Pricing new digital products is difficult and risky, often requiring weeks of research, analysis, and conversation. The stakes are high: getting the price right can mean the difference between making a profit — or operating at a loss.
The pricing process traditionally begins with market research. This analysis of what’s already in the market helps us understand the similar available products, their price ranges, and which products consumers perceive as premium.
Client collaboration also generates insights into a product’s literal value and target audience, among other considerations. Together, we can evaluate how different price points might affect a digital product's long- and short-term success.
These processes require significant time and effort.
As a product researcher, I love working with WillowTree clients to find the best price for their innovative digital product ideas. Because any new, potentially helpful tool should be explored (as my colleague Jen Ware notes in her recent early adoption article), I recently used ChatGPT to generate a pricing recommendation.
The results suggested this technology could save me anywhere from six weeks to six months of background research and testing. These time efficiencies could translate into as much as five- or six-figure reductions in cost for our clients. WillowTree’s AI and ML Consulting practice is unearthing even more efficiencies as we increasingly incorporate AI tools in our everyday work.
Still, ChatGPT did not replace, but instead complemented our traditional pricing strategy.
By engaging the AI tool in conversation, my goal was to generate a recommendation for a product I had already priced, then compare price points. The product was a learning app and facilitated access to in-person or online music instruction. I provided ChatGPT details about the product’s features, benefits, and functionality. I also described the target audience and current market conditions.
In its response, ChatGPT explained the target market's price elasticity and offered competitive intelligence regarding companies that had successfully implemented similar prices. The AI also noted that too many companies undervalue their products and that the lower end of this range would result in up to six-figure revenue loss.
Finally, the AI presented a price range of $9.99 to $19.99, which included the target price my team had recommended to the client: $19.99.
While I was pleased with the level of detail and thoroughness of ChatGPT’s response, it still fell short.
For instance, when I asked ChatGPT for the sources, references, articles, or other materials it used to draw its conclusions, the AI’s reply amounted to, “Can't help you with that.” In other words, it could give me a price but not fully support its recommendation.
ChatGPT could tell me what — but not how.
First, you must be an expert to thoroughly understand and critically evaluate ChatGPT's recommendations. You also need to have familiarity with AI prompt engineering to produce a quality response in the first place.
Here’s another example of why we must still incorporate human expertise in product pricing.
One client approached us with an idea for a learning app, and they had a potential audience in mind. As I began to dig into the competitive space and talk with potential customers, we identified a different target audience and value proposition from what our clients initially identified.
Without the in-depth conversations and end-user interviews our research team held, we wouldn’t have uncovered these insights, which ultimately helped us establish a more ideal market fit and price point.
In these conversations with clients and my role as a product researcher, I constantly drew on my prior knowledge of pricing best practices. That knowledge base was critical when I communicated with ChatGPT.
Whenever possible, ChatGPT and other large language models (LLMs) must consider the pricing strategies of similar products, analyze the value proposition of a new digital product, and consider the target audience.
Because of my digital product expertise, I know which types of market data are solid foundations for product pricing recommendations. I’ve also learned how to write ChatGPT prompts that receive the most valuable responses. (Check out my colleague and WillowTree Research Director Jill Heinze’s article on effective ChatGPT prompts and their nuanced implications.)
Here are two fictionalized examples of ChatGPT prompts that I found successful in the product pricing process:
Even though these prompts helped me unlock valuable research-backed resources and sentiment analysis, I nonetheless encountered setbacks. The tool responded to me multiple times during our chat with some variation of “Hire a pricing expert to conduct research.”
Even ChatGPT knows that it can help with only part of the process.
AI’s strength lies in its ability to crunch billions of rows of data without error and within seconds. But the technology’s shortcomings in creativity and intuition mean we still need humans. Product teams still need trained researchers to make complex pricing decisions that align with a company’s overall business strategy to identify pricing pitfalls that AI might miss (such as ethical or legal constraints) and to uncover unforeseen target audiences, value propositions, or pricing strategies that will capitalize on any revenue-generating opportunities.
WillowTree’s product research expertise draws on extensive market research, consumer behavior, and pricing psychology. When we combine that with the strengths and possibilities of AI, we can provide more value to our clients than ever before.
