Using AI in Pricing: An in-depth discussion
- FutureUP
- 7 days ago
- 7 min read
AI Pricing ≠ Dynamic Pricing 😯
Most people think AI pricing = dynamic pricing. But there’s far more happening behind the scenes…
FutureUP's founder, George Boretos, joins Speaker and Pricing strategist Claire Wang for an in-depth discussion on how AI is transforming pricing strategy and performance!
Here's what companies gain today when AI supports pricing:
✔️ Smarter recommendations per segment
✔️ Better upsell and cross-sell paths
✔️ Higher revenue and profit uplift
✔️ Faster decisions with fewer errors
🤖 Dynamic pricing adjusts in real time, based on demand or competition. But most teams, especially in B2B, use periodic AI pricing to optimize by country, segment, or product portfolio. Both matter. Both work.
🤲 AI + human judgment beats either alone. We’ve seen poor outcomes when AI runs unchecked. Human oversight adds common sense, instinct, and accountability.
🤔 Worried about data gaps? Good news:
✔️ AI can infer missing info from sales patterns
✔️ You can test accuracy before adding complexity
✔️ Proxies work when perfect data doesn’t exist
📈 Adoption is still early (below 20% in our survey). But results are clear:
✔️ Higher productivity
✔️ Stronger margins
✔️ Strong uplift from a few points… to 2x improvement
🔮 What’s next?
✔️ GenAI scraping unstructured pricing intel
✔️ Chat-based pricing interfaces
✔️ AI systems collaborating across functions
👉 Still hesitating? Remember, the biggest risk is being left behind.
Start now, but start small:
✔️ One or a few products
✔️ Selected markets
✔️ Quick, measurable wins
You can watch the full recording or read the discussion Q & A below :
Discussion Q & A
Can you explain what AI-driven pricing means? Is it simply dynamic pricing, or is there more to it?
With AI-driven pricing, a suggested price, usually the optimum price to maximize revenue or another KPI, is estimated by an AI algorithm instead of humans. The final price could then be automatically presented to the customer or forwarded to the commercial team for their final decision. This is often coupled with additional suggestions like product recommendations for upselling and cross-selling.
Dynamic pricing is just one form of AI-driven pricing, where prices are automatically estimated in real time by an AI algorithm, considering competitors' prices, market conditions, supply vs demand, or other parameters, usually without any or limited human intervention. Typical examples of dynamic pricing are airline tickets, hotel bookings, and online retail stores, as frequently seen in B2C applications.
Non-dynamic AI pricing can take into account all these factors, but it’s not real-time. This is the most frequent form today. For instance, if a company sells directly to corporate customers in different geographies with different spending capacities, you can optimize your list price per country or market segment to maximize your revenue. This usually happens a few times per year or quarter, so it is not real-time like dynamic pricing, although prices can be dynamically adjusted for new market or other conditions.
How is AI different from the traditional use of statistical analysis or prediction in Pricing? What is new that AI can bring?
The best way to describe this is that AI is statistics with a purpose!
It all depends on how we perceive AI. When AI was introduced in the mid-20th century, it was initially mainly neural networks in an attempt to mimic human intelligence, which is now referred to as Artificial General Intelligence (AGI). That was quite distinct from statistics, which, as a field, predates AI by centuries.
AI has grown beyond its original scope over the past few decades, and today, it is generally perceived as an “umbrella” encompassing machine learning and many statistical techniques.
Regression analysis, classification, and Bayesian inference are just some examples used intensively in many AI applications. However, there are other areas in AI, such as neural networks, deep learning, backpropagation, and numerous other AI methods, that are only partly based on statistics and are quite unique.
Apart from their scientific differences, there is also a key difference in their scope of use. Statistics focus on discovering data patterns to use them for predictions and reverse-engineer hidden relationships. AI focuses on imitating human behavior from simple to more complex tasks.
AI is more like statistics with a purpose, and I think this is perhaps the best definition of AI!
Many people worry about AI-driven pricing being unfair or opportunistic. How can we ensure AI is used responsibly in pricing?
This is called responsible AI.
We’ve all seen the Oasis concert fiasco in the UK, with AI raising prices by 2-3 times within hours, making everyone furious. The band had to eventually abandon their dynamic pricing model for their subsequent concerts.
