Explaining a pricing decision: Auditing a rule is not enough
.avif)
Auditing a rule verifies that a calculation was executed correctly. Explaining a decision reconstructs the data, the expected demand, elasticity, cannibalization, the rules applied, and margin-volume trade-offs. According to Sage & IDC, 71% of financial executives would reject an AI tool that is 99% accurate if it cannot explain its answer.
Verifying that a pricing rule has been properly executed has never been enough to explain the resulting decision. You can thoroughly audit the calculation and still be unable to answer the question that really matters: Why this price, for this product, in this store, on this date?
This guide explains what constitutes a truly explainable pricing decision and why a technical audit of a rule is just one of several factors to consider.
A rule that executes correctly is not an explained decision
Auditing a pricing rule involves verifying that a calculation was performed as expected: the correct data was entered, the correct formula was applied, and the correct result was obtained. This is a necessary check—and one that is often well-managed by mature pricing teams.
But this audit answers only one question: Did the system work? It does not answer the question posed by a sales manager, a customer disputing a discrepancy, or a regulator scrutinizing a pricing practice: Why was this specific price chosen, rather than another? A rule can be executed without the slightest technical glitch and produce a decision that no one in the organization can justify in hindsight.
The 6 Layers of a Truly Explainable Pricing Decision
Explaining a pricing decision requires being able to reconstruct, at any time, all the factors that led to it—not just the final formula.
What Led to the Calculation
Competitor prices, sales history, available inventory—accurate, up-to-date input data.
The selected forecast
The projected sales volume at this price, for this product, at this retail location.
The impact on the rest of the product line
The estimated impact on nearby references, not just on the adjusted product.
The business constraints identified
Which rules actually came into play, and which ones were set aside—and why?
The compromise reached
What balance was prioritized between competitiveness, profit margin, and price-image?
What the decision is intended to achieve
The anticipated impact on revenue and margin, which can be measured in hindsight.
A decision can only truly be explained if all six of these layers are traceable—not just the last one, which involves the calculation itself.
Reviewing a rule vs. explaining a decision
These two exercises are often confused, even though they address different questions.
| Dimension | Audit a rule | Explaining a Decision |
|---|---|---|
| Question asked | Did the calculation run correctly? | Why this specific price for this product on this date? |
| Scope | The formula and its technical implementation | Data, demand, elasticity, rules, arbitrage, impact |
| Recipient | The Technical or IT Team | The sales department, a customer, a regulator |
What our article on how an AI pricing engine works reveals is that understanding the mechanics of such an engine—data, recommendations, safeguards—is the first step—but it is not enough to explain why a specific decision, regarding a given product and date, was chosen over another.
71% of executives reject AI they don't understand
This is no longer just a theoretical issue. As generative AI and machine learning become integrated into pricing decisions, confidence in a recommendation depends less and less on its statistical accuracy alone—and more and more on the ability to justify it.
Financial executives would reject an AI tool that is 99% accurate if it is unable to explain its answer, according to a study of 2,275 financial decision-makers in North America and Europe (Sage & IDC, *The Emerging Economics of AI in Finance*, June 2026).
A pricing decision that cannot be explained will not be accepted, even when it is correct—that is the paradox of a high-performing but opaque algorithm: its accuracy alone is not enough to convince the teams that are ultimately held accountable for it.
The real risk isn't making a mistake; it's the inability to justify it.
The most common risk isn't that a pricing engine gets it wrong—it's that it gets it right without anyone being able to prove it.
Companies identify explainability as a key risk associated with the adoption of generative AI—but only 17% are actively working to address it, according to McKinsey & Company’s 2024 State of AI survey.
This gap between the identified risk and the action taken is telling: explainability is recognized as an issue, but is rarely treated as a design prerequisite. Our article on why generative AI gets prices wrong details the other side of the problem: AI that is poorly trained on data produces errors, and without traceability, these errors become as difficult to correct as they are to detect.
