When IKEA deployed its AI chatbot Billie to handle routine customer service inquiries, the bot resolved approximately 47% of all customer contacts without human intervention. Most organizations would have stopped there, logged the cost savings, and called it a win. IKEA asked a different question: what do we do with the 8,500 people who no longer need to answer those calls?
The answer was a close analysis of what Billie could not do — which pointed directly to unmet customer demand for interior design expertise. IKEA reskilled those employees as design consultants, launched a paid consultancy offering, and generated approximately €1 billion in new revenue in the first year. A technology deployment became a redeployment story, and that distinction is precisely what most organizations miss when they think about AI today.
Most organizations only capture the first half of that equation — the efficiency gain — which is why they measure cost savings but struggle to articulate broader impact. IKEA captured both halves. While the chatbot handled the volume, the leadership decided where the freed capacity went. The result was a new revenue stream built entirely from existing talent.
IKEA started with a use case that was operational, well-defined, and measurable. That discipline is replicable. The Impact Index is the framework that makes it repeatable.
The Two-Layer Filter
Before you select a use case, evaluate vendors, and scope a proof of concept, answer these two questions:
- Is this use case actually ready to deploy?
- When it works, how does the outcome get redeployed to produce value?
Gartner and MIT CISR have both mapped AI maturity at the organizational level. The Impact Index applies the same logic one level down to the use case itself. That is where organizations should be focusing, and what a CFO should be approving: the highest-leverage starting point.
The framework is also tool-agnostic. It does not evaluate whether a use case deserves AI. It evaluates where effort and investment will produce the highest return. Sometimes that means agentic AI. Sometimes simpler automation. Sometimes process redesign before any technology is introduced. The index has no agenda — only criteria.
Layer One: The Impact Index
Most AI use cases fail because the foundational work was not done before the technology was introduced. The Impact Index is designed to surface that gap early.
The Index runs five criteria. You score each one zero, one, or two, which gives you a maximum of ten points. It takes about fifteen minutes to work through. Find the full scoring guide is in the workbook at the end of this article.
Think of it as a fast filter. The goal is to catch the use cases most likely to fail before they consume budget and goodwill. If a use case scores low, that is useful information — it tells you what work needs to happen first. Low scores get sequenced, not shelved.
1. Process definition
CFO question: "Can you show me this process on one page today?"
If the answer is no, document first. Technology requires a defined process to improve.
2. Automation foundation and friction
CFO question: "Is any part of this already running in a system, and what is the fully loaded cost of what remains manual?"
The single strongest predictor of time to value. A process already partially automated has had the foundational work done — you are starting from stage two. It also surfaces whether a simpler, deterministic automation would be faster and cheaper than AI.
3. Data readiness
CFO question: "Where does the data live, who owns it, and when was it last audited?"
Poor data produces confident AI pointed in the wrong direction — quickly.
4. Team readiness
CFO question: "Would the people doing this work today welcome this change, and are they involved in scoping it?"
A time-starved team comfortable with tools is a built-in adoption engine. Skepticism and competing priorities slow everything down, and technology alone will not resolve them.
5. Financial line of sight
CFO question: "Which line on the P&L does this connect to, and how will we measure movement in 90 days?"
If the answer takes more than two sentences, the business case needs more work.
Reading the Score
8 to 10 — High confidence. The foundational work is largely in place. Define the outcome, engage the team, deploy.
5 to 7 — Proceed with caution. One or two gaps exist. Identify which criteria scored low and treat them as pre-work. This is still a strong use case but it needs a few steps before it is ready.
4 and below — Not ready. A useful answer, and a temporary one. The value of this use case may be real, but the foundational work has not been done yet. A use case that scores low today can score high in six months.
Layer Two: The Throughput Question
The Impact Index tells you whether a use case is ready. This second layer determines whether the business case is actually complete.
"400 hours saved" is a technology metric. "400 hours saved and redeployed to pipeline generation, driving an incremental $2M in revenue" is a business case.
Time saved does not automatically become margin captured. Without a named destination for the capacity an initiative frees, organizations end up with a tool that works and a P&L that does not move. IKEA named the destination before the chatbot went live. That decision is what produced €1 billion in new revenue.
Determine the destination first. Then build the path.
So before approving any initiative, ask one more question: when this works, where does the freed capacity go, and who is accountable for deploying it toward a specific outcome?
There are four possible destinations:
- Redeployment to higher-value work.
- Absorbing growth without adding headcount.
- Reducing cost over time through natural attrition.
- Accelerating revenue through faster cycle times.
Any one of them is a complete answer. The only wrong answer is leaving it unaddressed.
This is not a new skill — it just needs to enter the approval conversation before the budget is committed.
Close the Gap Between AI Investment and Business Outcome
Every AI initiative that stalls, underdelivers, or quietly gets shelved has something in common. The technology was evaluated. The business case was not.
The Impact Index and the redeployment question are not sophisticated tools. They are two disciplined questions asked before the budget moves. Most organizations skip them — not because they lack the capability, but because the approval process was built around technology adoption, not business outcomes.
That is the shift worth making. AI is not a productivity tool waiting to be deployed. It is a capacity-generation engine waiting to be directed. The organizations pulling ahead are not the ones with the most advanced models. They are the ones who decided where the value was going before they signed the contract.
The framework is in the workbook. The harder work is changing the conversation that happens before anyone opens it.
Sources: BLS Employer Costs for Employee Compensation (December 2025); Gartner AI Maturity Model (2025); MIT CISR Enterprise AI Maturity Model, Weill, Woerner and Sebastian (December 2024); IKEA Billie AI deployment and reskilling program, as reported by Brian Solis and Fluent Support (2024, 2026)