There is only one remedy for responsible and successful AI use: Human-AI collaboration.
In this “symbiosis,” people should have at least a high-level supervision of AI systems during design, deployment, and go-live. Why? Because we have common sense, instinct, and accountability that AI does not possess.
After all, it was a person who pulled the plug on the AI system in the Oasis concert case, not the AI itself.
We often say garbage in, garbage out. Do companies need to have a sound data infrastructure to start benefiting from using AI in Pricing? If not, what is your suggestion?
Yes and no.
As a general rule, the more quality data you have, the better the result.
But in many cases, it’s impossible or very difficult to collect all the necessary info. Take B2B competitive intelligence, for instance. Theoretically, you should know all your competitors’ prices. But this is proprietary and, possibly, confidential info that you have limited or no access to at all.
But that's not necessarily a problem.
AI is built to guess missing information, at least to some extent. Your competitors’ actions and prices are reflected to some extent in your sales, so you may be able to understand competitors’ impact from your sales data.
The trick here is to run your AI model first without competitive intelligence or other missing or scarce info and test the model for accuracy, i.e., how close your sales or price predictions are to reality. If it is okay, then you don't need the extra/missing information, as it is partially embedded in your existing sales or other data. If not, you could add a proxy, such as your sales team's estimates. And so on and so forth until you reach the desired outcome.
The key to all this, especially in the absence of key data (for instance, in B2B), is to use an advanced AI econometric model that can make informed predictions with limited data.
Have you seen any use cases of companies utilising AI in pricing? What are the lessons?
The current AI-Price Optimization adoption rate is 16%, according to a survey conducted by FutureUP with ValueBizbooster (a company specializing in value-based selling).
This means that AI-Price Optimization is still in its early adoption stages, but many companies have already taken action or begun experimenting in this area.
Based on our survey and my experience, most people are happy with the results they get, with the main focus areas being productivity improvement and, most importantly, financial gains.
Performance uplift for significant indicators like revenue, profit, or conversions can range from a few percentage points to 2-fold improvements, depending on the industry and the size or maturity of the company or product.
What future developments do you foresee for AI in pricing? Are there any new technologies or methods on the horizon?
I can see two important directions from the merging of Generative and Predictive AI technologies:
Gen AI can help retrieve and structure pricing data through web scraping or by analyzing information hidden in comments and text fields.
Second, Gen AI will gradually become a kind of operating system or UI for communicating with devices, software, and systems. This includes Predictive AI as well. Instead of using fixed menus, options, reports, and charts, you would simply ask the system to perform an action.
Imagine a Gen AI front-end through which you would ask, for instance, the AI system to identify hidden pockets of inelasticity in your product portfolio. Or to estimate the demand for a product if the price of a competitor changes. Or to collect additional and currently unstructured pricing information from your CRM system. The AI system will trigger the necessary processes and AI models, and generate, visualize, or export the requested info.
Seamless integration of Gen AI with Predictive AI, including Price Optimization, will be key in the future.
Fast forward many years, and AI systems will work together in a similar way, interacting with each other and humans as part of the Human-AI symbiosis mentioned before.
What would you say to businesses hesitant to adopt AI-driven pricing? What first steps should they consider?
The biggest threat of AI today is to be left behind. So, don’t hold back. Follow Nike’s advice, and just do it!
Don't overthink it or try to address all pricing challenges at once. Choose the simplest challenge, choose a partner for the AI and the pricing part, and start experimenting.
Aim for quick wins and start small. For instance, start with a product and a few countries where you have a lot of data. FutureUP’s system, for instance, identifies areas of low profitability or hidden pockets of inelasticity where it is a no-brainer to increase the price and gain a lot.
Always check the estimates of the system to see if they make sense. Don’t listen to data scientists claiming that statistically, the system is accurate and works fine. That's an important first test. The second is you. You should be able to identify and relate at least directionally with the results.
As you accumulate more experience and benefits from your quick wins, you can refine your approach and expand. Quick wins will also increase your organization's appetite for further investments and experimentation, providing you with additional resources and a stronger commitment from top management.
At all steps, focus on a specific challenge to address with clear benefits to demonstrate and seek alignment and internal buy-in! AI has a purpose, and so must you within your organization!
🎧 Want to learn more?
Interested in learning more about AI-Powered Price Optimization and Strategic Forecasting?
Comments