Research shows surprisingly low AI usage in product development, especially for expensive design and launch tools. Front-end ideation sees the most adoption (~25%), while German companies are generally further along the adoption curve than their US counterparts.
Studies from MIT, McKinsey, and S&P Global report staggering 80-95% failure rates for AI projects moving from pilot to production. The primary reasons are not technological but organizational: poor understanding of user needs, lack of executive support, and faulty data.
The leading cause of AI project failure is a failure to understand the internal user's needs, a problem dubbed the "shiny thing syndrome." This mirrors the classic product development mistake of building a solution without validating the customer problem first. It's an old lesson in a new context.
A major hurdle for AI adoption is poor data quality and structure. Data is often scattered across Word files, spreadsheets, and presentations. The process of centralizing and cleaning this data into a reliable, "gold" standard is far more challenging and time-consuming than most firms anticipate.
One company successfully implemented AI by repurposing its existing Stage-Gate new product development process. The key shift was treating internal teams as the "customer." This structured approach avoids chaotic, ad-hoc "guerrilla" adoption efforts that often fail.
Rather than replacing customer interaction, AI facilitates a more continuous and iterative feedback loop. It allows for the rapid creation of virtual prototypes that can be shared with customers multiple times throughout development, ensuring the product stays on track.
A study revealed that zero percent of product development leaders use or intend to use AI for core governance decisions. This stands in stark contrast to the financial sector, where AI-driven trading is common. The reluctance stems from a deep-seated fear of letting a machine control strategic business choices.
Contrary to expectations, the wealth of information from AI tools is not making governance easier. Leaders are experiencing information overload, which obscures rather than clarifies go/no-go decisions. The challenge is shifting from data generation to data synthesis and evaluation.