Fair and Ethical Pricing in the Age of AI: From Revenue to Responsibility

Advanced analytics and AI now allow companies to set highly granular, personalised prices. Airlines can increase fares six- or seven-fold for a single high-demand weekend. Industrial suppliers can model a customer’s switching costs and quietly embed that insight into price recommendations. These capabilities are powerful and risky.

Nick Boyer, Conga’s Senior Director of Strategic Consulting, prompted leaders at the Manufacturing Pricing Excellence 2026 – Power of 50 to ask themselves “What will the market bear?” and also “Can we defend this price to our customers, to our sales teams, and to ourselves?”

From Value-Based Pricing to Ethical Pricing  

Most pricing organisations are comfortable talking about value-based pricing: differentiating prices according to the value customers receive compared with alternatives. Ethical pricing goes a step further. It tests whether the way value is translated into price respects the relationship with the customer.

A useful framing is a spectrum with two ends:

• Ethical pricing: Prices reflect value, are set using transparent and defensible criteria, and are applied consistently to comparable customers. The customer may negotiate, but does not feel tricked.
• Exploitative pricing: Prices take advantage of customer ignorance, urgency, dependency, or lack of alternatives. Short-term gains are prioritised over the long-term relationship, often punishing loyalty rather than rewarding it.

Both can emerge from the same set of market facts. For example, a flight from London to Budapest for a Champions League final weekend suddenly priced at over £1,200, versus a fraction of that the week before or after. From a revenue management standpoint, it is a textbook response to a demand spike. From the customer’s standpoint, particularly a lifelong supporter with no realistic alternative if they want to attend the final, it feels like exploitation.

The same dilemma plays out in B2B when supply is constrained, or when a supplier has strong market power. The question is not only whether a customer is willing to pay, but whether using that willingness in full is consistent with the type of relationship the supplier claims to build.  

Scarcity, Power, and the Long Memory of Customers  

The COVID period offered several stark examples of how companies responded to scarcity.

Some semiconductor manufacturers reportedly raised prices by 10 to 50 times as OEMs scrambled for critical components. Steel and metals distributors invoked force majeure to exit fixed-price contracts, only to resell the same products back to the same customers at much higher prices. In each case, supply-and-demand logic supported price increases. But the commercial impact did not end with that quarter’s margin.

Customers responded by diversifying supply, building redundancies, and rethinking their dependence on any single supplier. What looked like a pricing win on paper became a trigger for customers to reduce their future exposure and the supplier’s future revenue.

An instructive counterexample came from the plaster market in the UK. A large manufacturer, heavily regulated due to its market share, was unable to raise prices significantly even when a competitor’s plant burned down, taking a sizeable portion of national capacity offline. Demand surged, prices did not.

Instead, the company designed an allocation mechanism based on past purchasing behaviour, effectively rewarding loyalty by prioritising volume for customers who had historically bought more. Smaller, unregulated competitors took the opposite route, raising prices sharply to monetise the shortage.

Once capacity normalised, the regulated supplier’s market share rose. Customers recognised that, under stress, this supplier had acted predictably and protected existing relationships, while others had not. Scarcity events are trust tests, and customers’ memories are long.

AI Pricing: Consistency at Scale and New Risks  

AI and advanced pricing engines promise something B2B organisations have struggled with for decades: consistent, data-driven price recommendations that account for dozens of factors at once. This can improve fairness. Similar customers, with similar cost-to-serve and in similar contexts, receive similar prices, regardless of which salesperson they talk to.

However, the same complexity that makes AI powerful also introduces new ethical risks.

For example, an AI pricing model built using a “switching cost” attribute to determine how difficult and costly it would be for a customer to move to a competitor, would increase the recommended price based on how high the switching cost would be. On paper, this is a rational monetisation of customer dependency.

In practice, it produced price differences of 18–30% for customers that were otherwise very similar, buying the same commodity product. By embedding a power asymmetry into its logic, it produced outcomes that would be difficult to justify in a direct conversation with a customer.

The more attributes a model uses, and some industrial AI models now operate with 60 or more, the greater the risk of such unintended consequences. Not all drivers of willingness to pay are acceptable to use, even if they are predictive.

In both B2B and B2C, the ability to measure and segment willingness to pay does not automatically grant the right to exploit every driver that appears in the data.  

Defining a Fair Price in B2B  

There is no universal formula for a fair price. However, four tests can help B2B leaders decide whether a price is defensible:

1. Value  

The price should reflect the value of the product or service versus alternatives, from the customer’s perspective. This includes performance, risk reduction, service levels, and any differentiation the supplier provides.

2. Consistency  

Comparable customers buying comparable products under comparable conditions should see comparable prices. This does not preclude segmentation; it demands that segmentation be based on objective, non-discriminatory criteria.

3. Explainability  

A sales leader should be able to explain how a price was derived and which factors influenced it without embarrassment. If the true logic behind a recommendation cannot be shared with a customer, it is a warning sign.

4. Empathy

Pricing must take into account the customer’s context and constraints. This does not mean underpricing systematically, but it does mean recognising when maximising short-term margin will damage trust or appear to punish dependency or loyalty.

These tests are especially important as organisations tighten control of their price corridors and automate more of their decision-making. Once AI models and agents are driving quotes at high velocity, unwinding a pattern of perceived unfairness becomes much harder.  

Guardrails: Automate the Arithmetic, Not the Accountability  

As AI matures, some organisations are already experimenting with pricing processes without human involvement. For most industrial and B2B companies, that level of autonomy is still distant. Humans remain firmly in the loop. And the more pricing becomes data-driven and automated, the more important it is that humans define the rules of the game.

Practical guardrails include:

Attribute selection discipline  

Carefully vet which customer and transaction attributes are used in pricing models. Exclude factors that exploit vulnerability, dependency, or subjective judgements (for instance, loosely defined notions of “customer potential” that can mask bias).

Boundary and scenario testing  

Before deploying models at scale, run scenarios for edge cases: captive customers, extreme scarcity, key accounts, and long-term contracts. Assess not only margin impact but whether the resulting prices pass the value, consistency, explainability, and empathy tests.

Consistency monitoring  

Regularly review whether similar customers are receiving similar prices. AI can improve consistency, but only if the training data and segmentation logic are sound.

Clear ownership  

Assign explicit accountability for pricing logic, not only for target margins and approval workflows, but for the ethical posture of the models and rules in use.

Ethical pricing is not a constraint on AI; it is a design requirement. When guardrails are built in from the outset, AI can support a more rational, defendable, and trust-preserving approach to price differentiation.  

Conclusion  

Pricing has always been about trade-offs, but AI and scarcity events expose those trade-offs more starkly. B2B organisations can raise prices sharply when customers are captive or desperate; they can also choose to protect trust and loyalty, even when the data says customers will pay more.

The most effective leaders treat ethical pricing not as a philosophical add-on, but as a core part of commercial strategy, asking:

• Are we rewarding or punishing loyalty?  
• Would we be comfortable explaining our pricing logic to our best customer?  
• If this model runs at scale for three years, what kind of reputation will it create for us?

In industrial markets and aftermarkets where relationships span decades, that reputation is an asset as tangible as any product line. AI will increasingly handle the arithmetic of pricing. The responsibility for fairness and for the long-term health of customer relationships will remain firmly with humans.