Simon-Kucher: Do you think AI models will soon replace pricing managers in large organizations?
ChatGPT: AI models can assist large organizations by automating tasks and providing data-driven insights for decision-making. But they’re unlikely to completely replace pricing managers anytime soon.
This interaction between our consultants and ChatGPT – Open AI’s cutting-edge large language model interface – illustrates the effect generative AI is having on pricing. It has a major role to play, but for now, human experts are still needed to wield this powerful new tool.
The quantitative nature of pricing makes it an obvious first area of application for new technologies. Many companies are already using artificial intelligence (AI) and machine learning to set prices but still face challenges, such as labor-intensive product benchmarking, non-actionable customer feedback, and unmanageable governance. A custom AI solution could address these challenges but would be costly and technically complex, making the risks high and the rewards uncertain.
With generative AI, particularly the new generation of ready-to-use products like ChatGPT, firms can now tap into unstructured data that previously needed great human effort to be interpreted. This article explores how generative AI can transform pricing processes by making pricing decisions faster, more responsive, and more accurate.
Simplifying benchmarking
A key challenge in pricing is benchmarking products and services. Large retailers often have thousands of stock-keeping units with descriptions that only humans can compare. Determining an item’s distinguishing features or perceived value takes so much time that the task is often neglected, making constructing product hierarchies and comparing products with competitors’ offerings difficult or even impossible.
With text- and image-based generative AI, an AI model can generate reasonable descriptions from a collection of pictures and group the products by description. The process is so quick that the only limit is the amount of data available to the company. Combining this process with internal data or data scraped from competitor websites allows product hierarchies and competitive landscapes to be created in minutes instead of months.
Capturing value from customer feedback and contract management
Often, a company must know not just if their customers are satisfied with their products, but also how they are being used so it can optimize prices and price tiers. Online reviews and blog posts offer a wealth of valuable insights, but extracting information a company can act on from this vast, unstructured dataset with a lot of vernacular language is daunting.
The large language models (LLMs) that drive the latest AIs offer remarkable advances. They can interpret slang, emojis, and unstructured data, allowing them to extract not only customer sentiment from reviews, but key product features and pain points associated with customers’ perception of value and willingness to pay. By understanding how customers use their products, businesses can intelligently segment customers and assign pricing tiers or bundle offerings, unlocking new sources of value quickly adapting to shifting customer sentiment.
The same principle applies to business-to-business (B2B) transactions. Generative AI can parse complex legal language to quickly identify crucial contractual clauses. This gives businesses an understanding of their past negotiations, including what concessions have been made and what negotiation position a potential partner or new team members might take. All this information empowers organizations to craft tailored, competitive offers that optimize revenue and foster long-term relationships with B2B partners.
Getting started with pretrained models
Just a year ago, implementing an AI solution within a company was a huge investment. It would involve hiring a team of data scientists and engineers and then assembling a data warehouse and technical infrastructure before finally building an AI solution over the course of weeks, months, or years.
With recent advances in generative AI, pretrained models have become available. Adopting one of these models avoids the laborious process of creating a model from scratch and produces useful results immediately without access to internal data. In a field such as pricing, where speed is paramount, pretrained models enable companies to capitalize on emerging market trends and opportunities before the competition.
The greatly overstated demise of human expertise
Generative AI is not a silver bullet – it requires expert management to mitigate its risks. Even with pretrained models, they are limited by being trained on public data only – ChatGPT can tell which products are available on a company’s website, but not how well each of those products has sold. At Simon-Kucher, we recommend the following steps when using generative AI in your pricing:
Map out your current pricing process: What are the key decision points? Where is most effort being made? What are the current blindspots? How can more speed or better information improve the process?
Identify underutilized sources of unstructured data and make a plan to acquire them: Which data sources have been underutilized because they required too much effort? Look both internally and at external sources of data from competitors.
Structure the method of using generative AI to interpret new sources of data: Don’t just think about what results you want, but how the results will map onto your current pricing project.
Create a small-scale pilot for integrating generative AI: The outcome of the pilot should highlight gaps in your approach from step 3 and provide a go/no-go decision on scaling up.
Scale up to the larger organization: Once you have a clear idea of the approach to and potential value from using generative AI in your organization, scaling up can begin!
Simon Kucher has decades of experience in designing, piloting, and implementing pricing processes all over the world. Contact us to learn more. The key to maximizing the value of cutting-edge technology lies not only in the technology itself but in the people who wield it.
Read more about Chat GPT and AI in the healthcare industry here.
Note to the reader: OpenAI’s GPT-4 model was used in the development of this document.
