How Procter & Gamble Uses AI to Unlock New Insights From Data

Few organizations can legitimately claim to have been doing analytical research for over a century. But in 1924, Procter & Gamble CEO William Cooper Procter asked Paul “Doc” Smelser, an economist at P&G for 34 years, to find out how many customers used Ivory soap to clean their bodies rather than to wash their dishes and clothes. The answer was 12% for dishwashing, 31% for washing their face and hands, 40% for bathing, and 17% for other uses (presumably, clothes washing). Ivory soap went on to be positioned as a personal care product instead of a household item, though internet commentators suggest that it’s still effective for washing clothes.

P&G was also a pioneer in the creation of common data across a global organization, establishing a large group of analytical professionals and then including them in business units. It has maintained a tradition of using customer and market insights into the present, and it’s now using analytical, generative, and agentic AI to address important business issues.

To find out what P&G is up to with artificial intelligence, we spoke with Jeff Goldman, vice president of enterprise data science and leader of business AI initiatives at the company. We first met Goldman more than a decade ago, when he was developing an innovative approach to analytics visualization at P&G through an initiative called Business Sphere. Since then, he’s gone on to assemble a group of several hundred data scientists and AI engineers who build and deploy AI algorithms at scale across the company’s marketing, digital commerce, supply chain, and sales organizations. Most of the data scientists are directly embedded in business units or AI product teams.

As Goldman put it, the century-old orientation to data and analytics still prevails. “The historical analytical ethos carries through to our current culture,” he said. “We’ve always had a drive for analytics and to understand the dynamics behind our business. But with AI, the nature of questions we can ask and the depth of answers we can provide have expanded dramatically, as has our ability to leverage AI to optimize our business processes.”

P&G’s ‘AI Factory’

Around 2021, Goldman and his colleagues observed that AI was playing an increasingly strategic role in P&G’s business. The complexity of algorithms in production was growing, and the delays involved in progressing from prototype to scale with AI algorithms were having a material financial impact. Algorithms were being developed in a bespoke manner, and each one required custom deployment by the company’s AI engineers.

To speed up AI model creation, P&G developed a capability that it calls an “AI factory.” It provides a vehicle — including data sources, software tools, methods, and defined security protocols — to rapidly develop, test, deploy, and monitor algorithms in production. Seth Cohen, P&G’s CIO, described the project in a podcast episode earlier this year: “The nice part of an AI factory approach is that it’s a platform to allow people instant access to the data within the data repository, but then also instant access to the AI algorithms and generative models. Therefore, a developer spends a lot less time worrying about ‘How do I scale it?’ — because that comes out of the box.”

Goldman said that the AI factory reduces the time to model deployment by roughly six months. He also pointed out that the factory evolves as technology does. For example, it now incorporates agentic AI capabilities, including monitoring agentic systems at scale, registering agents, and applying the requisite Agent2Agent Protocol and the Model Context Protocol to connect multiple agents and tools. Goldman said there are basically two sides to the company’s AI Engineering organization: One is responsible for building and continuously updating the factory, while the other focuses on scaling and operating the algorithms developed within it.

The AI factory capabilities facilitate testing different versions of models for specific business requirements. One factory-developed use case, for example, is the Pampers My Perfect Fit application, which uses the results of an AI-driven questionnaire to provide parents with diaper fit recommendations that are 90% accurate for preventing leaks, the most common pain point in the market. In another example, an analytical AI use case is being realized in Brazil: A system splits and schedules customer orders into truck-size loads, prioritized by shelf need. The system has reduced out-of-stock occurrences in that country by 15%.

P&G’s Generative and Agentic AI Products

Goldman noted that a huge percentage of value has and will continue to come from analytical AI, citing use cases involving supply chain management and media decisioning. But P&G was also early to embrace generative AI, and it has a variety of internal product offerings in place that employ that technology. After providing employees with secure access to a variety of underlying large language models that can be selected based on the business issue being addressed (through a product it has dubbed chatPG), the company has developed imagePG and askPG. ImagePG supports the generation and analysis of images and videos, including for the company’s advertising, while askPG incorporates curated internal unstructured data for use by employees.

