Impact Pricing Blog

There is a fantasy running around the business world that AI is finally going to take pricing off everyone’s plate. Plug in the model, feed it your data and watch it spit out perfect prices while you focus on something more fun. It is a nice idea. It just has very little to do with reality.

Pricing refuses to behave like a clean optimization problem. It is one of the least standardized disciplines in business. Finance has its models. Marketing has its funnels. Pricing has a loose collection of frameworks that rarely match from one practitioner to the next. No shared canon. No dominant approach. Every company ends up with its own doctrine built from habit, internal politics and scars.

That fragmentation matters. AI can only learn consistent logic if humans apply consistent logic. In pricing, they don’t. One team relies on value metrics. Another leans on gut feel. Another spends half its energy cleaning up the mess left by creative discounting. Since the field never aligned on a coherent playbook, AI has no stable pattern to absorb.

Even the data works against automation. Pricing data is full of exceptions. Sales teams cut deals they should not cut. Leaders panic at the end of the quarter. Customers bluff. Reps concede. Months later, those transactions sit in a database looking like strategic choices when they were really defensive scrambles. Train a model on that and it will learn behaviors that were never intended to be repeated.

People sometimes push back with the example of B2C, and they are right. In high-volume consumer markets, AI can handle most of the workload. The data is rich and predictable. Millions of small transactions create stable patterns. The machine finds them and works just fine. Until the environment changes. Then it has no clue how to respond because it has no understanding of why prices were set the way they were in the first place. It only knows what happened, not what it meant.

Move into B2B and the whole thing breaks. Deals are lumpy. Each customer has its own politics. A new stakeholder can show up and rewrite the value story overnight. Competitors change their packaging. Budgets shift. The context around a price moves faster than any model can follow. Pricing becomes a judgment problem, not a pattern recognition exercise.

This is where AI actually shines, but not in the way people want it to. AI works when humans supply the logic. A simple example. When I work with clients, they are rarely able to articulate their buyers’ actual problems. The instincts are there, but the language is vague. So we use the problems prompt in Pricing GPT and generate twenty possible problem statements. Some hit them hard. Some fall flat. The model gives us the breadth. The humans supply the judgment. Nothing about that process is autopilot. It is a collaboration. The structure comes from the human. The speed comes from the machine.

Even after working with AI tools for years, and even though they can learn my frameworks inside and out, they still cannot run pricing for me. They can map problems, results, and context. They can spot blind spots, surface possibilities, and sharpen the logic. But they cannot read the power dynamics inside an account. They cannot sense when a sales rep is exaggerating urgency. They cannot detect that a buyer is positioning for a budget conversation three months from now. Those signals live outside the data. They live in human interpretation. Pricing turns on those signals constantly.

The punchline is that AI is a force multiplier, not a replacement. It expands what pricing leaders can see and speeds up how they think. It cannot take responsibility for the decision because the work is rooted in context no model can fully capture. The companies that win will be the ones that use AI as leverage while keeping humans firmly in charge of the logic.

Now, go make an impact!