Every decision is documented, not just calculated
The AI Center allows you to ask, in natural language, why a specific recommendation was made—what data it was based on, what elasticity was used, and what rules were applied. GENIUS Admin maintains a comprehensive audit log of business rules, permissions, and decisions, which can be viewed at any time, not just at the time of calculation.
GENIUS Predict documents the expected demand and the influencing factors used in each forecast through its "AI Explanation" module—so that traceability relies not on a team's memory, but on the system itself.
Learn more about the platform on our MPS page : Booper, the modular pricing solution.
Could you explain your latest pricing decision?
Spend 30 minutes with our team to objectively assess which aspects of your current pricing would be difficult to justify in hindsight.
FAQ
Auditing a rule verifies that a calculation was executed according to the intended logic. Explaining a decision involves reconstructing everything that led to it: the data used, the expected demand, the elasticity and cannibalization taken into account, the rules actually applied, the margin-volume trade-off chosen, and the anticipated impact.
According to a Sage/IDC study of 2,275 financial executives, 71% of them would reject an AI tool that is 99% accurate if it cannot explain its answer. A pricing decision that cannot be explained will not be adopted, even when it is correct.
The data used, expected demand, price elasticity, cannibalization, the business rules applied, and the margin-volume trade-off selected—all the way through to the expected impact on revenue and margin.
No. Speed of execution says nothing about the ability to justify a specific decision. An algorithm can generate a price instantly without being able to explain, in hindsight, why that price was chosen over another.
Internally, before the sales or finance department. Externally, when dealing with a client who disputes a discrepancy, or in the event of a regulatory audit, particularly with regard to compliance (Omnibus) or non-discriminatory pricing.
On the contrary, when it is built in from the design stage rather than added as an afterthought, a decision that has already been documented is validated more quickly than one that must be reconstructed manually to satisfy an audit.
For each recommendation, present the determining factors (elasticity, competition, constraints, applied rule) in industry-specific terminology, with the option to re-run the calculation. A recommendation that cannot be explained will not be adopted. See “AI That Decides, AI That Executes.”
Also in this series
- How does an AI pricing engine work?
- Why Generative AI Gets Prices Wrong
- Data Quality: The Real Glass Ceiling in AI Pricing
- Why Pricing Decisions Should Never Be Left to a Single Person
- Implementing and Maintaining a Pricing Policy Over Time: Governance, Exceptions, and Waivers
Sources: Sage & IDC, *The Emerging Economics of AI in Finance*, a survey of 2,275 finance executives, published in June 2026 · McKinsey & Company, *The State of AI 2024* survey · Booper, internal product data (AI Center, GENIUS Admin, GENIUS Predict)
Further reading
It has become increasingly common to compare two implementations of the same general-purpose language model before deciding which one to use to drive pricing; but the real question isn’t which one to choose, but where to position each one.
A general-purpose LLM lacks four key components required for pricing decisions: access to real-world data, explicit business rules, impact simulation, and explainable governance—these are the responsibilities of a specialized solution, not the LLM alone.
Generative AI projects that combine in-house expertise with that of a specialized partner have a significantly higher success rate than those developed solely in-house: 67% versus 22%.
Auditing a rule verifies that a calculation was executed correctly. Explaining a decision reconstructs the data, the expected demand, elasticity, cannibalization, the rules applied, and margin-volume trade-offs. According to Sage & IDC, 71% of financial executives would reject an AI tool that is 99% accurate if it cannot explain its answer.
A pricing tool faithfully processes the data provided to it; the quality of the results depends first and foremost on the quality of the input data. Three tasks are priorities before any configuration: a product catalog free of duplicates, synchronized prices and costs, and a continuous sales history. There’s no need to wait for perfect data to get started: an initial setup using files allows you to gradually improve reliability.
This is precisely what BOOPER’s Change Management module addresses: supporting teams to ensure that this data analysis does not hinder the adoption of the new tool.