Pareto shows you where to look. Artificial intelligence now shows you faster, in nearly every discipline of the 80/20 business process. The discipline has not changed. The cost of running it has collapsed.
For decades, the customer-product matrix and true profitability analysis were projects: weeks of exports, pivot tables, arguments over cost drivers and reconciliation. That friction is the main reason most companies never ran the analysis at all.
Chapter 7 of The 80/20 Business Process runs the full diagnostic inside an ordinary AI-enabled spreadsheet, starting from a single tab of raw sales transactions. Twelve plain-language prompts, about two hours of work. The tool is interchangeable: Sheets with Gemini, Excel with Copilot, or a spreadsheet with Claude or ChatGPT connected to the data. What matters is the thinking behind the prompts and the decisions you make from the answers.
The AI even writes the formulas. Your job shifts from building them to checking them: test the results against a handful of customers and products you know well, confirm that the complexity reconciles to the cost pool, and only then trust the curve.
Illustrative dataset from the book: nine customers, about 230 products, $66.4M in sales.
A conventional P&L reports one blended number. Allocate complexity by activity and the business splits into parts that behave differently: a core to protect, hidden gems to scale, an operational-drag quadrant to simplify, and a tail to automate or exit. Those four verbs are not opinions. They fall straight out of the numbers. And because the rerun now takes an afternoon, the diagnosis can be repeated every quarter, watching the tail being tamed almost in real time.
AI runs a thousand SKUs through the decision tree in minutes, settling the obvious keeps and cuts so people can spend their time on the hard cases. The same prompts that built the diagnosis test the cut list against true profitability before anyone commits.
It flags every SKU with a negative true-profitability variance and ranks it by dollar gap. It reads scattered CRM notes, distributor sheets and competitive quotes to build a market price benchmark with a confidence level on each estimate, and reruns on schedule to report only what changed.
Behavioral scoring that took an analyst a week now takes minutes on a file you already have. AI scores customers and prospects against the ideal customer profile and estimates wallet size from public signals. Its output is a prioritized hypothesis, to be validated like any other estimate.
Rebuilding the matrix for every segmentation hypothesis once took weeks. Now a team can ask in plain language what happens to margin if the bottom 40% of SKUs leave a segment, and see the answer while the question is still live.
AI forecasts, routes and flags variability before it becomes a missed shipment, and cuts the time needed to grade inventory by value, efficiency and return on investment.
Automation serves the transactional tail instantly and at near-zero cost. The reward is the human capacity released at the top, pushed deliberately into the vital few. Two guardrails: no vital-few account without a named human owner, and every automated channel identifies itself and offers a fast route to a person.
An AI-enabled knowledge base becomes the memory of the innovation engine. It clusters hidden friction patterns from field observations and service logs, and retrieves the company's own prior solutions, and analogous ones from other industries, when a team faces a hard technical contradiction.
AI finds the same disproportion in a workforce that it finds in a customer file. It cannot know who is quietly mentoring the people around her. The leader supplies the context, and owes the same fairness to whoever lands on the wrong side of the list.
The judgment is not in the output. It never was.
From The 80/20 Business Process
The relationship runs both ways. ERP and CRM systems amass vast amounts of data, much of it noise. Organized around the vital few, that data gives AI a prioritized, well-structured framework, narrows its attention to the segments that matter, and lowers the risk of models that fit the past but fail on what comes next.
AI is strongest at scale and pattern recognition. People bring context, intuition and values. Together, on a shared 80/20 data model, they produce distinct but complementary insight. And a company that keeps accumulating data under a well-understood 80/20 model holds an asset whose value rises with every advance in the technology.
AI is a tireless analyst, not a decision-maker. It will hand you a ranked list with total confidence and no context. It cannot see the strategic account you are deliberately subsidizing, the loss-leader that anchors a profitable bundle, or the new product whose tail is an investment rather than a leak. Treat its freeloader list as questions to investigate, not a kill list to execute.
Decide what needs to be done before asking the machine anything.
The strategic exception is the one thing the data cannot see.
No dashboard can take responsibility for a decision.
Deliver the decision in a way people can trust.
Treat whoever is on the losing side of it fairly.
Next quarter, when the picture has already changed.
As the cost of analysis falls toward zero, judgment and resolve become the whole game.
The machine is the analyst. Named people hold the decision rights.
Run the prompt sequence on the full transaction file, then validate the numbers you intend to act on.
The tail grows back, and the rerun now costs an afternoon.