The wrong first question about AI: pick the P&L line first
“Where can we use AI?” produces pilots. “Which P&L line moves, by how much, and by when?” produces a business case a CFO will fund. The first question an executive team asks decides which of the two it gets.
Two questions, two outcomes
Most AI programmes start with a search for use cases. Teams list the places a model could help, rank them by enthusiasm, and fund the top few as pilots. The question sounds open-minded. In practice it points the organisation at the technology and away from the result.
The alternative starts at the other end. Name the line on the income statement that should move: cost to serve, credit losses, claims leakage, collections, revenue per client. Say by how much, and by when. Then work back to the process, the decision and the tool.
A use case describes what the technology does. A P&L line describes what the business gets.
Why the use-case question stalls
Boston Consulting Group’s 2025 research, The Widening AI Value Gap, found that 60% of companies report little or no value from AI, and only 5% achieve value at scale. The models are available to everyone, so access does not explain the gap. The choices made around the model do.
A use-case list produces pilots that are judged on whether the model works. A pilot can succeed on that test and still leave the P&L where it was, because nobody decided which budget line would change, who would sign for it, or what work would stop.
What a P&L-first case contains
- The line. One named line on the income statement, not “productivity”.
- The baseline. Today’s figure, from your own data, so the result is a comparison and not an opinion.
- The size and the date. How much the line should move, and by when, stated before anything is built.
- The owner. One executive who signs for the number and can change the process.
- The work that changes. Which steps stop, speed up or move between people and systems.
- The controls. Where a human decides, and how you will evidence it, including for decisions covered by POPIA section 71.
Ask every AI proposal on the table to restate itself in one sentence: “This will move [line] by [amount] by [date], owned by [name].” The proposals that cannot finish the sentence are not ready for funding yet. That is useful to know before the money is spent.
The point
Start from value, not tools. Picking the P&L line first does not slow an AI programme down. It removes the pilots that were never going to pay, and gives the rest a number someone has agreed to own.
Open the free AI-to-P&L Calculator
Sources
- Boston Consulting Group, The Widening AI Value Gap, 2025: 60% of companies report little or no value from AI; 5% achieve value at scale.
- Protection of Personal Information Act 4 of 2013 (South Africa), section 71: Automated decision making. Summary above is a paraphrase; refer to the Act for the authoritative text.