
| Internal team | 4 people part-time, i.e. 1 full-time equivalent for 4 months, at €6,000 a month including employer costs | €24,000 |
| External supplier | 20 days of consultancy, €18,000 in total | €18,000 |
| Licences, APIs, hosting | 4 months at €1,500 | €6,000 |
| Management time | Steering committees | €3,000 |
| Direct cost | €51,000 | |
| Training value to be deducted | 15 people, 2 days of training avoided, €21,000 in total | −€21,000 |
| Net direct cost | €30,000 | |
| Opportunity cost | The use case that would have returned €150,000 a year, delayed by 6 months | €75,000 |
| Total economic cost | €105,000 |
The learning is real, but it recovers only about 40% of the direct cost. The largest item appears in no budget: the deferred gain from the right use case. A business that launches five pilots and abandons 46% of them, the average reported by S&P Global, ties up some €117,000 in direct costs without result.
Frequently asked questions
Why do so many AI pilots come to nothing?
The studies cite the same causes: data not ready, unclear business value, costs poorly anticipated, processes not reviewed. A pilot launched without an indicator or a stopping point has no reason to stop, nor to succeed.
Is a pilot that trains the teams not useful in itself?
It is, but rarely in proportion to its cost. In our example, the training recovers about 40% of the direct cost. We therefore set an explicit learning objective, measured separately from the economic result.
Should we start small?
Yes, but on the right subject: a single process, an indicator already tracked, a quantified target, and a budget for data and training alongside the licence budget.