While P&G’s generative AI products are focused on personal productivity, the company is equally interested in use cases that can directly scale into business processes. For example, the company’s Great Idea Generator tool creates product and advertising concepts based on consumer trends and previous testing results. It can greatly accelerate the progress of a new idea, from concept to advertisement to shelf. Another AI tool is Project Genie, which synthesizes information from articles and help documents to provide information to over 800 customer service reps, greatly reducing question processing time.

P&G has long been focused on providing easy access to data on how different aspects of the business are performing. Business Sphere (the analytics visualization environment) and a desktop equivalent called Decision Cockpits were part of an earlier approach to that issue, but now the company offers a generative front end to business data called insightsPG. As CIO Cohen put it in an article earlier this year, “Why should you need a dashboard when you can talk to your data?” Goldman told us that although insightsPG is relatively new, it is transforming how the business interacts with data by putting advanced analytical and reasoning capabilities into the hands of a broader population inside the company.

P&G’s experiments with agentic AI have focused on a series of pilots to determine where the technology can be employed most effectively. Goldman commented that there have already been early successes with agents across the advertising, supply chain, and consumer relations areas. Maintaining a human in the loop for these and other processes remains critical, he said.

Improved R&D Through AI

Goldman’s data science and AI organization has a strong partnership with P&G’s R&D function. R&D at the company has a long history of using data and quantitative analysis to understand the chemistry and physics of products and how best to manufacture them. Recently, R&D teams have complemented the work of lab technicians with AI algorithms to enable greater speed in molecular discovery. One example is P&G’s Perfume Development Digital Suite, an ecosystem of digital tools powered by AI and advanced data processing to create new fragrances five times faster than was previously possible. AI algorithms analyze millions of data points and help create perfume character models based on consumer insights into what makes a fragrance smell good. The model integrates perfume character models to identify likely winning formulations that are then tested through fast prototyping and experimentation.

Together with IT, the R&D organization is testing new ways of working using generative AI. In partnership with Harvard Business School’s Digital Data Design Institute, P&G conducted a large-scale field experiment into the ways AI functions as a “cybernetic teammate.” It involved 776 commercial and R&D professionals who were assigned to use generative AI, either as individuals or on teams, to address real consumer issues across four business units with or without generative AI. The experiment found that individuals using chatPG achieved the same performance level as teams that did not use the AI tool, while teams that partnered with chatPG consistently produced the best outcomes. The use of AI also helped break down functional silos between R&D and commercial professionals: Professionals from either background using AI produced more balanced solutions.

Building Human-AI Capabilities

P&G has long had a focus on building human capabilities in data, analytics, and AI. The company has partnered with Harvard Business School and Boston Consulting Group for AI upskilling across the P&G workforce. So far, over 4,000 executives have been through an intensive eight-week AI upskilling program focused on the strategic impact of AI on the industry. This is complemented with in-house upskilling programs focused on using generative AI daily in employees’ work. Together, these programs are creating a growing cohort of leaders who rapidly pilot new AI capabilities and partner with Goldman’s team on the next generation of AI algorithms for the company.

In addition, P&G began offering its ongoing Friends of Data Science certification program several years ago. The company had long had a large quantitative analyst community, but the goal of creating the certification program was to upskill those individuals on analytical AI. The program requires 15 weeks of intensive study and focuses on not only how to build models but also, more importantly, on how they can go wrong. The content is continually updated for new cohorts and now includes content on the transformer models behind generative AI, and on graph machine learning for uncovering increasingly complex data signals.

“A digitally reluctant organization makes it hard to introduce new capabilities,” Cohen said. “We’re changing that.” Given the company’s long-term focus on data and analytics, and now its focus on various forms of AI, P&G appears to be one of the least “digitally reluctant” companies we’ve come across.