AI INVESTMENT
Why Most AI Initiatives Never Deliver ROI and How to Choose the Ones That Will
AI initiatives rarely fail because of the technology alone. The real challenge is choosing projects with a clear path to measurable business value.
Challengine · 23 September 2026 · 6 min read
AI has made it easier than ever to generate ideas.
Automate customer service. Build an internal copilot. Predict demand. Automate reporting. Add AI to sales. Improve forecasting.
The problem is no longer finding possible AI use cases.
The problem is deciding which ones are actually worth doing.
Many organizations move from idea to pilot too quickly. A promising use case gets attention, a team is formed and implementation begins before anyone has properly defined what business problem is being solved or what success should look like.
That is where ROI starts to disappear.
Evaluate the business case before the technology
An AI initiative should compete for resources just like any other investment.
Before asking “Can we build this?”, ask:
“Should we build this?”
A simple evaluation can reveal a lot:
| Question | What you need to know |
|---|---|
| Problem | What measurable problem are we solving? |
| Value | What changes financially or operationally if this works? |
| Baseline | What does the process cost or produce today? |
| Feasibility | Do we have the data, systems and capabilities required? |
| Ownership | Who is responsible for delivering the outcome? |
| Risk | What could prevent the expected value from materialising? |
| Measurement | How will we know whether the initiative worked? |
If these questions cannot be answered, the initiative is probably not ready for execution.
A good AI idea is not necessarily a good investment
Consider two AI initiatives.
Initiative A
A sophisticated AI assistant that could improve several internal processes.
Initiative B
Automating one repetitive process that currently consumes 2,000 employee hours every year.
Initiative A might sound more ambitious.
Initiative B might have the stronger business case.
This distinction matters. Organizations should not prioritize initiatives based on how impressive the technology sounds. They should compare them based on expected impact, cost, risk and execution readiness.
The strongest AI portfolio is not necessarily the one with the most AI projects.
It is the one where the best ideas receive resources, and weak ideas are stopped early.
Move from AI experimentation to AI investment decisions
A useful process is simple:
- 01Identify
- 02Structure
- 03Evaluate
- 04Compare
- 05Decide
Every initiative should eventually answer three questions:
- What value could this create?
- How confident are we that the value can be delivered?
- Is this the best use of our resources right now?
This creates something many AI portfolios are currently missing: comparability.
Instead of reviewing ten completely different presentations, spreadsheets and pilot proposals, decision-makers can compare initiatives using the same logic.
That makes it easier to identify the initiatives worth accelerating, and the ones that should never become projects.
The goal is not more AI projects
AI makes execution faster.
That makes good decisions even more important.
When execution becomes cheaper and easier, organizations can launch more initiatives than ever before. But scaling bad decisions simply creates waste faster.
The competitive advantage will increasingly come from knowing what is worth executing before execution begins.
That is the shift from AI experimentation to AI investment management.